From swarm-modelling@santafe.edu Tue Apr 1 18:30:50 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 11:30:50 -0700 Subject: swarm-modelling: testing Message-ID: <199704011830.LAA10170@grasshopper.santafe.edu> Testing the list. glen ================================== swarm-modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 18:37:22 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 11:37:22 -0700 Subject: swarm-modelling: testing Message-ID: <199704011837.LAA10181@grasshopper.santafe.edu> Just testing the list. glen ================================== swarm-modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 19:12:47 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 12:12:47 -0700 Subject: SMod: testing again Message-ID: <199704011912.MAA10209@grasshopper.santafe.edu> Just testing again. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 19:13:48 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 12:13:48 -0700 Subject: SMod: testing again In-Reply-To: <199704011912.MAA10209@grasshopper.santafe.edu> References: <199704011912.MAA10209@grasshopper.santafe.edu> Message-ID: <199704011913.MAA10213@grasshopper.santafe.edu> replying to the test glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 19:31:06 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 12:31:06 -0700 Subject: SMod: New List! Message-ID: <199704011931.MAA10232@grasshopper.santafe.edu> Hey guys, If you'll take a brief look up at the header "From:" field, you'll see that you've been subscribed to a new list, the Swarm-Modelling list. I've appended the info file (normally retrieved via a message to majordomo saying "info swarm-modelling" in the body. That has an attempted explanation of what this list if for. I'd appreciate comments telling me if you think it's worded correctly or if you think the target discussions are defined too narrowly. Let me know what you think. glen -------------------------------------------------------------------- [Last updated on: Tue Apr 1 11:59:27 1997] The Swarm Modelling list, , is an open mailing list intended to provide a forum for the discusion of high level modelling and simulation issues. Special attention should be devoted to using Swarm to do such modelling; but, posts to the list are not restricted to such. This list is not for installation or bug reports for the Swarm package. All posts should be understandable to those (esp. scientist and manager types) who have an interest in systems modelling but do not have a programming background. This list is archived: ask majordomo to "index swarm-modelling" There is a set of documents about Swarm on the World Wide Web at the URL http://www.santafe.edu/projects/swarm/. And there are two other lists of interest to Swarm programmers, the Swarm-Support list and the Swarm-GIS list. Use Majordomo to get the info files for those lists. The swarm developers at SFI can be reached as - this is a very small set of people, those who are currently actively developing the core of Swarm. Welcome aboard! -------------------------------------------------------------------- For list administration needs, please use the Majordomo server at the Santa Fe Institute. For help using Majordomo, send mail to: majordomo@santafe.edu with the following in the body: help -------------------------------------------------------------------- ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 19:48:40 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 12:48:40 -0700 Subject: SMod: List dynamics Message-ID: <199704011948.MAA10242@grasshopper.santafe.edu> With the addition of two new lists (swarm-modelling and swarm-gis), I've decided it's time to get a little more sophisticated with the delivery and content of the mailing lists. So, as usual, I'd like your input on a couple questions: 1) Since we now have 4 lists (announce, support, gis, modelling), simply subscribing the swarm-support list to the swarm-announce list doesn't cut it. It's conceivable that someone may want to be members of both swarm-announce and swarm-modelling but not swarm-support or swarm-announce and swarm-support but not swarm-modelling. (I'm presuming that most of the people on swarm-gis will be members of one of the other two lists, so it's not as much of a consideration, here.) Now, the *best* way I can see this happening is to subscribe anyone who is a member of any one of the other three lists to the announce list. That prevents duplicate posts to any list member and allows all the people on one of the other three lists to receive posts to swarm-announce, as well. Does this sound right? Of course, the problem is automating the subscription to swarm-announce and triggering that subscription from a user-initiated subscription to one of the other lists.... But, I presume I can deal with that. 2) I've added prefixes to the "Subject:" fields of messages from each of the support, gis, and modelling lists so that anybody out there using a mailer (or a mail filter) that is capable of sorting the mail can do so based on that prefix. a) Do you like this or dislike it? b) Would you prefer different prefixes? (currently, they're "SSup," "SGIS," and "SMod") 3) I've added a footer to each message with a short description of the lists purpose and a hint at how to use majordomo (as was suggested by a person who didn't know how to unsub). Again, let me know if this is ok and feel free to criticize my wording. 4) Digests are coming! We don't have them automated yet. But, it should be ok for now. The list is calm. I'll let you know when we have them up. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 20:36:43 1997 From: swarm-modelling@santafe.edu (Pietro Terna) Date: Tue, 01 Apr 1997 22:36:43 +0200 Subject: SMod: List dynamics Message-ID: <1.5.4.32.19970401203643.006897b8@alpcom.it> As 'an incoming true swarm user' I think that all four lists are useful and also necessary, without limitations. I'll work about economic models but ideas from GIS are also stimulating etc. ... So a little question (unfair?): Why not to keep a unique list? Pietro At 12.48 01/04/97 -0700, you wrote: > >With the addition of two new lists (swarm-modelling and >swarm-gis), I've decided it's time to get a little more >sophisticated with the delivery and content of the mailing >lists. So, as usual, I'd like your input on a couple >questions: > >1) Since we now have 4 lists (announce, support, gis, modelling), >simply subscribing the swarm-support list to the swarm-announce >list doesn't cut it. It's conceivable that someone may want to >be members of both swarm-announce and swarm-modelling but not >swarm-support or swarm-announce and swarm-support but not >swarm-modelling. (I'm presuming that most of the people on swarm-gis >will be members of one of the other two lists, so it's not as much >of a consideration, here.) > >Now, the *best* way I can see this happening is to subscribe anyone >who is a member of any one of the other three lists to the announce >list. That prevents duplicate posts to any list member and allows >all the people on one of the other three lists to receive posts to >swarm-announce, as well. > >Does this sound right? > >Of course, the problem is automating the subscription to swarm-announce >and triggering that subscription from a user-initiated subscription >to one of the other lists.... But, I presume I can deal with that. > >2) I've added prefixes to the "Subject:" fields of messages from each >of the support, gis, and modelling lists so that anybody out there >using a mailer (or a mail filter) that is capable of sorting the mail >can do so based on that prefix. > > a) Do you like this or dislike it? > b) Would you prefer different prefixes? (currently, they're > "SSup," "SGIS," and "SMod") > >3) I've added a footer to each message with a short description of >the lists purpose and a hint at how to use majordomo (as was >suggested by a person who didn't know how to unsub). Again, let >me know if this is ok and feel free to criticize my wording. > >4) Digests are coming! We don't have them automated yet. But, >it should be ok for now. The list is calm. I'll let you know when >we have them up. > >glen > > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 23:26:09 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 16:26:09 -0700 Subject: SMod: List dynamics In-Reply-To: <1.5.4.32.19970401203643.006897b8@alpcom.it> References: <1.5.4.32.19970401203643.006897b8@alpcom.it> Message-ID: <199704012326.QAA10318@grasshopper.santafe.edu> Pietro Terna writes: > As 'an incoming true swarm user' I think that all four lists are useful and > also necessary, without limitations. I'll work about economic models but > ideas from GIS are also stimulating etc. ... > So a little question (unfair?): Why not to keep a unique list? > Pietro This is a bit of a controversy. Several people feel we don't need more than one list. But, more people seem to feel we do. So, I think it's probably a good idea. I believe their argument goes like this (if any proponents for the list split want to correct me, feel free). Some Swarm users have been using Swarm for awhile, now. They are not as interested in seeing questions they've seen over and over again or in questions relating to some obscure platform. Since alot of people on this list may get several (like...over 100?) email messages a day, it is more convenient to be able to subscribe to lists on which they're fairly sure only "interesting" messages are posted. Right now, this isn't really that much of a problem, because there hasn't been that much traffic. But, in high traffic periods (like after a new release), it can be difficult to sift through the posts to find which ones are "interesting" and which ones aren't. Ideally, with the scheme I'm setting up, you can subscribe to all the lists, if you'd like. In fact, this might be a good idea for a new Swarm user. But, after awhile, you might realize that you're not that interested in one or the other. It should also be amenable to those people who are not actually using Swarm, but who employ (or advise) people who are. Then you could subscribe to the modelling and GIS lists to interact on one level without having to pour through intricacies you're not concerned about. So, basically, it was decided at SwarmFest to split the list into a "code-slinger" type list and a "theory-slinger" type list. And, depending on whether the scheme works or not, we can keep or trash this scheme. If the traffic just continues to grow, then it might be justified to make another split later on. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 2 16:52:35 1997 From: swarm-modelling@santafe.edu (Stephen C. Upton) Date: Wed, 02 Apr 1997 09:52:35 -0700 Subject: Modeling of Crowd Behavior Message-ID: <3.0.32.19970402095234.007255b4@tsa-po.lanl.gov> As a 1.7 Swarm Lurker (between class 1 and 2 -- closer to 2 - I've finally got Linux loaded on a new machine! I've previously had Swarm up and running at another job, but, alas it's not part of my job description here -- yet!), I appreciate the separation of the lists. As of now, I am more interested in modeling aspects than specifics about Swarm code -- that will come :P My question is: has, or is, anyone attempting to look at modeling crowd (of humans) behavior, whether it be with Swarm or without? This has obvious applications for the justice department and the military, for example. They would like to disrupt a crowd using non-lethal technologies, but certain actions may be more provocative than others. I would appreciate any references any one might have also. thanx upton *** * **** * ***** ********* ** ****** ***** *** ***** ******** ********* Stephen C. Upton TSA-5, MS F602 Los Alamos National Laboratory Los Alamos, NM 87545 505-667-9435 FAX 505-665-2017 upton@lanl.gov ******* ********* *** ** *** ******** **** ****** ** ****** **** *** *** ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 2 18:17:26 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Wed, 2 Apr 1997 11:17:26 -0700 Subject: Modeling of Crowd Behavior In-Reply-To: <3.0.32.19970402095234.007255b4@tsa-po.lanl.gov> References: <3.0.32.19970402095234.007255b4@tsa-po.lanl.gov> Message-ID: <199704021817.LAA10633@grasshopper.santafe.edu> Stephen C. Upton writes: > My question is: has, or is, anyone attempting to look at modeling crowd (of > humans) behavior, whether it be with Swarm or without? This has obvious > applications for the justice department and the military, for example. > They would like to disrupt a crowd using non-lethal technologies, but > certain actions may be more provocative than others. > > I would appreciate any references any one might have also. I'm not sure if this is a good reference; but, it sure looks relevant. http://cbl.leeds.ac.uk/rodw/papers/tiemec-95/main.html Also, this seems perfect for a multi-agent system. I guess the problem lies in the agent motivation. If we could identify a common motivation and behavior that's purely local and that gives rise to a riot, we could ethically experiment with riot control. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 00:45:49 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Wed, 2 Apr 1997 17:45:49 -0700 Subject: Taking Snapshots Message-ID: <199704030045.RAA11710@grasshopper.santafe.edu> Hey! Well, I got fed up with Sven's needling [grin] and hacked xwd so that it's callable as a subroutine. It's relatively painless to compile it as a library, link it with Swarm, and put a method in, say, the observerSwarm that is called at whatever frequency you'd like. It's not in distributable form, yet. For instance, the xwd package still tries to do a link when you make it and you have to create your own libxwd.a with an ar rvs command. But, if anybody's got ants in their pants, I can send it to them. I doubt this is the way we should really go. XWD files are pretty large (105583 bytes for 80x80 default zoom heatbugs). Plus, we're trying to move *away* from X if possible, not towards it. So, I've started looking into HDF and AVS. Mind you, however, that this is still not high priority. The hackish stuff is so simple that it can never be considered high priority. And the long-term stuff ties in with our general data visualization problem and our portability problem, both of which are superceded by ||-Swarm, the Manual, and the Alpha port. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 01:08:59 1997 From: swarm-modelling@santafe.edu (Sven N. Thommesen) Date: Wed, 02 Apr 1997 19:08:59 -0600 Subject: Taking Snapshots Message-ID: <3.0.32.19970402190854.0099d700@spidle2.humsci.auburn.edu> At 05:45 PM 4/2/97 -0700, you wrote: > >Hey! > >Well, I got fed up with Sven's needling [grin] and hacked xwd so that Halleluja ! The squeaky wheel finally got a little grease! >it's callable as a subroutine. It's relatively painless to compile it >as a library, link it with Swarm, and put a method in, say, the >observerSwarm that is called at whatever frequency you'd like. > >It's not in distributable form, yet. For instance, the xwd package >still tries to do a link when you make it and you have to create your >own libxwd.a with an ar rvs command. But, if anybody's got ants in >their pants, I can send it to them. > Well, don't know 'bout them ants -- but I'd like some, anyhow ... >I doubt this is the way we should really go. XWD files are pretty >large (105583 bytes for 80x80 default zoom heatbugs). Plus, we're >trying to move *away* from X if possible, not towards it. So, I've >started looking into HDF and AVS. Mind you, however, that this is >still not high priority. The hackish stuff is so simple that it can >never be considered high priority. And the long-term stuff ties in >with our general data visualization problem and our portability >problem, both of which are superceded by ||-Swarm, the Manual, and the >Alpha port. > >glen > Well, we humble peons out here gratefully take whatever crumbs come our way ... :-) Thanks, Glen Sven ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 14:27:04 1997 From: swarm-modelling@santafe.edu (Brian Ruth (CBNED/SMSB) ) Date: Thu, 3 Apr 1997 09:27:04 -0500 (EST) Subject: Modeling of Crowd Behavior In-Reply-To: <3.0.32.19970402095234.007255b4@tsa-po.lanl.gov> Message-ID: On Wed, 2 Apr 1997, Stephen C. Upton wrote: > As a 1.7 Swarm Lurker (between class 1 and 2 -- closer to 2 - I've finally > got Linux loaded on a new machine! I've previously had Swarm up and running > at another job, but, alas it's not part of my job description here -- > yet!), I appreciate the separation of the lists. As of now, I am more > interested in modeling aspects than specifics about Swarm code -- that will > come :P > > My question is: has, or is, anyone attempting to look at modeling crowd (of > humans) behavior, whether it be with Swarm or without? This has obvious > applications for the justice department and the military, for example. > They would like to disrupt a crowd using non-lethal technologies, but > certain actions may be more provocative than others. > > I would appreciate any references any one might have also. > > thanx > upton > > > > *** * **** * ***** ********* ** ****** ***** *** ***** ******** ********* > Stephen C. Upton > TSA-5, MS F602 > Los Alamos National Laboratory > Los Alamos, NM 87545 > 505-667-9435 FAX 505-665-2017 > upton@lanl.gov > ******* ********* *** ** *** ******** **** ****** ** ****** **** *** *** I'm currently a Swarm user wannabe (I'm waiting for the source+binary release for SGIs), but I have looked a bit into modeling crowd behavior. Dana Eckart of Radford University and I are currently developing a cellular automata model, using his Cellular simulation system (available at http://rucs2.sunlab.cs.runet.edu/~dana/ca/cellular.html), which demonstrates the emergence of panic within a unit of soldiers when exposed to one or more battlefield threats. The model determines whether a particular soldier will panic by assigning a probability of panic conditional on the number of wounded and/or panicking neighbors within the soldier's extended Moore neighborhood (21 x 21 cells on a 2D lattice), and then performs a random draw to determine the soldier's panic state (panicking/not panicking). Flocking behavior is also considered within the unit (for non-panicking soldiers only) , where a soldier's speed and direction of travel is determined by that of his neighbors. This type of model could easily be extended to a civilian crowd being fired upon by a sniper or being broken up by law enforcement officers, where the probability of panic associated with a civilian would be somewhat higher than that for a soldier under similar circumstances, and flocking (or anti-flocking) might be governed by a hierarchy based on parameters such as uniform (national guard, policeman), visible weapon (gun, club), and so on. Also, check out the examples link on Prof. Eckart's Cellular page (referenced above), where he presents an implementation of a flocking algorithm developed by Tamas Vicsek and colleagues at Eotvos University in Budapest. Hope this helps. Brian Ruth *===================================================================* | Brian G. Ruth ** Voice: Comm. (410)612-8687 | | U.S. Army Research Laboratory ** FAX: Comm. (410)671-2375 | | ATTN: AMSRL-SL-CM ** | | Bldg. E3331 ** | | Aberdeen Proving Ground, MD ** | | 21010-5423 ** email: bruth@arl.mil | |===================================================================| | Army Research Laboratory | | Survivability/Lethality Analysis Directorate | | Chemical-Biological, Nuclear & Environmental Effects Division | | Survivability Modeling & Simulation Branch | *===================================================================* ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 14:16:06 1997 From: swarm-modelling@santafe.edu (Randy Gimblett) Date: Thu, 03 Apr 1997 08:16:06 -0600 Subject: Modeling of Crowd Behavior Message-ID: <1.5.4.32.19970403141606.006f3bf4@ag.arizona.edu> Glen and SWARM Users In response to modeling human behavior, we have developed some software that merges agents with GIS for simulating recreation behavior in complex wilderness settings. We examined SWARM as the modeling framework, but to get a prototype up and running developed our own software in Visual Basic 4.0 running under windows 95. A version of the software referred to as RBSim - Recreation Behaviour Simulator and is available with enought documenation from the web: http://www.dlsr.com.au/software/rbsim The prototype is currently being written in SWARM by Bohdan Durnota in Australia. For more detailed information on the software you can contact with Bob Itami at the above web site or Randy Gimblett (gimblett@ag.arizona.edu). Randy Gimblett ___________________________________________________________________ Randy Gimblett Associate Professor School of Renewable Natural Resources The University of Arizona Tucson, Arizona 85721 USA Email: gimblett@nexus.srnr.arizona.edu World Wide Web: http://nexus.srnr.arizona.edu/~gimblett Phone: (520) 621-6360 Fax: (520) 621-8801 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 21:47:39 1997 From: swarm-modelling@santafe.edu (Sven N. Thommesen) Date: Thu, 03 Apr 1997 15:47:39 -0600 Subject: Catalog of agents Message-ID: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> At 09:27 AM 4/3/97 -0500, Brian Ruth wrote: > >I'm currently a Swarm user wannabe (I'm waiting for the source+binary >release for SGIs), but I have looked a bit into modeling crowd behavior. >Dana Eckart of Radford University and I are currently developing a >cellular automata model, using his Cellular simulation system (available >at > > http://rucs2.sunlab.cs.runet.edu/~dana/ca/cellular.html), > Brian, thanks for that reference to Prof. Eckart's ca stuff. Very interesting. Worth looking at! Which leads me to a suggestion for 'someone' to do: the essence of Swarm modelling is, of course, the behavioral methods of our agents. I'd like to see a web site collect a catalog of different behavioral methods people have used, described in pseudo-code and/or source code. (Similar to Prof. Eckart's collection of CA models.) This would allow others to test out the posted methods, and to critique them. Over time, we might get an idea of which algorithms are useful and which not for given applications. Any takers ? (If none, I may do it myself after random is done ...) Sven ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 23:07:10 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Thu, 3 Apr 1997 16:07:10 -0700 Subject: What *is* Time? (was Re: re getCurrentTime(): solved) In-Reply-To: <3.0.32.19970403153149.009a2100@spidle2.humsci.auburn.edu> References: <3.0.32.19970403153149.009a2100@spidle2.humsci.auburn.edu> Message-ID: <199704032307.QAA14420@grasshopper.santafe.edu> I moved this to the modelling list because it's really a very large fuzzy issue versus an implementation issue. I realize that the boundaries between the two lists are not well defined, yet; but, I'm trying to help that de-fuzzification. Sven N. Thommesen writes: > Rick has accurately described my initial expectations, > given that (time-wise, at least) my app is not more > complex than Heatbugs. Then I suppose that means we should re-evaluate the docs and what they explain. Of course, we'll need y'all to help us! > I can now hit my button before the sim starts, while > it is running, and while it is stopped, without any > rude surprises. Good! The point of the whole exercise is to get you to where you can do what you want, eh? > PS: Perhaps Glen could further muddy the waters: > is it, in Swarm, possible for different (sub-)Swarms to > have their own sense of time get out of sync with > each other? [And how does the answer depend on > whether we have serial or parallell Swarm? :-) ] > If they do, what is the relationship between such > sub-swarm clocks and any master Global Time clock? > > If the answer is yes, they *can* get out of sync, > then perhaps the macro is a dangerous thing that > shouldn't be there? I can always count on Sven to just cavalierly pop open that can o' worms. First, the short form (correct me if I get anything wrong, Roger): There is a "Relative Time" for each Swarm. This is probably the time that should be used inside subswarms. For instance, the modelSwarm has a "modelTime" that the agents inside that swarm should access. These relative times will almost certainly get out of sync in a || Swarm. But, that concern should be handled by the constraints that will be specified for the given schedules inside each subswarm. Examples of these constraints are the ability to have a a schedule in a swarm containing a sequence of ConcurrentGroups with "DefaultOrders" of Concurrent, Sequential, or Randomized. The long form: There has been alot of consideration given to how these models are integrated. And while time may seem like an inherent feature of what is happening in Swarm, it isn't intended as such. What we're calling the "logical model of concurrency" in Swarm is based on partially ordered sets. (See *) This basis can be considered completely independent of time. Just because all processes implicitly contain a concept of the "passage of time," doesn't mean they have to refer to or be based upon time. So, it's completely reasonable to think about all processes in terms of the order of events. These events can be ordered by any constraint imaginable.... heat index, risk dependency, relationships, etc. Now that i've gone quite off the deep end, I'll return to the question: "Is it possible for different (sub-)Swarms to have their own sense of time get out of sync with each other?" Not only is it possible, but, it's preferable, since most real systems have subsystems with clocks out of sync with each other. They have sync points, obviously. One perfect example is a human-computer interface. The computer's polls on, say, the keyboard device are much much faster than the human's strikes of the keys. In fact, the keystrokes of the human are some neurally generated sequence with some (hopefully) random noise, which is certainly not "sync'ed" with the computer's high frequency poll. But, the two processes are forcibly synced because any keystroke that may actually happen in between polling events of the computer waits for the next cycle of the computer. This is an extremely simple synchronization. Much much more complicated sycing goes on in virtually any system we might want to model. So, on to the next point: "perhaps the macro is a dangerous thing that shouldn't be there?" Well, scissors are also dangerous in the hands of running children? [grin] Seriously, yes I think the macro is very dangerous. But, I can't imagine telling users that there's no way to get the absolute time of the simulation. And, it is a useful tool, especially if you have ever worked inside the paradigm where your simulation contains a "truth" model and a "system" model. In this type of simulation, "error" plays a large role... And one can only have "error" if one has "truth." So, the macro should stay. And it is dangerous. But, it's a useful tool. We could either let new users run with the scissors and learn the error of there ways after a couple of untoward accidents.... Or we could prescribe restrictions on their behaviour for reasons they don't understand and expect them to grow into that understanding. Both work; but, you come off as limited and pretentious in the latter. glen p.s. I'm aware that this was not well-written and may, as Sven put it, "muddy the waters;" but, since I'm in manual writing phase, I felt it was a good chance to just toss out some things that need to be said without being too self-critical of the way they're said. Any responses will help mold the way this subject is handled in the manual. * Sets upon which relationships are defined that don't have relevance to all members of the set. E.g. I can have a set of objects, some of which are cows and some of which are cars. I can have a partial order relation, say, "produces-better-milk-than". Then it makes complete sense to say "cow1 produces-better-milk-than cow2"; but, it makes no sense to say "car1 produces-better-milk-than car2". But, it certainly also makes sense to say that "cow2 produces-better-milk-than car1". ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 18:08:40 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Mon, 7 Apr 1997 18:08:40 +0100 (BST) Subject: Modeling of Crowd Behavior In-Reply-To: Message-ID: On Thu, 3 Apr 1997, Brian Ruth (CBNED/SMSB) wrote: > I'm currently a Swarm user wannabe (I'm waiting for the source+binary > release for SGIs), but I have looked a bit into modeling crowd behavior. > Dana Eckart of Radford University and I are currently developing a > cellular automata model, using his Cellular simulation system (available > at > > http://rucs2.sunlab.cs.runet.edu/~dana/ca/cellular.html), > > which demonstrates the emergence of panic within a unit of soldiers when > exposed to one or more battlefield threats. The model determines whether > a particular soldier will panic by assigning a probability of panic > conditional on the number of wounded and/or panicking neighbors within the > soldier's extended Moore neighborhood (21 x 21 cells on a 2D lattice), > and then performs a random draw to determine the soldier's panic state > (panicking/not panicking). Flocking behavior is also considered within > the unit (for non-panicking soldiers only) , where a soldier's speed and > direction of travel is determined by that of his neighbors. Why shouldn't panicking soldiers flock as well? Don't people who flock panic and people who panic flock? Just wondering, Jan. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 19:15:23 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Mon, 7 Apr 1997 12:15:23 -0600 Subject: Catalog of agents In-Reply-To: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> References: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> Message-ID: <199704071815.MAA16011@grasshopper.santafe.edu> Sven N. Thommesen writes: > Which leads me to a suggestion for 'someone' to do: > > the essence of Swarm modelling is, of course, the behavioral > methods of our agents. I'd like to see a web site collect > a catalog of different behavioral methods people have used, > described in pseudo-code and/or source code. (Similar to > Prof. Eckart's collection of CA models.) This would allow > others to test out the posted methods, and to critique them. > Over time, we might get an idea of which algorithms are > useful and which not for given applications. > > Any takers ? This is an excellent idea. Of course, more than just the agent methods should be described, I would guess. But, it might be reasonable to compress the "essence" of a model with descriptive pseudo-code for the agent types and their respective methods, and the environment design and it's respective methods. The one element left is the scheduling. Developing compact descriptions of these three elements might even help us with the Schedule-language specification when and if we start working seriously on that. A language needs well-specified data types (agents and environments) as well as well-specified operator types. As far as takers for the task.... Brad? glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 22:23:53 1997 From: swarm-modelling@santafe.edu (Jae Chan Oh) Date: Mon, 7 Apr 1997 17:23:53 -0400 (EDT) Subject: Which Machine for swarm: Pentium Pro, or SUN Ultra 140? In-Reply-To: <199704071815.MAA16011@grasshopper.santafe.edu> from "glen e. p. ropella" at Apr 7, 97 12:15:23 pm Message-ID: <199704072123.RAA00927@homer.cs.pitt.edu> Hi, We have choice of buying either Pentium Pro or Sun Ultra 140 for swarm application developement. Which machine do you think we should go with? (In terms of computing power, swarm compatibility/friendliness, etc.) Does anyone have any preference between the two? The Specs are: Pentium-Pro: 200 MHz, 64 M Main Memory, 3.8 G HD, SoundBlaster, CD Player, 17" Monitor, Cost around $3,000. (we can add extra 64Meg) Sun Ultra Sparc 140: 143 MHz, 128 Meg, 2.1 G HD, CD player, 17" monitor, Cost around $5000 We will use Linux for Pentum-Pro and Solaris for SUN, of course. Thanks, -Jae ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 23:00:35 1997 From: swarm-modelling@santafe.edu (Brian Haugh) Date: Mon, 7 Apr 1997 18:00:35 -0400 Subject: Which Machine for swarm: Pentium Pro, or SUN Ultra 140? In-Reply-To: Jae Chan Oh "Which Machine for swarm: Pentium Pro, or SUN Ultra 140?" (Apr 7, 5:23pm) References: <199704072123.RAA00927@homer.cs.pitt.edu> Message-ID: <9704071800.ZM1551@fasolt.csed.ida.org> We have found that our Swarm application runs faster on a Pentium Pro than on a Sun Ultra 1/140. Unfortunately, Swarm can't take any advantage of the Ultra's 64 bit architecture since the gcc Objective-C compiler does not compile for it. Of course you can compile with gcc on an UltraSparc with gcc, but the code is no different from a regular Sparc. Thus, the main difference in performance is just due to clock speed - since the Pro's clock is faster than an Ultra 1 at 140 Mhz, then the code run's faster. Our app runs at about the same speed on a Ultra Enterprise 2 200Mhz server as the Pentium Pro, but the Ultra 2 is a bit more expensive. So, if Swarm performance is the only consideration, I would recommend the Pentium Pro under Linux. If you are going to use the system for other applications, though, the Ultra might have some advantages. The Ultra is especially fast on multi-media, i.e., video compression/ decompression, from what I've read, due to special parallel execution for MPEG instructions. Brian -- Brian A. Haugh, Ph.D. Institute for Defense Analyses Computer & Software Engineering Division 1801 North Beauregard Street Alexandria, Virginia 22311-1772 phone: (703) 845-6678 fax: (703) 845-6788 email: bhaugh@ida.org ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 23:21:56 1997 From: swarm-modelling@santafe.edu (Randall Gray) Date: Tue, 8 Apr 1997 08:21:56 +1000 (EST) Subject: Which Machine for swarm: Pentium Pro, or SUN Ultra 140? In-Reply-To: <199704072123.RAA00927@homer.cs.pitt.edu> (message from Jae Chan Oh on Mon, 7 Apr 1997 17:23:53 -0400 (EDT)) Message-ID: <199704072221.IAA01474@njal.ml.csiro.au> > From: Jae Chan Oh > Date: Mon, 7 Apr 1997 17:23:53 -0400 (EDT) > We have choice of buying either Pentium Pro or Sun Ultra 140 for swarm > Does anyone have any preference between the two? In Australia (at least) the answer is fairly easy: go with the intel based machine. We recently purchased a dual processor machine with roughly 2/3 the grunt of our "central number crunching" machines at work for about 1/5 the price tag. Buy the extra memory for the PC -- it is even nicer. Linux can make huge disk buffers when there is no other demand for the memory and this is a *very* nice use which speeds things up quite a lot. > We will use Linux for Pentum-Pro and Solaris for SUN, of course. Linux is available on SPARCS: there is a port to Ultras which is running (though still in development). Apparently it flys. Dunno what the exchange rate is today, so I guess this is worth what ever I get for it ;-) -- Randall ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 02:41:51 1997 From: swarm-modelling@santafe.edu (Todd Allen) Date: Mon, 07 Apr 1997 21:41:51 -0400 Subject: Which Machine for swarm: Pentium Pro, or SUN Ultra 140? References: <199704072123.RAA00927@homer.cs.pitt.edu> Message-ID: <3349A25F.4E79@jhunix.hcf.jhu.edu> Jae, For the money the Pentium is the clear choice. I run a dual Pro-200 machine under Linux 2.0 at home and a Sun Sparc Ultra 1 at school and have found that the performance is nearly identical when running Swarm. With the huge drop in pentium pro 200 prices this week due to the release of the 233 and 266 processors, I'd recommend getting a dual pentium pro box. Swarm can't directly take advantage of the second processor, but performance is still enhanced due to the off-loading of much of the OS and X windows overhead onto the second processor. If you're going to go with a single processor you may want to look into one of the new faster processors. The Pentium II 233 is selling for about what the Pro-200 was a month ago, and the 266 is about a hundred more. Regards, Todd -- Todd Allen The Johns Hopkins University Department of Economics tallen@jhu.edu 410-516-7571 (office) "Rational expectations is rigorous deduction based upon faulty assumptions." - Brian Arthur ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 16:50:41 1997 From: swarm-modelling@santafe.edu (Brian Ruth (CBNED/SMSB) ) Date: Tue, 8 Apr 1997 11:50:41 -0400 (EDT) Subject: Modeling of Crowd Behavior In-Reply-To: Message-ID: On Mon, 7 Apr 1997, Jan Kreft wrote: > Why shouldn't panicking soldiers flock as well? Don't people who flock > panic and people who panic flock? > > Just wondering, > > Jan. Naturally they do (in a manner of speaking ;-)). The flocking of panicking soldiers (or people in general) should, however, probably reflect less centralization and more random variations relative to the flocking of non-panicking soldiers (people). Brian ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 17:43:13 1997 From: swarm-modelling@santafe.edu (John A. Lopez) Date: Tue, 08 Apr 1997 09:43:13 -0700 Subject: Catalog of agents References: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> <199704071815.MAA16011@grasshopper.santafe.edu> Message-ID: <334A759F.2A67@pop.calweb.com> glen e. p. ropella wrote: > > Sven N. Thommesen writes: > > Which leads me to a suggestion for 'someone' to do: > > > > the essence of Swarm modelling is, of course, the behavioral > > methods of our agents. I'd like to see a web site collect > > a catalog of different behavioral methods people have used, > > described in pseudo-code and/or source code. (Similar to > > Prof. Eckart's collection of CA models.) This would allow > > others to test out the posted methods, and to critique them. > > Over time, we might get an idea of which algorithms are > > useful and which not for given applications. > > > > Any takers ? > > This is an excellent idea. Of course, more than just the > agent methods should be described, I would guess. But, > it might be reasonable to compress the "essence" of a > model with descriptive pseudo-code for the agent types > and their respective methods, and the environment design > and it's respective methods. The one element left is the > scheduling. > > Developing compact descriptions of these three elements > might even help us with the Schedule-language specification > when and if we start working seriously on that. A language > needs well-specified data types (agents and environments) > as well as well-specified operator types. > > As far as takers for the task.... Brad? > > glen Sven and Glen, I think this is more than an excellent idea, since this list would necessarily evolve to discipline-specific lists. Such posted methods could be improved upon or varied in ways which could form the basis of a dialog (parallelog?) within a research discipline (mine is anthropology/archaeology). I'd be more than happy to post my kinship, mimicry, and the other behavioral methods I've developed or am working on. In light of the recent discussion regarding future funding and non-profit organization structure, we could take a lesson from Swarm. This idea of a methods listing could be a source for SFI income: users could pay some individual or institutional fee for specific downloads of methods (available free to contributers within a class of course). Perhaps the SFI folks could continue to make its server available and act as a GA and cull those discussions which have little or no interest. John ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 17:47:56 1997 From: swarm-modelling@santafe.edu (John A. Lopez) Date: Tue, 08 Apr 1997 09:47:56 -0700 Subject: Modeling of Crowd Behavior References: Message-ID: <334A76BA.5E72@pop.calweb.com> Brian Ruth (CBNED/SMSB) wrote: > > On Mon, 7 Apr 1997, Jan Kreft wrote: > > > Why shouldn't panicking soldiers flock as well? Don't people who flock > > panic and people who panic flock? > > > > Just wondering, > > > > Jan. > > Naturally they do (in a manner of speaking ;-)). The flocking of > panicking soldiers (or people in general) should, however, probably > reflect less centralization and more random variations relative to > the flocking of non-panicking soldiers (people). > > Brian Jan & Brian, I think paniced people give up their individual attributes for the global elements and consequences governing the crowd. Two cents worth. John ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 19:19:46 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 8 Apr 1997 12:19:46 -0600 Subject: Catalog of agents In-Reply-To: <334A759F.2A67@pop.calweb.com> References: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> <199704071815.MAA16011@grasshopper.santafe.edu> <334A759F.2A67@pop.calweb.com> Message-ID: <199704081819.MAA16505@grasshopper.santafe.edu> John A. Lopez writes: > I think this is more than an excellent idea, since this list would > necessarily evolve to discipline-specific lists. Such posted methods > could be improved upon or varied in ways which could form the basis of a > dialog (parallelog?) within a research discipline (mine is > anthropology/archaeology). I'd be more than happy to post my kinship, > mimicry, and the other behavioral methods I've developed or am working > on. > > In light of the recent discussion regarding future funding and > non-profit organization structure, we could take a lesson from Swarm. > This idea of a methods listing could be a source for SFI income: users > could pay some individual or institutional fee for specific downloads of > methods (available free to contributers within a class of course). > Perhaps the SFI folks could continue to make its server available and > act as a GA and cull those discussions which have little or no > interest. Hmmmm. This would complement the idea that someone had during SwarmFest of providing an archiving service of source code. The suggestion was that, since part of Swarm's purpose is to help with the documentation and repeatability aspects of doing research via simulation, it would help to have some repository in which simulants could deposit their source code for posterity. Along these lines, it would be reasonable to assume that the swarm.org organize and manage this repository as well as some kind of Agent-Reuse library/catalog, as mentioned above. Charging for this service wouldn't be a bad idea, at all. There are three ways that could be done: 1) charge the researcher for archiving his stuff (this wouldn't be effective for the agent catalog but might be ok for registering simulations), 2) charging for access to the database, or 3) grant style funding, where money is sought from institutions who have an interest in making simulation more mature. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 11 19:14:35 1997 From: swarm-modelling@santafe.edu (Phil Knight) Date: Fri, 11 Apr 1997 19:14:35 +0100 Subject: modelling CAs in swarm? Message-ID: I'm a delurking swarm user who rates somewhere just over 3 on Glen's user scale (3 - those who have successfully installed Swarm and played with the provided demos, but haven't made their own models, 4 - those who are actively using Swarm). I'm interested in using swarm to explore cellular automata, specifically to model physical phenomena (i.e. digital physics or Fredkin's universe as a CA). In order to get started I set up a sim of Conway's Life. This is literally a two minute job to wrap a model swarm around the pre- supplied life class. Since I'm likely to want to utilise relatively large lattice spaces (ultimately in 3d), and the life class is implemented with a bog standard 2d array, I set about cobbling together a sparse matrix version of life and I've now succeeded in getting this up and running. The initial world size is specified and seeded with the desired probability, the initial world area is constantly displayed on a raster, but the lattice iteslf extends to 32k x 32k (it should be possible to increase this even further). In actual fact the app will also run more than just the life rule, allowing any 2d, binary, totalistic, Moore neighborhood CA to be simulated by specifying the appropriate birth/survival rule before running. For anyone who's interested, both apps are available at: http://www.pknight.demon.co.uk/ I should emphasise though that they are very hacked together and not intended to be the best or most efficient means of construction. At this stage I'm just trying to explore methods of modelling CAs on large lattices rather than providing serious models. Any constructive criticism is more than welcome. The real purpose of this post is to ask firstly if anyone else here is using swarm for CAs, especially on large lattices and if so, the approaches being used in the models. Secondly, and perhaps more importantly, one area that I'm really going to want to work on is visualisation tools (esp. 3d). For example, the sims above only display the initial lattice size, but the "active" lattice in the sparse version quickly becomes much larger, so there is a requirement to be able to zoom in/out of different areas of the lattice. I haven't yet thought too much about 3d visualisation tools as yet, but hope this post might stimulate some discussion :-) -- Phil Knight ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sun Apr 13 17:12:11 1997 From: swarm-modelling@santafe.edu (Brad Leydorf) Date: Sun, 13 Apr 1997 12:12:11 -0400 Subject: Behavior modelling?? References: Message-ID: <335105DB.240E@one.net> Is anyone out there in Swarmland into behavior modelling? (of people). If so, I'd love to hear any suggestions or cautions you might have... Thanks, Brad bleydorf@one.net ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sun Apr 13 22:58:42 1997 From: swarm-modelling@santafe.edu (Pietro Terna) Date: Sun, 13 Apr 1997 23:58:42 +0200 Subject: modelling CAs in swarm? Message-ID: <1.5.4.32.19970413215842.0069f7bc@alpcom.it> Hi Phil, I have tested your helpful examples, many thanks. Only conwaylife2d has been creating some problem in my Linux box. I my directory (SWARMHOME)/include/graph/ I just had a ConwayLife2d.h file, which resulted imported in your application via #import so generating errors such as 'duplicate interface declarations' and, finally, no linking. My solution hacked solution has been that of modiying all instances of ConwayLife2d in ConwayLifeBis2d in your package and so ... all is running. Yours, Pietro At 19.14 11/04/97 +0100, you wrote: >I'm a delurking swarm user who rates somewhere just over 3 on Glen's >user scale (3 - those who have successfully installed Swarm and played >with the provided demos, but haven't made their own models, 4 - those >who are actively using Swarm). > >I'm interested in using swarm to explore cellular automata, specifically >to model physical phenomena (i.e. digital physics or Fredkin's universe >as a CA). In order to get started I set up a sim of Conway's Life. This >is literally a two minute job to wrap a model swarm around the pre- >supplied life class. > >Since I'm likely to want to utilise relatively large lattice spaces >(ultimately in 3d), and the life class is implemented with a bog >standard 2d array, I set about cobbling together a sparse matrix version >of life and I've now succeeded in getting this up and running. The >initial world size is specified and seeded with the desired probability, >the initial world area is constantly displayed on a raster, but the >lattice iteslf extends to 32k x 32k (it should be possible to increase >this even further). In actual fact the app will also run more than just >the life rule, allowing any 2d, binary, totalistic, Moore neighborhood >CA to be simulated by specifying the appropriate birth/survival rule >before running. > >For anyone who's interested, both apps are available at: >http://www.pknight.demon.co.uk/ >I should emphasise though that they are very hacked together and not >intended to be the best or most efficient means of construction. At this >stage I'm just trying to explore methods of modelling CAs on large >lattices rather than providing serious models. Any constructive >criticism is more than welcome. > >The real purpose of this post is to ask firstly if anyone else here is >using swarm for CAs, especially on large lattices and if so, the >approaches being used in the models. > >Secondly, and perhaps more importantly, one area that I'm really going >to want to work on is visualisation tools (esp. 3d). For example, the >sims above only display the initial lattice size, but the "active" >lattice in the sparse version quickly becomes much larger, so there is a >requirement to be able to zoom in/out of different areas of the lattice. >I haven't yet thought too much about 3d visualisation tools as yet, but >hope this post might stimulate some discussion :-) > >-- >Phil Knight > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 12 20:45:46 1997 From: swarm-modelling@santafe.edu (Brad Leydorf) Date: Sat, 12 Apr 1997 15:45:46 -0400 Subject: anyone behavior modelling? References: Message-ID: <334FE66A.1729@one.net> Is anyone out there in Swarmland into behavior modelling? (of people). If so, I'd love to hear any suggestions or cautions you might have... Thanks, Brad bleydorf@one.net ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 16 20:27:09 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Wed, 16 Apr 97 13:27:09 MDT Subject: Simulating Individual Behavior References: <199704161351.HAA00352@seamus.dischordia.com> Message-ID: <9704161927.AA12178@sfi.santafe.edu> I would be interested in knowing how the SWARM project might impact social simulation. We can start this off as a research discussion, and then perhaps, this will help to stop the flood of "unsubscriptions." The idea of simulating a system by going down to the atomic level is something that most disciplines have dealt with in one form or another. Consider, for example, ecosystem modeling. In ecosystem modeling, there is some controversy about individual-based modeling (IBM) and yet this modeling approach holds promise in predicting the evolution of a ecosystem. A key problem with IBM is not that it is necessarily computationally prohibitive, but that not enough data are available to calibrate the model, or individuals participating within the model. If this is a problem for modeling alligators or wood storks, I would venture that the problem for human systems would be manifold. What kind of model is used to model the human. A logical choice of model type for human objects in the simulation is one based on AI techniques: rule-based or operator based systems. On the other hand, what level of detail is required to model the human element? Perhaps, it need not be detailed. A good example of high level modeling of humans can be found in emergency planning simulation, which is a subject of interest in the Society for Computer Simulation since they have sponsored many conferences in this area. If you have a hazard or peril within an enclosed structure, how should the humans react? Modeling the human in this instance could be much easier than modeling for a less-streneous social goal. Thoughts? -paul -- Paul A. Fishwick E-Mail: fishwick@cise.ufl.edu Dept. of Computer & Info Phone & FAX: (352) 392-1414 Science and Engineering WWW: http://www.cise.ufl.edu/~fishwick University of Florida (PGP Key available at above WWW address) P. O. Box 116120 332 Bldg. CSE, Gainesville, FL 32611-6120 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 11:50:51 1997 From: swarm-modelling@santafe.edu (Philippe LAVAL) Date: Thu, 17 Apr 1997 09:50:51 -0100 Subject: Simulating Individual Behavior Message-ID: <3.0.32.19970417095049.006ebc8c@poseidon.obs-vlfr.fr> Paul Fishwick said : >A key problem with IBM is not that it is necessarily >computationally prohibitive, but that not enough data are >available to calibrate the model... Perhaps you are talking about ecosystems from your simulation-oriented point of view, as the word 'calibrate' indicates. I think that a complex system like an ecosystem cannot be modeled with some large FSM or even with a set of stochastic equations (except in the short-term): in an ecosystem there is probably no unique latent 'model' to be discovered, against which we could compare some data and make adjustments. Instead there probably exists many possible modes, with the system unpredictably switching from time to time to one or another. The first thing to do could be to try to recognize these modes. Because switching is unpredictable (ecosystems seem to exhibit self-organized critical modes), it makes no sense to try to 'calibrate' the whole system, in the same manner simulation engineers calibrate a model of a factory or a flexible workshop. -------------------------------------------------------------------- Philippe Laval Station zoologique B.P. 28 - 06234 Villefranche-sur-Mer CEDEX (France) laval@ccrv.obs-vlfr.fr ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 11:46:50 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: Thu, 17 Apr 1997 04:46:50 -0600 Subject: Simulating Individual Behavior Message-ID: <199704171046.EAA21778@santafe.santafe.edu> Paul Fishwick wrote: >A key problem with IBM is not that it is necessarily >computationally prohibitive, but that not enough data are >available to calibrate the model... Paul - why is data calibration more of a problem for IBM models than for other modelling technologies? What sorts of problem domains are you particularly concerned about? I can allow as how it would be difficult to calibrate an IBM model of a forest precisely so that each tree-agent is parameterized via data taken from its respective real-tree in the forest. Yet, all modelling technologies make do with some degree of approximation. A reservoir-flow model of tree-species interaction in a forest would simply treat all of the trees of each species as "one big tree" of that species with respect to some data (such as concentration, nutrient uptake, waste-production, and etc.) while ignoring other data (such as spatial distribution, variety within the species, and etc.) This will be justified for certain questions about forest dynamics, but not for others, and might make more sense for some problem domains than for others. Thus, there is data and there is data. All modeling technologies must pick and choose among the data, and one always has to focus on some reasonable subset of data. You seem to be suggesting that this is fundamentally more of a problem for IBM models than for other modelling technologies - can you elaborate? ...and, please!, it almost *hurts* to use the acronym IBM! could we use "ABM" for Agent-Based Models? I think it fits better anyway, as an agent in this class of models is not always an "individual" in the common sense of that term.... I know the term has some historical precedent for models in this class, but the acronym IBM induces a certain amount of, shall we say, cognitive dissonance, no? (not that the acronym "ABM" itself is inviolate with respect to prior cognitive content....but, still!.....) Chris Langton ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 18:43:53 1997 From: swarm-modelling@santafe.edu (Mike Brown) Date: 17 Apr 1997 09:43:53 -0800 Subject: Simulating Individual B Message-ID: I think Paul's question raises two important points - though they may be more about the organization of science than about the inherent difficulties of SWARM or any other ABM (Thanks, Chris. I like that acronym better too.) Ecologists are concerned with aggregations of individuals; they have not had to know as much about individual species as, for example, zoologists. Similarly, microeconomists focus on the behavior of individual firms and macroeconomists on the aggregate behavior of the economy, etc. ABM create two problems. First, the ecologist and macroeconomist have to "ratchet down" and study the behavior of individual entities. Moreover, they have to know enough about individual behavior to determine which specific behaviors might be relevant to the question under investigation. While this is not an insurmountable problem, it does demand a new focus for researchers. Second, there are some disciplines where data on the behavior of individual "agents" simply has not been studied. To take a bad example, look at economics. Macro studies the behavior of the aggregate, and micro the study of the firm -- but who has been looking at the behavior of the consumer? We have been able to make a "rational actor" assumption for so long that we have not bothered to collect data about the real-life behavior of induividual consumers. For these reasons, I think Paul is very right -- modeling and validating the behavior of individual entities can be very tricky. Mike ------------------------------ Date: 4/17/97 4:47 AM To: Brown, Mike From: swarm-modelling@santafe.edu Paul Fishwick wrote: >A key problem with IBM is not that it is necessarily >computationally prohibitive, but that not enough data are >available to calibrate the model... Paul - why is data calibration more of a problem for IBM models than for other modelling technologies? What sorts of problem domains are you particularly concerned about? I can allow as how it would be difficult to calibrate an IBM model of a forest precisely so that each tree-agent is parameterized via data taken from its respective real-tree in the forest. Yet, all modelling technologies make do with some degree of approximation. A reservoir-flow model of tree-species interaction in a forest would simply treat all of the trees of each species as "one big tree" of that species with respect to some data (such as concentration, nutrient uptake, waste-production, and etc.) while ignoring other data (such as spatial distribution, variety within the species, and etc.) This will be justified for certain questions about forest dynamics, but not for others, and might make more sense for some problem domains than for others. Thus, there is data and there is data. All modeling technologies must pick and choose among the data, and one always has to focus on some reasonable subset of data. You seem to be suggesting that this is fundamentally more of a problem for IBM models than for other modelling technologies - can you elaborate? ...and, please!, it almost *hurts* to use the acronym IBM! could we use "ABM" for Agent-Based Models? I think it fits better anyway, as an agent in this class of models is not always an "individual" in the common sense of that term.... I know the term has some historical precedent for models in this class, but the acronym IBM induces a certain amount of, shall we say, cognitive dissonance, no? (not that the acronym "ABM" itself is inviolate with respect to prior cognitive content....but, still!.....) Chris Langton ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== ------------------ RFC822 Header Follows ------------------ Received: by cpqm.saic.com with ADMIN;17 Apr 1997 04:44:33 -0800 Return-Path: Received: from sfi.santafe.edu by cpmx.mail.saic.com; Thu, 17 Apr 97 04:45:16 -0700 Received: by sfi.santafe.edu (4.1/SMI-4.1) id AA21897; Thu, 17 Apr 97 04:46:49 MDT Date: Thu, 17 Apr 1997 04:46:50 -0600 From: cgl@santafe.edu Message-Id: <199704171046.EAA21778@santafe.santafe.edu> To: swarm-modelling@santafe.edu Subject: Re: Simulating Individual Behavior Sender: owner-swarm-modelling@santafe.edu Precedence: bulk Reply-To: swarm-modelling@santafe.edu ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 16:12:57 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: Thu, 17 Apr 1997 09:12:57 -0600 Subject: Simulating Individual B Message-ID: <199704171512.JAA22183@santafe.santafe.edu> Mike Brown writes: > I think Paul's question raises two important points - though they may be more > about the organization of science than about the inherent difficulties of > SWARM or any other ABM (Thanks, Chris. I like that acronym better too.) > Ecologists are concerned with aggregations of individuals; they have not had > to know as much about individual species as, for example, zoologists. > Similarly, microeconomists focus on the behavior of individual firms and > macroeconomists on the aggregate behavior of the economy, etc. Excellent point, and I do think that it is more about the organization of science than inherent problems with models... In many scientific fields (but ecology and economics will serve nicely as examples), there are those (usually theorists) who study the macro behavior, and those (usually field workers) who study the micro behavior - and the two communities rarely interact with each other. The models built by each community often simply make enormously simplified assumptions about the other's results as the starting point and/or boundary conditions for their own models. One of the main reasons I am interested in these ABM models is that they provide an opportunity to close the loop and look at the way in which the behavior treated by the theorists at the macro level is grounded in the micro-level behaviors treated by the field workers - it all has to come together here - the macro behavior should emerge out of the micro behavior, as it does in the natural system. Thus these ABM models tend to be "meso"-scale, bigger than the small slices of systems typically treated by the field-workers, but necessarily smaller than the full scale macro-system: just big enough to study the way in which the macro emerges out of the micro, but not so big that it rivals the real system in all its unwieldy scale... It is precisely because these models can serve as a common venue for both the macro and micro communities in a discipline that they are of so much interest (and, I think, importance). > First, the ecologist and macroeconomist have to > "ratchet down" and study the behavior of individual entities. Moreover, they > have to know enough about individual behavior to determine which specific > behaviors might be relevant to the question under investigation. While this is > not an insurmountable problem, it does demand a new focus for researchers. > Second, there are some disciplines where data on the behavior of individual > "agents" simply has not been studied. To take a bad example, look at > economics. Macro studies the behavior of the aggregate, and micro the study of > the firm -- but who has been looking at the behavior of the consumer? We have > been able to make a "rational actor" assumption for so long that we have not > bothered to collect data about the real-life behavior of induividual > consumers. > For these reasons, I think Paul is very right -- modeling and validating the > behavior of individual entities can be very tricky. Right again - however I consider this as a feature rather than a bug! I think it gets at the essence of what "modelling" is all about... The term "model" means a lot of different things to different people. However, I find that the most useful characterization of a model is as an artifact that you construct that helps you think about the system under study - this can range from obviously over-simplified models that simply help you understand what data you still lack, and what parameters might be the interesting ones, to a full blown *theory* that claims to explain exactly how the real system works. The point is that in building these ABM models, often the first thing we learn from them is that we have to go back to the real system with new questions....but this is good! It is very useful to have a tool that provides you with a new way to look at your problem and which generates new questions about it. Mike (and Paul) may be right about the general lack of data, although I'm willing to bet that some of this perception is due to the lack of communication between the macro and micro scale research communities. But, I also think that this is because we haven't been thinking about some of these natural systems in the right way, I think ABM models provide the proper venue to help us think about the models in the right way, and if the result of that help in thinking is to force us to collect new data about the system, then that's a useful and necessary lesson.... So - again, I don't think of this as a problem, but a feature of these models......and I don't think of it as unique to ABM models, per se - I think new tools often offer us new "opportunities" to collect data... Chris Langton ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 16:51:50 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Thu, 17 Apr 1997 16:51:50 +0100 (BST) Subject: Simulating Individual Behavior In-Reply-To: Message-ID: OK, I agree with Paul and Mike that it can be very tricky to determine the parameters for individuals. Bacteria as individuals, e. g., are simply so small that only the latest high-tech methods could give estimates for the parameters you would have to use in a model (and that costs a lot of timemoney). Almost always microbiologists measure the population average and disregard the heterogeneity of the population. But I can see another way to solve the problem. If you start with a conceptual model with all the parameters and mechanics you need and only a rough idea of actual values (from population average measurements) you can run the sim and compare the output with the real system to be modeled. Then you optimize the parameters to achieve the desired output. Perhaps call it back-calibration. (Is there an official term for it?) For that purpose, you need a parameter manager (see previous discussion with "parameter" in the subject line) to allow a search through param space in an evolutionary fashion. As far as I can tell, Gecko's param manager is (the only one?) suited for that task. BTW, does IbM hurt also? If so, sorry for that ;-). Point being is that that's the term used in the literature and if I want to search for new literature, I rely on this keyword. Therefore I would want to use this keyword myself sometime. Cheers, Jan. Mike Brown wrote: > I think Paul's question raises two important points - though they may be more > about the organization of science than about the inherent difficulties of > SWARM or any other ABM (Thanks, Chris. I like that acronym better too.) > Ecologists are concerned with aggregations of individuals; they have not had > to know as much about individual species as, for example, zoologists. > Similarly, microeconomists focus on the behavior of individual firms and > macroeconomists on the aggregate behavior of the economy, etc. > > ABM create two problems. First, the ecologist and macroeconomist have to > "ratchet down" and study the behavior of individual entities. Moreover, they > have to know enough about individual behavior to determine which specific > behaviors might be relevant to the question under investigation. While this is > not an insurmountable problem, it does demand a new focus for researchers. > > Second, there are some disciplines where data on the behavior of individual > "agents" simply has not been studied. To take a bad example, look at > economics. Macro studies the behavior of the aggregate, and micro the study of > the firm -- but who has been looking at the behavior of the consumer? We have > been able to make a "rational actor" assumption for so long that we have not > bothered to collect data about the real-life behavior of induividual > consumers. > > For these reasons, I think Paul is very right -- modeling and validating the > behavior of individual entities can be very tricky. > > Mike > > Paul Fishwick wrote: > > >A key problem with IBM is not that it is necessarily > >computationally prohibitive, but that not enough data are > >available to calibrate the model... > > Paul - why is data calibration more of a problem for IBM > models than for other modelling technologies? What sorts > of problem domains are you particularly concerned about? > > I can allow as how it would be difficult to calibrate > an IBM model of a forest precisely so that each tree-agent > is parameterized via data taken from its respective > real-tree in the forest. Yet, all modelling technologies > make do with some degree of approximation. A reservoir-flow > model of tree-species interaction in a forest would simply > treat all of the trees of each species as "one big tree" > of that species with respect to some data (such as concentration, > nutrient uptake, waste-production, and etc.) while ignoring > other data (such as spatial distribution, variety within the > species, and etc.) This will be justified for certain questions > about forest dynamics, but not for others, and might make more > sense for some problem domains than for others. > > Thus, there is data and there is data. All modeling technologies > must pick and choose among the data, and one always has to > focus on some reasonable subset of data. You seem to > be suggesting that this is fundamentally more of a problem > for IBM models than for other modelling technologies > - can you elaborate? > > ...and, please!, it almost *hurts* to use the acronym IBM! > could we use "ABM" for Agent-Based Models? I think it fits > better anyway, as an agent in this class of models is not > always an "individual" in the common sense of that term.... > I know the term has some historical precedent for models > in this class, but the acronym IBM induces a certain amount > of, shall we say, cognitive dissonance, no? (not that the > acronym "ABM" itself is inviolate with respect to prior > cognitive content....but, still!.....) > > > Chris Langton ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 16:44:57 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Thu, 17 Apr 97 11:44:57 EDT Subject: Simulating Individual Behavior In-Reply-To: <199704171512.JAA22183@santafe.santafe.edu>; from "cgl@santafe.edu" at Apr 17, 97 9:12 am Message-ID: <9704171554.AA25869@sfi.santafe.edu> Paul and Phillippe and Mike and Chris, I've been enjoying this thread a great deal. From my ecology model, I have been somewhat frustrated by the lack of the kind of data I wanted, but in practice, taking the tiniest bit of relevant detail, instead of all possible detail, buys a lot.... You really get to see what the ramifications are of that seemingly little detail. And there are a lot of those little details "we all know", but haven't necessarily realized how important they are. As a trivial example, when deciding a herbivore's freedom of movement per round, I found that ~215 degrees (similar to a vision path) worked really well, and the "obvious" 360 degrees (random direction) was devastating to the plants. Gecko tells me these crazy things a lot, and I'm forced to think it through and hunt data. Another one I just ran into (predictably, while trying to do something else that Gecko balked at ;), was that my crazy program told me that terrestrial producers should be more productive than aquatic. "Dim-witted bug-generator!" I muttered.... However, to settle the argument, I got the data. Though I'd always "known" that oceanic plankton were "of course" the dominant producers on earth, turns out I was wrong. The "dim-witted bug-generator" was right. If it's right for the right reasons (always a danger, being right for the wrong reason ;), I accidentally found something out that really matters to ecological policy! For it suggests what kind of terrestrial ecosystems (rather safer and easier to manipulate than open ocean :) could conceivably offset CO2 emissions. Assuming I could prove it, of course.... In other words (Chris's :): > So - again, I don't think of this as a problem, but a feature > of these models......and I don't think of it as unique to > ABM models, per se - I think new tools often offer us new > "opportunities" to collect data... Ditto. Cheers, Ginger ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 17:01:03 1997 From: swarm-modelling@santafe.edu (Kevin Crowston) Date: Thu, 17 Apr 1997 12:01:03 -0400 Subject: Simulating Individual B Message-ID: > who has been looking at the behavior of the consumer? We have >been able to make a "rational actor" assumption for so long that we have not >bothered to collect data about the real-life behavior of induividual >consumers. Actually, consumer behaviour is an important research area in the field of marketing (which seems to be a mixture of applied microeconomic and psychology, at least to an outsider). Kevin ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 17:13:28 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: Thu, 17 Apr 1997 10:13:28 -0600 Subject: Simulating Individual B Message-ID: <199704171613.KAA22412@santafe.santafe.edu> > Actually, consumer behaviour is an important research area in the field of > marketing (which seems to be a mixture of applied microeconomic and > psychology, at least to an outsider). Right - I think that there is a tremendous amount of data on consumers that have been taken in market "demographic" studies, and of course there are Nielson ratings, the Billboard top-10 and etc... These data exist in corporations, and they may or may not be available to scientists for research use. One group working with such data in the context of ABMs is the Emergent Solutions group at Coopers and Lybrand, who have been building models of the propagation of fads and fashions throughout (initially music) markets - they used a demographic data base of 150,000 americans, provided by (I'm assuming) the entertainment industry.... Just one example of how there might, in fact, be data out there that has been collected by other groups for other purposes ... Chris ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 17:49:20 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Thu, 17 Apr 97 12:49:20 EDT Subject: Simulating Individual Behavior In-Reply-To: <199704171613.KAA22412@santafe.santafe.edu>; from "cgl@santafe.edu" at Apr 17, 97 10:13 am Message-ID: <9704171658.AA28614@sfi.santafe.edu> P.S., the analogue of consumer research <-> economics is possibly agriculture & fisheries & lumber <-> ecology. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 18:20:34 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Thu, 17 Apr 1997 13:20:34 -0400 (EDT) Subject: IBMs.. Message-ID: <199704171720.NAA22086@tide.cise.ufl.edu> Chris and all: Let me address your points: > why is data calibration more of a problem for IBM > models than for other modelling technologies? What sorts > of problem domains are you particularly concerned about? As a general statement, it is always more difficult to collect data from small scale phenomena than for large scale phenomena regardless of domain. Consider the ecological domain as an example. Improved sensing, radio and satellite methods are providing data for individuals, where this was not possible in the past; however, getting population data is generally easier than obtaining specific data from individuals in the population. The concepts of population and sample in statistics gives us some justification for this. What holds true for ecology most likely holds true for other fields---that is, procuring the data can be expensive, time consuming and difficult. > I can allow as how it would be difficult to calibrate > an IBM model of a forest precisely so that each tree-agent > is parameterized via data taken from its respective > real-tree in the forest. Yet, all modelling technologies > make do with some degree of approximation. A reservoir-flow > model of tree-species interaction in a forest would simply > treat all of the trees of each species as "one big tree" > of that species with respect to some data (such as concentration, > nutrient uptake, waste-production, and etc.) while ignoring > other data (such as spatial distribution, variety within the > species, and etc.) This will be justified for certain questions > about forest dynamics, but not for others, and might make more > sense for some problem domains than for others. Yes, agreed. > You seem to be suggesting that this is fundamentally more of a problem > for IBM models than for other modelling technologies. ...In the sense that it is generally less economic and more difficult to obtain individual data for basic reasons of time, cost and other factors. New sensing methods are helping but, yes, I would say that IBMs are more difficult to create and calibrate due to the sheer amount of data that are needed. I have copied by colleague in Miami to comment on this aspect since I am not a field person--i.e., not going out to tag gators :) This is not a theoretical issue; it is one of practicality relating to cost, time and availability of data. > ...and, please!, it almost *hurts* to use the acronym IBM! > could we use "ABM" for Agent-Based Models? I think it fits As you say, IBMs have precedence in the ecological community and so I see no need for changing this term when speaking of ecological matters. Regarding something more generic, I actually prefer the word "Object". "Agent" sounds like something human. Anyway, I realize we are discussing semantics at this point. -paul Paul A. Fishwick E-Mail: fishwick@cise.ufl.edu Dept. of Computer & Info Phone & FAX: (352) 392-1414 Science and Engineering WWW: http://www.cise.ufl.edu/~fishwick University of Florida (PGP Key available at above WWW address) P. O. Box 116120 332 Bldg. CSE, Gainesville, FL 32611-6120 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 18:32:34 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Thu, 17 Apr 1997 13:32:34 -0400 (EDT) Subject: calibration Message-ID: <199704171732.NAA22113@tide.cise.ufl.edu> Well, I am glad that this new mailing list called 'simsoc' is finally encountering some discussion and somehow we managed to bring in the SWARM mailing list. So much the better as this is appropriate. To respond to Philippe: > Perhaps you are talking about ecosystems from your simulation-oriented > point of view, as the word 'calibrate' indicates. Perhaps I am. > I think that a complex > system like an ecosystem cannot be modeled with some large FSM or even > with a set of stochastic equations (except in the short-term): in an > ecosystem there is probably no unique latent 'model' to be discovered, > against which we could compare some data and make adjustments. Instead > there probably exists many possible modes, with the system unpredictably > switching from time to time to one or another. I am not sure that we could argue this one way or the other. I see no reason why a model could not be developed. When you say "many possible modes", this can easily be modeled by stepping up a level of abstraction so that the modes are modeled as well. > The first thing to do could be to try to recognize these modes. Yes, as part of our systems identification process. > Because > switching is unpredictable (ecosystems seem to exhibit self-organized > critical modes), it makes no sense to try to 'calibrate' the whole system, > in the same manner simulation engineers calibrate a model of a factory or a > flexible workshop. I cannot see your logic. You say it makes no sense but do not provide any justification for your feeling. I think we'd need to get into specifics in order to continue along these lines. -paul ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 18:50:33 1997 From: swarm-modelling@santafe.edu (J.J. Merelo Guervos) Date: Thu, 17 Apr 97 19:50:33 +0200 Subject: Simulating Individual B In-Reply-To: <199704171613.KAA22412@santafe.santafe.edu> References: <199704171613.KAA22412@santafe.santafe.edu> Message-ID: <9704171750.AA09973@kal-el.ugr.es> >>>>> "cgl" == cgl writes: cgl> One group working with such data in the context of ABMs is cgl> the Emergent Solutions group at Coopers and Lybrand, who have cgl> been building models of the propagation of fads and fashions Anything published on that topic? Any reference? Of course, I could also push our own model, in which agents look at products, and decide to buy them or not; then they are punished or rewarded according to how much each product is publicized. Results show that it has an uncanny similitude with reality. But I dont have much quantitative data to support that model, only qualitative. I would like to have it, though. See ya! JJ -- JJ Merelo | http://kal-el.ugr.es/htbin/jj-plex Grupo Geneura ---- Univ. Granada | http://kal-el.ugr.es/ ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 20:34:15 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Thu, 17 Apr 1997 15:34:15 -0400 (EDT) Subject: ABMs Message-ID: <199704171934.PAA22512@tide.cise.ufl.edu> I think we are converging on an agreement: ABMs or IBMs (whatever your preference) will provide good models as long as 1) these particular models happen to answer the set of questions that you are asking of the model in the first place, and 2) enough quality data exist to use the ABMs effectively. In the ATLSS Project (across-trophic simulation of the Everglades), we are all pretty-much "pro-ABM/IBM" however, I was surprised to find out that IBMs were still controversial in the Ecology community, anyway. Maybe some of our Eco-experts could shed some light on this? -paul ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 20:46:35 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Thu, 17 Apr 97 15:46:35 EDT Subject: Simulating Individual Behavior In-Reply-To: <9704171658.AA28614@sfi.santafe.edu>; from "Ginger Booth" at Apr 17, 97 12:49 (noon) Message-ID: <9704171956.AA03098@sfi.santafe.edu> Paul, Re "controversy". Kinda depends on whether: a.) You believe you need to convince a middle-aged world-reknowned expert on doing things "the other way" (doesn't matter which way) that "your way" is better, or, b.) You believe that "your way" is validated (or not) by results and follow your own piper. I suspect a.) is impossible. If you pursue b.) and are really successful, you get to be the expert of a.) sometime. If you're really good, you may be dead by "sometime". :) My two cents. :) Ciao, Ginger ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 22:50:32 1997 From: swarm-modelling@santafe.edu (Doug Donalson;) Date: Thu, 17 Apr 1997 14:50:32 -0700 (PDT) Subject: ABM/I(C?)BM Message-ID: Ah, you have all jumped into the topic closest to my heart (well maby not the closest, but close). I am working on IBM's in ecology and am going through the wars of justification of a new technique. 1.) To Chris (ABM/IBM) You can call a gas station attendent a petroleum transfer engineer but he still smells like benzene! ;-) (Sorry Chris, I couldn't resist.) By the way, IBM in ecology doesn't just imply individual agents. It can also refer to differential equation based biomass change models where detailed physiology is included. See DeAngelis and Gross Individual Based Models. WARNING: PERSONAL OPINION NOW RUNS AMUCK!!!! 2.) To understand the attitude of mainstream theoretical ecology you have to start with it's emergence in the early '70s. Many of the builders were transfers from physics and were hoping to bring the idea of a nice clean set of paradigms (sorry for all the mis-spells) to ecology. There is a hugh litrature of stability boundries and equlibrium boundries for large groups of differential equation based models. Those who have put in so much hard work probably don't take kindly to being told that the assumptions behind their models are broken by almost every real ecological system! (The truth does hurt doesn't it?) The major arguments against complex models is that including parameters that are not precisly known can lead to large error propigation and misleading results. In particular, movement behavior is considered especially suspect. Unfortunatly for the detractors of complex models (see how easily I got out of the ABM/IBM discussion!) is that their argument points a lot of fingers at the simple models as well. If a model is highly sensitive to a single parameter or a group of parameters then what does that say about the results of a model that solves this problem by ignoring them! In a discussion with Alan Hastings (UC Davis) he stated that we should probably not yet study more than two or three species interactions because we (after 20+ years?) don't yet know enough about them. Maby this is because there is no such thing as a two or three species interaction in the real world. My work to date (pre-swarm) with a model I call the Heuristic Acynchronous Discrete Event Simulation or HADES, has shown that the assumptions of simple, differential equation based models can compleatly change the outcomes of a simulation. We used the original Lotka-Volterra predator-prey models as a base line and then successivly removed two of the main assumptions, that demographic stochasticity can be ignored and that the populations can be modeled as "well-mixed" at all times. The results showed that the equlibrium resullts predicted by the differential equation based model held fairly true over a large range of system sizes but that the time to extinction varied greatly with space playing both a stabilizing and destabilizing role at different system sizes. Bottom line. There is a lot of work that needs to go into understanding this new realm of complex models. One of the key areas is that of error propigation due to a combination of multiple parameters and inaccuracies in their measurements. Right or wrong, we have to prove the validity of the models to the main stream theoretical ecological community. Be of strong heart, remember, the young doctor who first proposed sterile surgical techniques was just about driven out of the profession by the greatest medical minds of his time! Doug Donalson *************************************************************************** * Doug Donalson * Office: (805) 893-2962 * * Ecology, Evolution, and Marine Biology * Home: (805) 961-4447 * * UC Santa Barbara * email donalson@lifesci.ucsb.edu* * Santa Barbara Ca. 93106 * * *************************************************************************** * * * The most exciting phrase to hear in science, the one that * * hearlds new discoveries, is not "EUREKA" (I have found it) but * * "That's funny ...?" * * * * Isaac Asimov * * * *************************************************************************** ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 23:14:03 1997 From: swarm-modelling@santafe.edu (Doug Donalson;) Date: Thu, 17 Apr 1997 15:14:03 -0700 (PDT) Subject: Simulating Individual Behavior In-Reply-To: Message-ID: Be careful of this method. There are at least two main problems you can run into. The first is that you can force fit a set of parameters to a model and get a result out that matches some experimental data but you have no guarentee that you have any more that a pretty picture. For a good example of what not to do see Gary Harrision differential equation based modeel of Luckinbill's Paramecium/Didinium experiments in the Jan 94 Ecology issue in the concepts section. Real systems are not deterministic. If you run a field experiment 20 times you will get 20 results and hope that there is enough commonality to get statistical power. Therefore, the result you are to which you are callibrating is just a sample from a statistical distribution. If that distribution has a wide varience then you may have fine-tuned your model to an point far from the expected behavior. A second trap is that if your assumption of the basic mechanisim is incorrect you may fine-tune to a different system that has the same result for that set of parameters. Unfortunatly, I don't has a good set of procedures to eleminate these hazzards. I am also not saying that this is a bad method, it happens to be one I saw abused in an attempt to justify the use of a differential equation based mode. Doug Donalson *************************************************************************** * Doug Donalson * Office: (805) 893-2962 * * Ecology, Evolution, and Marine Biology * Home: (805) 961-4447 * * UC Santa Barbara * email donalson@lifesci.ucsb.edu* * Santa Barbara Ca. 93106 * * *************************************************************************** * * * The most exciting phrase to hear in science, the one that * * hearlds new discoveries, is not "EUREKA" (I have found it) but * * "That's funny ...?" * * * * Isaac Asimov * * * *************************************************************************** On Thu, 17 Apr 1997, Jan Kreft wrote: > OK, I agree with Paul and Mike that it can be very tricky to determine the > parameters for individuals. Bacteria as individuals, e. g., are simply so > small that only the latest high-tech methods could give estimates for the > parameters you would have to use in a model (and that costs a lot of > timemoney). Almost always microbiologists measure the population average > and disregard the heterogeneity of the population. > > But I can see another way to solve the problem. If you start with a > conceptual model with all the parameters and mechanics you need and only a > rough idea of actual values (from population average measurements) you can > run the sim and compare the output with the real system to be modeled. > Then you optimize the parameters to achieve the desired output. Perhaps > call it back-calibration. (Is there an official term for it?) > > For that purpose, you need a parameter manager (see previous discussion > with "parameter" in the subject line) to allow a search through param > space in an evolutionary fashion. As far as I can tell, Gecko's param > manager is (the only one?) suited for that task. > > BTW, does IbM hurt also? If so, sorry for that ;-). Point being is that > that's the term used in the literature and if I want to search for new > literature, I rely on this keyword. Therefore I would want to use this > keyword myself sometime. > > Cheers, > > Jan. > > Mike Brown wrote: > > > I think Paul's question raises two important points - though they may be more > > about the organization of science than about the inherent difficulties of > > SWARM or any other ABM (Thanks, Chris. I like that acronym better too.) > > Ecologists are concerned with aggregations of individuals; they have not had > > to know as much about individual species as, for example, zoologists. > > Similarly, microeconomists focus on the behavior of individual firms and > > macroeconomists on the aggregate behavior of the economy, etc. > > > > ABM create two problems. First, the ecologist and macroeconomist have to > > "ratchet down" and study the behavior of individual entities. Moreover, they > > have to know enough about individual behavior to determine which specific > > behaviors might be relevant to the question under investigation. While this is > > not an insurmountable problem, it does demand a new focus for researchers. > > > > Second, there are some disciplines where data on the behavior of individual > > "agents" simply has not been studied. To take a bad example, look at > > economics. Macro studies the behavior of the aggregate, and micro the study of > > the firm -- but who has been looking at the behavior of the consumer? We have > > been able to make a "rational actor" assumption for so long that we have not > > bothered to collect data about the real-life behavior of induividual > > consumers. > > > > For these reasons, I think Paul is very right -- modeling and validating the > > behavior of individual entities can be very tricky. > > > > Mike > > > > Paul Fishwick wrote: > > > > >A key problem with IBM is not that it is necessarily > > >computationally prohibitive, but that not enough data are > > >available to calibrate the model... > > > > Paul - why is data calibration more of a problem for IBM > > models than for other modelling technologies? What sorts > > of problem domains are you particularly concerned about? > > > > I can allow as how it would be difficult to calibrate > > an IBM model of a forest precisely so that each tree-agent > > is parameterized via data taken from its respective > > real-tree in the forest. Yet, all modelling technologies > > make do with some degree of approximation. A reservoir-flow > > model of tree-species interaction in a forest would simply > > treat all of the trees of each species as "one big tree" > > of that species with respect to some data (such as concentration, > > nutrient uptake, waste-production, and etc.) while ignoring > > other data (such as spatial distribution, variety within the > > species, and etc.) This will be justified for certain questions > > about forest dynamics, but not for others, and might make more > > sense for some problem domains than for others. > > > > Thus, there is data and there is data. All modeling technologies > > must pick and choose among the data, and one always has to > > focus on some reasonable subset of data. You seem to > > be suggesting that this is fundamentally more of a problem > > for IBM models than for other modelling technologies > > - can you elaborate? > > > > ...and, please!, it almost *hurts* to use the acronym IBM! > > could we use "ABM" for Agent-Based Models? I think it fits > > better anyway, as an agent in this class of models is not > > always an "individual" in the common sense of that term.... > > I know the term has some historical precedent for models > > in this class, but the acronym IBM induces a certain amount > > of, shall we say, cognitive dissonance, no? (not that the > > acronym "ABM" itself is inviolate with respect to prior > > cognitive content....but, still!.....) > > > > > > Chris Langton > > > > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 23:33:17 1997 From: swarm-modelling@santafe.edu (Doug Donalson;) Date: Thu, 17 Apr 1997 15:33:17 -0700 (PDT) Subject: Simulating Individual Behavior In-Reply-To: <9704171956.AA03098@sfi.santafe.edu> Message-ID: Amen! But possibly not as bad as all that, there are top researchers such as Roger Nisbet and Bill Murdoch at UCSB who see relvance and need for this type of model. (Even if Roger and I don't always agree on all the details.) My present project is setting up an individual-based spatially-explicit model to begin to explore the dynamics of the Red Scale/Aphytis host-parasitoid system that Bill has been studying for some time. I was approched by Bill because he felt that this new method might be able to shed further light on the mechanisms that lead to the observed system dynamics. Now if only we could get the concurrent schedule part of the SWARM schedule working, oh well ... (Actually Roger Burkhart is being most helpful!) Sorry Ginger, I went and contaminated you modeling group with a tech statement. Doug Donalson *************************************************************************** * Doug Donalson * Office: (805) 893-2962 * * Ecology, Evolution, and Marine Biology * Home: (805) 961-4447 * * UC Santa Barbara * email donalson@lifesci.ucsb.edu* * Santa Barbara Ca. 93106 * * *************************************************************************** * * * The most exciting phrase to hear in science, the one that * * hearlds new discoveries, is not "EUREKA" (I have found it) but * * "That's funny ...?" * * * * Isaac Asimov * * * *************************************************************************** On Thu, 17 Apr 1997, Ginger Booth wrote: > Paul, > > Re "controversy". Kinda depends on whether: > > a.) You believe you need to convince a middle-aged world-reknowned expert > on doing things "the other way" (doesn't matter which way) that "your way" is > better, or, > > b.) You believe that "your way" is validated (or not) by results and > follow your own piper. > > I suspect a.) is impossible. If you pursue b.) and are really > successful, you get to be the expert of a.) sometime. If you're really good, > you may be dead by "sometime". :) > > My two cents. :) > > Ciao, > Ginger > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 23:44:40 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Thu, 17 Apr 1997 15:44:40 -0700 Subject: Simulating Individual Behavior References: Message-ID: <3356A7D8.15A2426B@ix.netcom.com> This is my first posting to any of the Swarm mailing lists, so maybe I'm supposed to introduce myself: Hello, I'm Mark, and I've been fooling around with microsimulation (and other approaches to what is now called ABM) off and on since about 1976. My main interest is in the ability of ABM to unify the scientific enterprise: different scientific communities talk about the same agents in different, sometimes contradictory, ways [this is _especially true in social and behavioral science], and the explicit realization that these _are_ the same agents can be the focal point for a deeper level of consistency among disparate theories and between theory and observation. That's why this particular thread has interested me a great deal. Jan Kreft wrote: > > But I can see another way to solve the problem. If you start with a > conceptual model with all the parameters and mechanics you need and only > a rough idea of actual values (from population average measurements) you > can run the sim and compare the output with the real system to be > modeled. Then you optimize the parameters to achieve the desired output. > Perhaps call it back-calibration. (Is there an official term for it?) There is a tradition of (social) microsimulation going back to G. Orcutt (1957) "A new type of socio-economic system", _Review of Economics and Statistics_ 58: 116-123 and then the classic Orcutt, G. et al. (1961) _Microanalysis of Socioeconomic Systems: A Simulation Study_ (NY/Evanston/London). There's also the more recent Orcutt, G. et al (eds.) (1984) _Microanalytic Simulation Models to Support Social and Financial Policy_ (Amsterdam). There's no doubt much more recent stuff, but I've been out of touch with this particular scene for a while. The Germans have been doing (social) microsimulation like this in grand style for about twenty years now, including official applications to federal policy. Most approaches that I'm aware of distribute parameters across agents based on dependent frequency distributions that have been derived from disparate aggregate data sources. I think this may be equivalent to the kind of "back-calibration" you suggest here, in that failure to validate a model must necessarily feed back into the assumptions used to estimate the initial population (and its agents' parameters). These models never (AFAIK) used anything more agent-specific than statistical cohorts derived from the dependent distributions. Although there is a mountain of literature in German, my favorite is a published dissertation: Helmut Vetterle (1986) _Konstruktion und Simulation mikroanalytischer Modelle. Die Methode der Mikrosimulation und ihre Anwendung_ ("Design and Simulation of Microanalytical Models: Method and Application of Microsimulation"). Augsburg: MaroVerlag. >150 refs. I suspect that this book is out of print, and that there is no English translation. If I can get Vetterle's consent, I'd be willing to put the dissertation on my website in the foreseeable future, and even translate part of it into English -- if there's any interest. (Maybe this is all old hat and I'm making a fool of myself -- somebody kick me if that's the case.) -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 00:41:54 1997 From: swarm-modelling@santafe.edu (Steve Emsley) Date: Fri, 18 Apr 1997 00:41:54 +0100 Subject: Simulating Individual Behavior In-Reply-To: <3.0.32.19970417095049.006ebc8c@poseidon.obs-vlfr.fr> (message from Philippe LAVAL on Thu, 17 Apr 1997 09:50:51 -0100) Message-ID: <199704172341.AAA00780@dial1199.csv.warwick.ac.uk> Just some thoughts: All of the individual-based models in ecology that I am aware of seek to model spatial resource competition. These range between 'artificial ecosystems", such as Gecko, and specific applications, such as Dave Green's CA study of 'Crown-of-Thorns' outbreaks. These models belong to the 'divide and conquer' school. Personally, my interest is in developing a model as a 'surrogate experimental system" (Olson 1995). Perhaps 'simulation' is a more accurate description. As such, I question Philippe Laval's statement that "in an ecosystem there is probably no unique latent 'model' to be discovered, against which we could compare some data and make adjustments." Although I concur that there is no 'whole ecosystem model' surely, in terms of emergent phenomena, one would not expect such a model to exist. However, there are latent models for physiological & behavioural processes within organisms which, by virtue of their interaction, determine one of many 'ecosystem' models. Mike Brown says "Ecologists are concerned with aggregations of individuals; they have not had to know as much about individual species as, for example, zoologists." In my field, plankton ecology, mean-field modellers have dominated the field for decades. Mathematically-endowed researchers will reduce any mean field coupled system of differential equations into an idealised excitable system. It is their goal in life. Forget individual differences or interaction between individuals - it just muddies the water. "First, the ecologists ... have to 'ratchet down' and study the behavior of individual entities ... it does demand a new focus for researchers." This is exactly what is required. In order to approach a plankton SWARM simulation this 'controversial (mad!)' researcher has had to enlist help from researchers into cell-orientated mechanistic models of physiology and behaviour. Even then we're only dealing with a drop in the ocean!! At least, it's not a multicellular drop! Jan Kreft says " If you start with a conceptual model with all the parameters and mechanics you need and only a rough idea of actual values (from population average measurements) you can run the sim and compare the output with the real system to be modeled. Then you optimize the parameters to achieve the desired output. Perhaps call it back-calibration." Excellent. I attempted this argument with a mean-field modeller PI on our "Testable Model" project by suggesting that my model will be intrinsically testable. I argue that for my simulation if I can adequately model the light field (easy) and turbulent environment (more difficult) then seed my 'ocean' with agents having a log normal spread of parameter values I can spin up and, assuming my mechanistic models are realistic, evolve a population that has parameter values optimised for fitness in their artificial environment (and comparable with organisms to be found in real oceanonographic settings). Sort of a spatially-extended genetic algorithm. Ginger says " For it suggests what kind of terrestrial ecosystems (rather safer and easier to manipulate than open ocean :) could conceivably offset CO2 emissions." Suggest you glance at Wally Broeker (1991) Keeping global change honest". Anti-intuitively, calcitic organisms PRODUCE CO2 when forming their skeletons. Although once it gets down into the deep-sea sediments it's there for a residence time c. 1 million years you have to find a way of increasing productivity first. Apart from dumping all our cars into the Pacific (Martin iron-feritilisation hypothesis) chances are that global warming transients furthur stabilise the mixed layer leading to REDUCTION in productivity. Lovelock is probably correct - grow more trees (and build timber houses, furniture or bury the crop!!). With that warming thought, I'll sign off! -- Steve Emsley Ecosystem Analysis & Management Group, University of Warwick, UK sme@oikos.warwick.ac.uk ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 00:45:00 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: Fri, 18 Apr 1997 10:45:00 +1100 (E ) Subject: Simulating Individual Behaviour In-Reply-To: <199704171743.LAA19514@grasshopper.santafe.edu> from "glen e. p. ropella" at Apr 17, 97 11:43:15 am Message-ID: <9704172345.AA06980@malthus> >From: cgl@sfi.santafe.edu >To: swarm-modelling@santafe.edu >Subject: Re: Simulating Individual B >Date: Thu, 17 Apr 1997 10:13:28 -0600 > >> Actually, consumer behaviour is an important research area in the field of >> marketing (which seems to be a mixture of applied microeconomic and >> psychology, at least to an outsider). > > Right - I think that there is a tremendous amount of data on >consumers that have been taken in market "demographic" studies, >and of course there are Nielson ratings, the Billboard top-10 >and etc... > > These data exist in corporations, and they may or may not be >available to scientists for research use. > > One group working with such data in the context of ABMs is >the Emergent Solutions group at Coopers and Lybrand, who have >been building models of the propagation of fads and fashions >throughout (initially music) markets - they used a >demographic data base of 150,000 americans, provided by >(I'm assuming) the entertainment industry.... > I'm pleased that the question of who studies consumer behaviour has been taken up, and answered. As well as the databases mentioned by Chris, there are also the databases from scanners in supermarkets, although it is not so easy to obtain access to these. (As a plug, see a recent paper by my coauthors and me which uses some scanner data in an ABM model: Midgley DF, Marks RE, and Cooper LG, "Breeding hybrid strategies" _Management Science_ 43(3): 257-275, March 1997.) Another area of data on individual consumer behaviour in economics comes from experimental economics: examining the choices of subjects in laboratory settings, when faced with varying settings. Robert Marks ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 02:07:29 1997 From: swarm-modelling@santafe.edu (Randall Gray) Date: Fri, 18 Apr 1997 11:07:29 +1000 (EST) Subject: Simulating Individual B In-Reply-To: (MICHAEL.L.BROWN@cpmx.saic.com) Message-ID: <199704180107.LAA11737@njal.ml.csiro.au> ABM's do indeed pose some problems w.r.t. "calibration". In a project of days gone by we skirted the issue somewhat by attempting to "bracket" reality in our range of assumptions of behaviour. We were looking at potential contamination of a marine trophic chain due to period (regular) dumping of a low level hazardous waste at sea. We ran a range of models which went from the most pessimistic to quite optimistic in terms of behaviour with respect to tainted waters and assessed the results. I am certain that what we did was not optimal, but collecting the sort of data we'd have required would have been (and still is) prohibitively expensive. I suppose the issue is "How close to reality do we need to get before the simulation exhibits the same sorts of strange attractors or ritical modes we see in Real Life?" Moreover, how do we correctly identify that we've arrived? I suppose that the problem with the ecosystem work is that you may only see a few critical modes in an ecosystem under study, but with your simulation you might actually come across dozens more. Some of the strange attractors arise from the implementation of the model, some are critical modes of the simulation, and some are also critical modes of the system being simulated. You always *hope* that the strange attractors of the system are strange enough to raise your suspicions, but differentiating the other two seems to be one of those harder problems. -- Randall ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 14:56:51 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Fri, 18 Apr 97 9:56:51 EDT Subject: Simulating Individual Behavior In-Reply-To: <199704172341.AAA00780@dial1199.csv.warwick.ac.uk>; from "Steve Emsley" at Apr 18, 97 12:41 (midnight) Message-ID: <9704181406.AA15206@sfi.santafe.edu> Hey, Steven! > Ginger says " For it suggests what kind of terrestrial ecosystems (rather > safer and easier to manipulate than open ocean :) could conceivably offset > CO2 emissions." Suggest you glance at Wally Broeker (1991) Keeping global > change honest". Anti-intuitively, calcitic organisms PRODUCE CO2 when > forming their skeletons. Although once it gets down into the deep-sea > sediments it's there for a residence time c. 1 million years you have to > find a way of increasing productivity first. Apart from dumping all our > cars into the Pacific (Martin iron-feritilisation hypothesis) chances are > that global warming transients furthur stabilise the mixed layer leading > to REDUCTION in productivity. Lovelock is probably correct - grow more > trees (and build timber houses, furniture or bury the crop!!). > > With that warming thought, I'll sign off! Another paper: "A large northern hemisphere terrestrial CO2 sink indicated by the 13C/12C ratio of atmospheric CO2", Science, Aug 25 1995. Since the title doesn't do much, from the abstract: "A strong terrestrial biospheric sink was found in the temperate lattitudes of the Northern Hemisphere in 1992 and 1993, the -magnitude of which is roughly half that of the global fossil fuel burning emissions for those years-." Later, Ginger ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 15:05:25 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Fri, 18 Apr 1997 15:05:25 +0100 (BST) Subject: Simulating Individual Behavior In-Reply-To: Message-ID: Doug, thanks for your comments. Doug Donalson wrote: > Be careful of this method. There are at least two > main problems you can run into. The first is that you can > force fit a set of parameters to a model and get a result out > that matches some experimental data but you have no guarentee > that you have any more that a pretty picture. The way to avoid such a force fit that I'm thinking about is optimizing/evolving the parameter set. Suppose that the parameter set is optimized with respect to the individuals' fitness (one or a few definitions of fitness or performance could be used). Then you compare the result of only those sims with optimized params with real data (in my case lab data, therefore the variation will be better known hopefully). (Real data also are the result of (evolutionary) optimized behavior). If you don't get a reasonable match the assumptions about the basic mechanism will be incorrect or lacking important points. Does this make sense to you? Do you see a better way to get around the lack of appropriate, not theoretically biased quality data? Simply setting parameters to experimental values beforehand could get you into serious problems also if a tiny deviation of a param has a large effect and it happens that that experimental value is just a bit wrong... Jan. > [...] > If you run a field experiment 20 times you > will get 20 results and hope that there is enough commonality > to get statistical power. Therefore, the result you are > to which you are callibrating is just a sample from a statistical > distribution. If that distribution has a wide varience then you > may have fine-tuned your model to an point far from the expected > behavior. > > A second trap is that if your assumption of the basic mechanisim > is incorrect you may fine-tune to a different system that has > the same result for that set of parameters. > > Unfortunatly, I don't has a good set of procedures to eleminate > these hazzards. I am also not saying that this is a bad method, > it happens to be one I saw abused in an attempt to justify > the use of a differential equation based mode. > > Doug Donalson ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 15:32:54 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Fri, 18 Apr 1997 15:32:54 +0100 (BST) Subject: Simulating Individual Behavior In-Reply-To: <3356A7D8.15A2426B@ix.netcom.com> Message-ID: Mark, thanks for the pointers. Mark P. Line wrote: > ...... > There is a tradition of (social) microsimulation going back to G. Orcutt > (1957) "A new type of socio-economic system", _Review of Economics and > Statistics_ 58: 116-123 and then the classic Orcutt, G. et al. (1961) > _Microanalysis of Socioeconomic Systems: A Simulation Study_ > (NY/Evanston/London). > > There's also the more recent Orcutt, G. et al (eds.) (1984) > _Microanalytic Simulation Models to Support Social and Financial Policy_ > (Amsterdam). > ...... > The Germans have been doing (social) microsimulation like this in grand > style for about twenty years now, including official applications to > federal policy. Most approaches that I'm aware of distribute parameters > across agents based on dependent frequency distributions that have been > derived from disparate aggregate data sources. I think this may be > equivalent to the kind of "back-calibration" you suggest here, in that > failure to validate a model must necessarily feed back into the > assumptions used to estimate the initial population (and its agents' > parameters). These models never (AFAIK) used anything more > agent-specific than statistical cohorts derived from the dependent > distributions. > > Although there is a mountain of literature in German, my favorite is a > published dissertation: > > Helmut Vetterle (1986) _Konstruktion und Simulation mikroanalytischer > Modelle. Die Methode der Mikrosimulation und ihre Anwendung_ ("Design > and Simulation of Microanalytical Models: Method and Application of > Microsimulation"). Augsburg: MaroVerlag. >150 refs. > > I suspect that this book is out of print, and that there is no English > translation. If I can get Vetterle's consent, I'd be willing to put the > dissertation on my website in the foreseeable future, and even translate > part of it into English -- if there's any interest. (Maybe this is all > old hat and I'm making a fool of myself -- somebody kick me if that's > the case.) > That would be fine! Would you make that diss available? I, personally, don't need a translation but most people will :-(. Jan. > ....... ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 17:00:09 1997 From: swarm-modelling@santafe.edu (Rick Riolo) Date: Fri, 18 Apr 1997 12:00:09 -0400 (EDT) Subject: Simulating Individual Behavior In-Reply-To: Message-ID: Jan, I think evolutionary algorithms (EAs) can contribute to the exploration of parameter value spaces, but I think one should use such techniques with care and with eyes open. Some of the issues include: 1. How do you make sure your definition(s) of fitness, or more generally, your implementation of selection pressures, will be the same as those that exist in the world you are trying to model? If they are not, then your model may evolve some parameter values but they won't reflect the underlying world "parameter" values. 2. EA's are notoriously touchy. You have to make choices about mechanisms and parameter values for the EA itself, you have to make representation choices, and so on. They definitely don't come with any kind of guarantee that they will find optimal (or even near optimal) "solutions" (ie sets of model parameter values in this case). 3. Part of the reason for 2 is that the spaces one searches with EA's are almost always "complex", often with many many "local hills" of pretty good solutions. Thus you might run the same algorithm from random starts 20 times and come up with 20 different "solutions" (parameter value settings). Then you have to choose amoung those, and of course there are all the other nearly equally good parameter settings you didn't ever see. How does one do that? Usually you go back to whatever empirical data, first principles, etc, you know about domain in question...that is, you are (almost) back to where you started from! Now I don't want to leave you with the impression that I have it out for EA's in general or EA's in this context....I use them all the time! They can help in the search for parameter values that at least lead to desired behavior...but you still have to make the case that the values found are plausible parameters values, given whatever other evidence you can bring to bear. I just think one has to be careful not ask or expect EAs to do more than they can do. One interesting use of EA's in connection to model parameter searches is exemplified by John Miller's ANT paper (available as an SFI working paper): 96-03-011 Active Nonlinear Tests (ANTs) of Complex Simulation Models. The basic idea is to use EA's to find weak or questionable aspects of a model's design. The important conceptual change is that rather than finding some "one optimal parameter settings" one is looking for *any* of the (perhaps many) parameter settings that can lead to really bad model behavior. Steven Bankes at RAND has also suggested similar uses of EA's for various parts of "exploratory modeling." - r Rick Riolo rlriolo@umich.edu Program for Study of Complex Systems (PSCS) 1061 Randall Lab University of Michigan Ann Arbor MI 48109-1120 http://pscs.physics.lsa.umich.edu/PEOPLE/rlr-home.html On Fri, 18 Apr 1997, Jan Kreft wrote: > Date: Fri, 18 Apr 1997 15:05:25 +0100 (BST) > From: Jan Kreft > To: swarm-modelling@santafe.edu > Subject: Re: Simulating Individual Behavior > > Doug, > > thanks for your comments. > > Doug Donalson wrote: > > > Be careful of this method. There are at least two > > main problems you can run into. The first is that you can > > force fit a set of parameters to a model and get a result out > > that matches some experimental data but you have no guarentee > > that you have any more that a pretty picture. > > The way to avoid such a force fit that I'm thinking about is > optimizing/evolving the parameter set. Suppose that the parameter set is > optimized with respect to the individuals' fitness (one or a few > definitions of fitness or performance could be used). Then you compare the > result of only those sims with optimized params with real data (in my case > lab data, therefore the variation will be better known hopefully). (Real > data also are the result of (evolutionary) optimized behavior). If you > don't get a reasonable match the assumptions about the basic mechanism > will be incorrect or lacking important points. > > Does this make sense to you? Do you see a better way to get around the > lack of appropriate, not theoretically biased quality data? > > Simply setting parameters to experimental values beforehand could get you > into serious problems also if a tiny deviation of a param has a large > effect and it happens that that experimental value is just a bit wrong... > > Jan. > > > [...] > > If you run a field experiment 20 times you > > will get 20 results and hope that there is enough commonality > > to get statistical power. Therefore, the result you are > > to which you are callibrating is just a sample from a statistical > > distribution. If that distribution has a wide varience then you > > may have fine-tuned your model to an point far from the expected > > behavior. > > > > A second trap is that if your assumption of the basic mechanisim > > is incorrect you may fine-tune to a different system that has > > the same result for that set of parameters. > > > > Unfortunatly, I don't has a good set of procedures to eleminate > > these hazzards. I am also not saying that this is a bad method, > > it happens to be one I saw abused in an attempt to justify > > the use of a differential equation based mode. > > > > Doug Donalson > > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 17:04:36 1997 From: swarm-modelling@santafe.edu (Stephen C. Upton) Date: Fri, 18 Apr 1997 10:04:36 -0600 Subject: Simulating Individual Behavior Message-ID: <3.0.32.19970418100435.00754004@tsa-po.lanl.gov> This is a great thread!!! First, I'm reminded of a lyric: IBM, UBM, We all BM for IBM (I believe this was from the science fiction book "With a Finger in my I" by D. Hofstader -- if not, then from some other 1970's scifi book on AI :-) Second, I'll paraphrase Hamming (R. W. Hamming, Numerical Methods for Scientists and Engineers, 1962): The purpose of modeling is understanding, not numbers. Third, I'll add some of my own thoughts: I am currently attempting to understand warfare. (For those that find this offensive, consider the better we understand the origin and causes of warfare, the better we can prevent it) There are quite a few models and simulations, as you may imagine, of various aspects and levels of warfare, from the individual soldier to the campaign level, e.g., Desert Storm. Each abstracts certain features which are relevant to a particular analysis. There have been several good points made that seem to cross problem domain boundaries: 1. The resolution or the level of abstraction of the agent (actor, individual, entity) 2. The resolution or the level of abstraction of the interaction between agents 3. Collecting, gathering, identifying data for 1 and 2 4. Identifying patterns or structures, emergent or not (modes, or the resultant behavior of the system as a whole) 5. The difference in modeling philosophy between ABM and ODE's or PDE's (The Newtonian legacy) 6. The relationship of the model to the "real world" (Maybe this was obvious!) There have been several attempts to model the campaign level using ABM's. However, for computational reasons,these were arguably unsuccessful. But the attempts bring up some good questions: At what level do you model your agent or their interactions? Is a Battalion sufficient or do I need to model individual tanks? How much detail is required? (More detail, More detail is the current rallying cry) How does that level of detail relate to system characteristics, e.g., location of strange attractors, etc.? How do you know when you have the level of detail and interactions correct, especially if there isn't a whole lot of data, at either the agent level or the system level? Steve Emsley says, "Personally, my interest is in developing a model as a 'surrogate experimental system". I concur. I also don't have much of a choice, except that historical data provides some clues as to what the important parameters, plus some thinking about the processes in general. But then I'm not sure if those parameters were a function of some other interactions, e.g., the evolution of new tactics as a function of technology. I also sense some "My model is better than your model", i.e., ABM's are "better" than ODE models, or ODE's are "better" than ABM's. This is an easy trap to fall into. Each is abstracting system characteristics and behavior differently. Certainly, thermodynamics is as useful as statistical mechanics, within their respective regimes. However, I currently believe the ODE folks are more guilty of this, but we don't want the pendulum to swing all the way to other side either. Finally, we are all probably guilty of thinking our models represent the real world in some useful fashion. We should always maintain a certain amount of sceptism for any model and remember the paraphrase, modeling is for understanding. Love to hear your comments. thanx *** * **** * ***** ********* ** ****** ***** *** ***** ******** ********* Stephen C. Upton TSA-5, MS F602 Los Alamos National Laboratory Los Alamos, NM 87545 505-667-9435 FAX 505-665-2017 upton@lanl.gov ******* ********* *** ** *** ******** **** ****** ** ****** **** *** *** ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 18:30:45 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Fri, 18 Apr 1997 18:30:45 +0100 (BST) Subject: Simulating Individual Behavior In-Reply-To: <3.0.32.19970418100435.00754004@tsa-po.lanl.gov> Message-ID: Stephen C. Upton wrote: > ........ > There have been several attempts to model the campaign level using ABM's. > However, for computational reasons,these were arguably unsuccessful. But > the attempts bring up some good questions: > At what level do you model your agent or their interactions? Is a Battalion > sufficient or do I need to model individual tanks? > How much detail is required? (More detail, More detail is the current > rallying cry) > How does that level of detail relate to system characteristics, e.g., > location of strange attractors, etc.? > How do you know when you have the level of detail and interactions correct, > especially if there isn't a whole lot of data, at either the agent level or > the system level? You can only know that if you vary the level of detail and look at the results. Maybe there is some sort of optimum, then that's what you would have to search for. > ......... > Finally, we are all probably guilty of thinking our models represent the > real world in some useful fashion. We should always maintain a certain > amount of sceptism for any model and remember the paraphrase, modeling is > for understanding. Yes! Cheers, Jan. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 18:54:45 1997 From: swarm-modelling@santafe.edu (Daniel J Shapero) Date: Fri, 18 Apr 1997 13:54:45 -0400 Subject: Simulating Individual Behavior In-Reply-To: Message-ID: hi, this is my first message to the group. I have been reading the dialogue for about two weeks and feel that something that happened to me yesterday pertains to Dougs message about parameters and conditions of models. I am an undergraduate at Johns Hopkins working on a Buffer Stock Model for a professor with Todd Allen. After getting the program to work (in SWARM) and the the GA values to converge to specific values, we showed the program to the overseeeing prof. he said that our values were all wrong. Todd and I then looked at the model and realized that we took into account something that the professor did not. Our agents adopted rules that had good average utilities, but also the length of time over which that rule was successful was important... a good rule for 10 years was better than one for 2 years. This parameter definately mimics life. But me realized that we did not know which should be more important to the agent, length of duration or actual average utility. So how did the professor get an answer... he did not take length of time importance into account when deriving the analytic answer. once we eliminated that part of the adoption rule, the converged values were correct. The moral is that I thought that the most interesting part of the experience is that ABM(IBM, OBM, whatever) can be used to do parameter sweeps over the importances of length of rule v. average utility over that time. This sweep would derive a pretty picture of a fitness landscape in another variable... Then these new results could be compared to data(if there is any out there). On the other hand, this macroeconomics professor is not interested in these other mechanisms, the ones i feel are most interesting. It is going to be interesting how these results are going to be published (ABM along side MAcroeconomics) Macro using SWARM. I love it!!! hope you didn't mind my jump into the group, Daniel Shapero ps. doug, i like your quote at the end of your messages ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 21:03:21 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Fri, 18 Apr 1997 13:03:21 -0700 Subject: Simulating Individual Behavior References: Message-ID: <3357D389.610C2A23@ix.netcom.com> Jan Kreft wrote: > > Mark P. Line wrote: > > > Helmut Vetterle (1986) _Konstruktion und Simulation mikroanalytischer > > Modelle. Die Methode der Mikrosimulation und ihre Anwendung_ ("Design > > and Simulation of Microanalytical Models: Method and Application of > > Microsimulation"). Augsburg: MaroVerlag. >150 refs. > > That would be fine! Would you make that diss available? I, personally, > don't need a translation but most people will :-(. In the short term, I'm willing to loan the book out if it does happen to be out of print. Check your bookstore to see, and send me your snail address if you want me to loan you my copy. The address of the publisher given in the book is MaroVerlag, Benno Ka"smayr Riedingerstr. 24 [new zipcode?] Augsburg The copyright for the book is held by Beratungsgesellschaft fu"r angewandte Systemforschung mbH - Augsburg Haunstetterstr. 19 [new zipcode?] Augsburg (phone +49-821-571093) The ISBN is 3-87512-501-0. Please let me know what you find out about the book's print status. If it's out of print, I'll try to get permission from the outfit in Augsburg to put relevant excerpts on the Web (the parts about state-of-the-art (anno 1986) hardware and software technology are not very relevant, except historically) as soon as I can get to it. ==== I should have also mentioned Martin Clarke's work in spatial microsimulation, which he had been doing for quite a while by the mid-80's; I assume he's still doing it. See, for instance, Clarke, Martin (1986) "Demographic processes and household dynamics: a microsimulation approach", in Robert Woods and Philip Rees (eds.) _Population Structures and Models_. London: Allen & Unwin. ==== There was a whole Sonderforschungsbereich (SFB; a federally funded research focus) in Germany on microsimulation, especially with federal policy applications, which has probably ended by now. It was SFB 3, "Mikroanalytische Grundlagen der Gesellschaftspolitik" ['Microanalytical Foundations/Underpinnings/Bases of/for Social Policy'], in Mannheim or Frankfurt, I believe. The DFG (Deutsche Forschungsgemeinschaft, like NSF in this country) would certainly know how to access their mountain of grey literature. ==== After posting that message yesterday, I started wondering what had happened in microsim since I last took a look. I found a lot on the Web searching under the usual terms ('microsimulation', 'mikrosimulation', 'microanalytical', 'mikroanalytisch'). Here are a few items I liked: http://www.tc.cornell.edu/Edu/SPUR/SPUR96/Greta/report.html http://www-cpr.maxwell.syr.edu/demogctr/micropap/microlst.htm http://petty.econ.rochester.edu/nas.htm http://itkwww.kub.nl:2080/TUP/Fondslijst/Eco/9536-5.html http://www1.ifs.org.uk/research/personal/CzechModel.HTM and especially http://misic.soc.cornell.edu/ -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 19 02:56:57 1997 From: swarm-modelling@santafe.edu (Steve Emsley) Date: Sat, 19 Apr 1997 01:56:57 +-100 Subject: Simulating Individual Behavior Message-ID: <01BC4C65.3E212A20@p62.ascend3.is2.u-net.net> ------ =_NextPart_000_01BC4C65.3E212A20 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable ---------- From: Stephen C. Upton[SMTP:upton@lanl.gov] Sent: Friday, April 18, 1997 05:04 To: swarm-modelling@santafe.edu Subject: Re: Simulating Individual Behavior This is a great thread!!! Steve Upton's problem domain seems well suited to a hierarchical swarm approach. In my domain each bucket of seawater can contain more = particles than my computer has bytes of RAM and there is no obvious aggregation rule. However, it seems inherent in military modelling that there is a hierarchical structure and a chain of command. So, the agents are: soldiers -> platoon -> company -> regiment -> division -> army corresponding to a chain of command: lieutenant ->=20 captain -> major -> colonel -> general. The lieutenant polls his grunts = and passes the data up the line to the captain. The captain reports his = synopsis of the data from the lieutenants to the major and so on to the general = who, far from the conflict, passes directives down the chain of command.=20 (Apologies for my ignorance of military organisation, whether UK or = US!). "How much detail is required? (More detail, More detail is the current rallying cry)" My point is: As far as the general is concerned there are no individual = soldiers. However, individual soldiers could be modelled - but only a few, and = less=20 (than actually involved in a conflict) platoons, and even less regiments = etc.=20 Whereas modelling a general would require a context-dependent rule-based system I should imagine that, the furthur down the chain of command, = rules become more fuzzy until, at the level of the individual platoon, one may = as well use stochastic differential equations. Of course, one might argue = that engagements are won or lost not because of the 'normal' = distribution but due to the effect of the few on the many. Since heroism is probably = as elusive an AI concept as creativity I imagine that the aim of modelling = warfare is more an exercise in risk assessment than understanding. Anyway, the main point of this posting is: (1) SWARM has been designed with hierarchicies of subswarms in mind although, apart from a few posting, this feature isn't being exploited. = Perhaps the computational overheads of running individual swarms has been a disincentive. If that is the case, roll on parallel swarm! (2) Just because SWARM is suited for ABM does not preclude the inclusion of ODEs. Despite my last post, which may have suggested that I was denigrating mean-field modelling in respect of ABM, I use ODEs in my model. My ModelSwarm sends a step message to a ModelState class that polls my LightSpace, OceanSpace and PhytoPlankton objects for variables = and Runge-Kutta's them for 24 hours to produce the next day's state. More of = an Infinite State Machine than a FSM!=20 (3) Assuming hierarchical order then a "surrogate experimental system" = is=20 a WELL-DEFINED "descriptive model" of the level above. In addition, the = "surrogate system" requires a "descriptive model" of the level=20 below to satisfy the criteria of testability with real world data. = Before disappearing down the hole with diameter =3D Planck's length the hopeful = aim of the exercise is to define the ranges of the parameters used in the level = above ... if only to prevent mean-field modellers from tuning their systems = to their heart's conten... Sorry, [diatribeMode: OFF]. Excuse the musings of a lapsed Type 4 lurker but I couldn't resist this = thread - hail patch dynamics, there's noting like an invasion of territory (ie = the mailing-list) to bring out a spurt of evolution. Maybe it was an = experiment!? Steve Emsley, Ecosystems Analysis & Management Group, University of Warwick, UK. ------ =_NextPart_000_01BC4C65.3E212A20 Content-Type: application/ms-tnef Content-Transfer-Encoding: base64 eJ8+IjkAAQaQCAAEAAAAAAABAAEAAQeQBgAIAAAA5AQAAAAAAADoAAENgAQAAgAAAAIAAgABBJAG AEQBAAABAAAADAAAAAMAADADAAAACwAPDgAAAAACAf8PAQAAAFUAAAAAAAAAgSsfpL6jEBmdbgDd AQ9UAgAAAABzd2FybS1tb2RlbGxpbmdAc2FudGFmZS5lZHUAU01UUABzd2FybS1tb2RlbGxpbmdA c2FudGFmZS5lZHUAAAAAHgACMAEAAAAFAAAAU01UUAAAAAAeAAMwAQAAABwAAABzd2FybS1tb2Rl bGxpbmdAc2FudGFmZS5lZHUAAwAVDAEAAAADAP4PBgAAAB4AATABAAAAHgAAACdzd2FybS1tb2Rl bGxpbmdAc2FudGFmZS5lZHUnAAAAAgELMAEAAAAhAAAAU01UUDpTV0FSTS1NT0RFTExJTkdAU0FO VEFGRS5FRFUAAAAAAwAAOQAAAAALAEA6AQAAAAIB9g8BAAAABAAAAAAAAAMsQgEIgAcAGAAAAElQ TS5NaWNyb3NvZnQgTWFpbC5Ob3RlADEIAQSAAQAjAAAAUkU6IFNpbXVsYXRpbmcgSW5kaXZpZHVh bCBCZWhhdmlvcgCHDAEFgAMADgAAAM0HBAATAAEAOAA5AAYAYwEBIIADAA4AAADNBwQAEwABADcA CwAGADQBAQmAAQAhAAAAOEFGOEY5RDk0MEI4RDAxMUE2Q0Q0NDQ1NTM1NDAwMDAABwcBA5AGAAwM AAASAAAACwAjAAAAAAADACYAAAAAAAsAKQAAAAAAAwA2AAAAAABAADkAwB2zlFxMvAEeAHAAAQAA ACMAAABSRTogU2ltdWxhdGluZyBJbmRpdmlkdWFsIEJlaGF2aW9yAAACAXEAAQAAABYAAAABvExc lLPZ+fiLuEAR0KbNREVTVAAAAAAeAB4MAQAAAAUAAABTTVRQAAAAAB4AHwwBAAAAGAAAAHNtZUBv aWtvcy53YXJ3aWNrLmFjLnVrAAMABhDye3qMAwAHEIILAAAeAAgQAQAAAGUAAAAtLS0tLS0tLS0t RlJPTTpTVEVQSEVOQ1VQVE9OU01UUDpVUFRPTkBMQU5MR09WU0VOVDpGUklEQVksQVBSSUwxOCwx OTk3MDU6MDRUTzpTV0FSTS1NT0RFTExJTkdAU0FOVEFGAAAAAAIBCRABAAAAgwoAAH8KAAAKEgAA TFpGdXqe54r/AAoBDwIVAqgF6wKDAFAC8gkCAGNoCsBzZXQyNwYABsMCgzIDxQIAcHJCcRHic3Rl bQKDM3cC5AcTAoB9CoAIzwnZO/EWDzI1NQKACoENsQtg5G5nAdA1Nwr7FFEL8hpjAEAgCoUKi2xp MQQ4MALRaS0xNDTPDfAM0BzTC1kxNgqgA2D1E9BjBUAtHvcKhx2rDDD1HnZGA2E6H/4edgyCBgAJ E9BwaAnwIEMuIAZVBTACIFtTTVRQFDp1JHJAGPFsLmf4b3ZdH58grQZgAjAh34Mi6yGQaWRheSwU sA8TUAMRHFAqgDE5OTdiIBlAOjA0Jg8grVSebyhPIusD4ArAbS0EYmZsHDAZEEBzAHABkGY0ZS4J gHUr7ycedWJOah6xLg8i61JlNABTFQdwdQtgdDChIEluKGRpdipAdQdAIEI+ZRGANxAFsBr/HAMz NlcddxpFLWdoBAAgOyFhRiAJwTZwIHRoO7FkPiE8YDgPOR8aWCOxdmU9JFQnBCAecQJgE+AgZK8D cQuAL9AJ4G0EIHcwcakv0HVpE9BkO/BvO3HPOxAEkArAEXBpYzdRL+PZCoVhcB5xANBoJEA20Dgg bXlApjvAEXAgYoh1Y2sRwCBvZkER/mEv8BPQBcBDIAOgBaAxAb9A8QRgFhBAIArANoBjQHA+cwqF PAADkUTxBaBtcM51RtIRgAQgYnkT0AQg8UZRUkFNO3A24DvxBJC7P5A7IW5CUEBQN8F1O2G6ZwnB ZzZxAiAKhXI2ULExQCBIb0GAP4ByKoD/QfBBFQuAS5ICME9RROADEH9B8ArARQAwR0kSO+JLpWH7 CoVCnHRN4B7ACHA/kEtCPzuAEXFA8UZRSaEDgWQu+TycU28qgEuBTLEoETthexYQNABzBvA28ASQ BCAt9j5AIDZhbwIgWDJJogBw/0UAWEFM8QdxT8FYQAqFNvIvAJBYxDABSYJyFhBzcH8CIDbwUSJC UlSONAAcMGX9SeFuMPFYMgqFQyAFMEDiu1hBAMBqBbFY8xWgbjBw/1gyVwFCsSWgOvE/kF5ZXGDv MIAEIDsSCcB1VyM24AqFvwqwBBAHkVayKlABkCAlEP9WozCRP5BCQVayX6ViBF+mTxYQXGAAIGNU c3lMEHD/AJBIpkZRZTcDUmXlXnZlEf9mlGBUS0JXsEZAA6BmhWGV7UFwaFaACoVmbuJqxkdR+xjg QxB0KoBkxTbwFhAewH83AHDCTlBtUWbCVJ4a5ih/KqAVkVnwB5ECEAXARPFp5mdMEELAbmM/kEZR UCf/BbBNEAMAMOBNMiqAbmARwBtLkSRQS3YxJFBTISn9VY0iTkFE4EXwRcANsEeBNwMgOyEWEHFB 4BYRPyD9c7BNR+J51CqAezk7QmazzwhwT6JNdgdAbHk2kgUA6HkpIjycTUUAXGALgPVP0XM0AEF0 UUPyZRRt5v87IUdRdVAEoEISS5NXYUwC/wuANvdXtlWGTkiDv4ISNlDdQiBibFEwU0IRLUXRRjH/ JZBFAGqRB9AqgEtCSIEEIL9zVkkjANBT0H4iT1F2BvBfP4BCIEDxXRFv9SlYZnN/iUQ/cQOgiaNZ 1gQgEcBj/XM3V0uSSkFQuDuBbfaHM0d6dYwEE9B4dC0NsHDPCfANsE/BTeItYkpACYD/CoVpABPC NsAv0G5wh0IHcO9MwGZDUWFWlGYIcDwACHD/cZ9U9yqATeJIpoeASaGHkvNH8ZbAenpFAGPRe8JR dP+JkT+AAyBp9YXpWHUqgGERf2BBiNFIpkGDTJA/kBPAb/8RcRPAQxBa4Q3QT5MHMUWA53qQTSOE wCBPVPIIcBGw252ldOBoBUAKwGcKUFFE/wnwTRBXAI6UgzKRcFTBBcDfFaATwEwBBUCZsWGfImn1 Zid1AQDAbCda4VOBaf+IYU1IiGIxcGZnDcEesmn1/4kRbTNsQ1lwJEA2IHVCS5H9gBBzQJA7IUAy AaCIwkim/zBwTJBxUUsxFLCVEIIzBTHvSkEFADvBNwF0RQCVEJWqv1akB3B1c1DHL/FvEWUKhd87 IUfTA5GSsASQYwQAnGL7TdAEAGuBAWTxq9BPskkjv2PQBIETwEtBMKFVjUFZcP8v8CpxbDRA8YAE afOsAqURJzaSgGE8nCgxjLBTV/xBUksgSjMJ4XnBAJB08P9CEQPwPABCiXQyRlIzoC/j/09CUBFk NwdAPAAIYKKgiUH/SCJqpIjzuRZWkjshMTA2cP9T4gQAP/ClcjaSkrALUIAQ+0IBJEBQBJARgGlA SLdvwv9Jwk0jN1El0MPhPDFKs2PB/wMAuXKF+b4Uu1dSNqcRqzL/muE/gEShagI70XzGk/GY0X9j IW0yCrF+IWExL+M8jSj6MoywSkyQpXi69GjSQeT5dHJBQksgQLAHkaVSE1B/BZAKQA2wnDXRIVsy aYhP9ERFoWFEXEFB8UTiC2D/pSG5IncCQxBFwJ4CN6GfQf+/oFcAE8FLYjvRlRAv8Eimv5NBdOBC wDaDB4AAcC0coP8wcEIgULhoE1xQqWXQISqA75UQnyLS8k/jeQqFMEMkQPd/0XswMGFTL/JBETbg O2L/E8Fl0AeBMOBXAEI03KTFQX9m0dQBZRI2cGRWYyNE8UzdopJTCrB1UCqAT3VQAHCT4ZNLM1Bo SoBvUBjx3mskgUwxM8J0VHYKwAcwm0BhZApSY9BXAC1LSeCfAZBAAUuBQJB0cjI0QoD/odJCMh5x MXB1UVayYSCSwP1lYXlAAbXBE9DcUUfiRlF/AHAKhTbQHKADANOh30RNV0RxleWL8kYkwCEa7Sj+ M4ywgKBB0L6hNrBCmwWw9wSBS3KL8iJB0FwgdBDfYt/DIQZxWhJDMhOzIjtCQ7Yhj6BFTEwt0wBG SfhORUR5ILvhBQOtkjBD//JQafWbtAGgxdEkQESyPED/NvB2xFayCoXwavH2enU7Yn/z//UPmUcV oAfgQkF2onP+ZkUAZrMFEEbRBzBp49XhvwGgUDJFALxzO7GRUnKHUf9lciRAN4B0cbI2pxFEITvA fwUQNqGXR25wQHC8ZDbwYfsHgEbSPcPAGPFGAEABQHD/GRC8kQK0kxCWwPXxsNPESf+zaGvjDbGV 5D+QdSHV0Wnmf8wCA8NSEJ8hi8P1fTGVLv8MIDtARmCIo+fTjbIpsNgf/1gCarRj0FEUwtCUhmvm cQD7UkYB8XRAAZJzZ1AMMVZw231QKnFbA5GnQmXcojQAkE9GRl1VjUV4fTD/nzFsM61xGRBKszuA MiDEIGlCEVR5kxAg51DR8HL/RhCoJK4Sh0HCglwxpyG4xP88BB7gUkZ6AjHAjwB5sWkQ/wOwQxCN QUuDQAGlUTaSPdD/RhCtwotRSkBbQv5zfVBB8Pt30EUAKHQwS3LblnoBMKHeLT3QECCMsEJBYgIT 54D/osEv0EnQwCFGUT9wV8Cng3/cUSpgh4FO0daxsyTxRiH2P1WdP2NFQVBAcCpwFLa/XAAQBrcw zDAQEDshJuvRc16wo9QgRzMgZcBullXfxtBOgbwAr2FGUVcx0ANAf0YA2nB3oFWGOZ8zNDGVfQUx kAAtkAADABAQAAAAAAMAERABAAAAQAAHMKBpxlVcTLwBQAAIMKBpxlVcTLwBHgA9AAEAAAAFAAAA UkU6IAAAAAAbcQ== ------ =_NextPart_000_01BC4C65.3E212A20-- ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 19 05:23:10 1997 From: swarm-modelling@santafe.edu (Scott Christley) Date: Fri, 18 Apr 1997 21:23:10 -0700 (PDT) Subject: Simulating Individual Behavior In-Reply-To: <01BC4C65.3E212A20@p62.ascend3.is2.u-net.net> Message-ID: Fascinating! Now that its the weekend; I have the time to step out of my role of striving to make GNUstep a reality; I wish I had such a discussion group when I was in graduate school...anyways let me pose some questions. first a metaphysical one: I am tending to notice a separation between two modelling paradigms: the ABMers (I was an IBMer once! ;-) and the ODEers. Are there true differences between these two paradigms or is it only a perception? Meaning is not a Swarm program a symbolic (mathematical) description of a model, just not as concise as a differential equation? Steve Emsley was talking about military simulation... Has anybody read the recent Wired article about the Marine's use of DOOM for battle simultion? Any comments about the merits of using VR in simulations besides the obvious: scenario iteration (as in responding to a terrorist threat)? Does anybody(or know of anybody) have interest in seeing Swarm expand in VR areas (versus 3D visualization)? Continuing with military simulation; is such simulation geared at controlling and predicting our own country's troups? Or is it more related to training(as in DOOM above) to prepare our troups for possible unknowns? From my understanding of VR use in the military its towards the later, but concentration on the former seems to be more of the theoretical nature. And the iteration I didn't mention where simulation is to geared towards acting out the opposing force. Imagine this scenario: You have some terrorists who have taken over a building and are holding a bunch of hostages. You have a small number of marines who are going to attempt to overtake them. It would seem to me that being able to simulate the most realistic terrorist is going to be both the hardest and most beneficial part of the simulation. Its simulations like this where individual actions have a huge impact upon the outcome. Did I actually read someone mention Planck's constant?!! :-) I'm currently reading a biography of Einstein by Albrecht Folsing (good book!); it got me thinking about how as humans we attempt to find the "parameters" so that we can "model" ourselves after other humans that we idolize. It makes me wonder if the search for the fundamental laws of the universe have progressed from physics to the higher level sciences of biology, sociology, history. excuse my musings Scott ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 19 07:36:35 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Fri, 18 Apr 1997 23:36:35 -0700 Subject: Simulating Individual Behavior References: Message-ID: <335867F3.63414CEB@ix.netcom.com> Scott Christley wrote: > > I am tending to notice a separation between two modelling paradigms: the > ABMers (I was an IBMer once! ;-) and the ODEers. Are there true > differences between these two paradigms or is it only a perception? > Meaning is not a Swarm program a symbolic (mathematical) description of a > model, just not as concise as a differential equation? I don't see it that way at all. Obviously, both an ABM and an ODE model are abstractions of our observations of behavior on the ground. The abstractions made in the two paradigms are different, of course. In ABM's, the abstractions made are informed directly by observation. There's ants, so we model ants. They cut up leaves and carry them around, so we have leaves and pieces and leaf-cutting behavior in our model. The more we know, the more constraints we can build into our ABM. There is no ODE system (no matter how observationally vacuous) that cannot be couched in terms of an ABM, unless we don't really know what it is our model is doing (in which case its uses are limited to crunching out trajectories). In ODE-based and PDE-based models, the abstractions made are the ones that are imposed by the kind of symbolic manipulation (analytic solution) that is possible with simple ODE and very simple PDE systems. In other words: calculus tells us the function has to be continuous, so suddenly we are forced into an abstraction of our population of wildebeests such that population size is a real number and population growth is a continuous function. Biologically (or sociologically, or whatever), we tend to remain rather unconvinced that either one of these assumptions is particularly realistic. ODE's and PDE's were invented so that problems could be solved analytically. But few interesting problems that we'd model with these formalisms are soluable analytically anyway, so there's no longer any good reason to use them, and one very good reason not to use them: they force on us an abstraction that serves merely a by-gone purpose and which is usually not warranted in the biology (or sociology). [As long as you're just doing some simple thumbnail models of biomass and energy balances and what-not, you might be safe up to a point just doing your ODE's, of course. I don't want to deny that.] Now I can say what I think the answers are to the questions you pose above. Are there true differences between these two paradigms? Yes, because ABM's don't force abstractions on us that are artifacts of a method whose day is past and which are not desired otherwise in our models. Is a Swarm program just a less concise description of a the same thing a differential equation describes? No, certainly not. A Swarm program does not normally describe the effects of fractional wildebeests pairing up and producing fractional young. It describes real, whole wildebeests pairing up and producing real, whole young. If anything, a Swarm model is _more_ concise. Swarm: "How many parents did this baby gnu have? Exactly two." ODE: "How many parents does every baby gnu have? 1.998785622342179 +/- 5% at the instant of conception, with a first-derivative slope of +0.73." That's my two gnus' worth... > It makes me wonder if the search for the fundamental laws of > the universe have progressed from physics to the higher level sciences of > biology, sociology, history. I certainly hope not. I would have hoped that the experience of physicists over the last 100 years would have taught biologists, sociologists and historians not to look for fundamental laws of the universe, but rather for theories that help understand the universe. :) -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 19 08:04:22 1997 From: swarm-modelling@santafe.edu (Theodore C. Belding) Date: Sat, 19 Apr 1997 03:04:22 -0400 Subject: Simulating Individual Behavior In-Reply-To: References: <01BC4C65.3E212A20@p62.ascend3.is2.u-net.net> Message-ID: Hi Scott- At 9:23 PM -0700 4/18/97, Scott Christley wrote: >I am tending to notice a separation between two modelling paradigms: the >ABMers (I was an IBMer once! ;-) and the ODEers. Are there true >differences between these two paradigms or is it only a perception? >Meaning is not a Swarm program a symbolic (mathematical) description of a >model, just not as concise as a differential equation? Using ODEs to model an agent-based model like Conway's Game of Life may be theoretically possible, but it's often a very bad representation. It's like using a Fourier series of sine waves to represent a square (pulse) waveform. The Fourier series representation needs a huge number of terms to even approach the accuracy that another type of representation would have (say step functions). On the other hand, using step functions to represent a sine wave is not the best representation. Doyne Farmer and Norman Packard, as well as John Holland, have written about the need for using an appropriate, efficient representation of a dynamical system. For something like the Game of Life, I think the most appropriate representation is in term of agents, not ODEs. If we agree that in some cases we want to use ODEs and in others we want to use agent-based models to model a system, then I think we also have to recognize that each modeling paradigm has different tools that are needed to work with them. A lot can be determined about ODEs using traditional mathematical techniques, as well as numerical simulation techniques. In agent-based models we don't have many mathematical tools at present and usually must rely much more on computer simulation of the agents' behavior. Using ODE tools to analyze an agent-based model (or vice-versa) is like using a hammer to pound a screw in. In this sense, an agent-based model is very different from an ODE, even though they're both mathematical formulations at some level. -Ted -- Ted Belding University of Michigan Program for the Study of Complex Systems ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sun Apr 20 15:19:36 1997 From: swarm-modelling@santafe.edu (Martina Schretzenmayr) Date: Sun, 20 Apr 1997 16:19:36 +0200 Subject: Simulating Individual Behavior/ ZIP codes Message-ID: >Jan Kreft wrote: >> >> Mark P. Line wrote: >> >> > Helmut Vetterle (1986) _Konstruktion und Simulation mikroanalytischer >> > Modelle. Die Methode der Mikrosimulation und ihre Anwendung_ ("Design >> > and Simulation of Microanalytical Models: Method and Application of >> > Microsimulation"). Augsburg: MaroVerlag. >150 refs. >> >> That would be fine! Would you make that diss available? I, personally, >> don't need a translation but most people will :-(. > >In the short term, I'm willing to loan the book out if it does happen to >be out of print. Check your bookstore to see, and send me your snail >address if you want me to loan you my copy. > > >The address of the publisher given in the book is > > MaroVerlag, Benno Ka"smayr > Riedingerstr. 24 > [new zipcode?] Augsburg ----------- The new ZIP Code is 86153 ----------- phone ++49/821/416033 > >The copyright for the book is held by > > Beratungsgesellschaft fu"r angewandte Systemforschung mbH - Augsburg > Haunstetterstr. 19 > [new zipcode?] Augsburg > (phone +49-821-571093) ----------- The new ZIP Code is 86161 > >The ISBN is 3-87512-501-0. > Augsburg is my home town. Very nice and very old (about 2000 years) city. If you ever can visit ... Martina Schretzenmayr ------------------------------------------------------------------------------ Martina Schretzenmayr Dipl.-Geogr. Raumplanerin ETH/NDS Institut fuer Orts-, Regional und Landesplanung (ORL-Institut)/ Institute for Local, Regional and National Planning Eidgenoessische Technische Hochschule Zuerich / Swiss Federal Institute of Technology Zurich ETH Hoenggerberg HIL H 41.3 CH - 8093 Zuerich Tel.: ++41 - 1 - 633 29 47 Fax: ++41 - 1 - 633 10 98 e-mail: Schretzenmayr@orl.arch.ethz.ch HomePage: http://www.orl.arch.ethz.ch/~Schretzenmayr/index.html ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 21 15:24:22 1997 From: swarm-modelling@santafe.edu (Mattias V. Bertelsen) Date: Mon, 21 Apr 1997 09:24:22 -0500 (CDT) Subject: Simulating Individual Behavior In-Reply-To: <335867F3.63414CEB@ix.netcom.com> Message-ID: I have a question: Why would a model need to be either ABM/IBM or ODE/PDE? It would seem that a model which even approached natural complexity would be extremely difficult to construct or analyze if one held strictly to one paradigm or the other. I think that Swarm lends itself quite handily to constructing models where the framework is agent-based, but the behavior of the agents themselves can be modeled using differential equations (for cases where the continuity assumption is justified), statistical models, or explicit behavioral descriptions. As soon as finals are done with, I am hoping to get started on my "summer coding project": A Swarm object using multi-layer perceptron code (neural networks) that functions as a black-box statistical model. You give it an input file of data and train a network to fit a statistical distribution. The object can then be used along the lines of the current distribution objects in Swarm--hook up a pseudorandom number generator, and pull numbers out of a distribution which came from real data. I could see this being used, for example, as a "weather object": the intervals between rain events in a simulated prairie ecosystem are reasonably modeled without having to model butterflies flapping their wings. Any thoughts? By the way--I am looking for a graduate program in CS or Ecology to work with this stuff, starting around the fall of '98. Any suggestions? Mattias V. Bertelsen mattias@spaceship.com http://www.spaceship.com/~mattias On Fri, 18 Apr 1997, Mark P. Line wrote: > Scott Christley wrote: > > > > I am tending to notice a separation between two modelling paradigms: the > > ABMers (I was an IBMer once! ;-) and the ODEers. Are there true > > differences between these two paradigms or is it only a perception? > > Meaning is not a Swarm program a symbolic (mathematical) description of a > > model, just not as concise as a differential equation? > > In ODE-based and PDE-based models, the abstractions made are the ones > that are imposed by the kind of symbolic manipulation (analytic > solution) that is possible with simple ODE and very simple PDE systems. > In other words: calculus tells us the function has to be continuous, so > suddenly we are forced into an abstraction of our population of > wildebeests such that population size is a real number and population > growth is a continuous function. Biologically (or sociologically, or > whatever), we tend to remain rather unconvinced that either one of these > assumptions is particularly realistic. ODE's and PDE's were invented so > that problems could be solved analytically. But few interesting problems > that we'd model with these formalisms are soluable analytically anyway, > so there's no longer any good reason to use them, and one very good > reason not to use them: they force on us an abstraction that serves > merely a by-gone purpose and which is usually not warranted in the > biology (or sociology). > > [As long as you're just doing some simple thumbnail models of biomass > and energy balances and what-not, you might be safe up to a point just > doing your ODE's, of course. I don't want to deny that.] > > Now I can say what I think the answers are to the questions you pose > above. > > Are there true differences between these two paradigms? Yes, because > ABM's don't force abstractions on us that are artifacts of a method > whose day is past and which are not desired otherwise in our models. > > Is a Swarm program just a less concise description of a the same thing a > differential equation describes? No, certainly not. A Swarm program does > not normally describe the effects of fractional wildebeests pairing up > and producing fractional young. It describes real, whole wildebeests > pairing up and producing real, whole young. If anything, a Swarm model > is _more_ concise. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 21 17:17:51 1997 From: swarm-modelling@santafe.edu (alan penn) Date: Mon, 21 Apr 1997 16:17:51 +0000 Subject: calibration Message-ID: <43859.9704211611@bas-a.bcc.ac.uk> To throw in twopenny worth, >> Perhaps you are talking about ecosystems from your simulation-oriented >> point of view, as the word 'calibrate' indicates. > >Perhaps I am. > >> I think that a complex >> system like an ecosystem cannot be modeled with some large FSM or even >> with a set of stochastic equations (except in the short-term): in an >> ecosystem there is probably no unique latent 'model' to be discovered, >> against which we could compare some data and make adjustments. Instead >> there probably exists many possible modes, with the system unpredictably >> switching from time to time to one or another. > >I am not sure that we could argue this one way or the other. I see >no reason why a model could not be developed. When you say "many >possible modes", this can easily be modeled by stepping up a level >of abstraction so that the modes are modeled as well. > >> The first thing to do could be to try to recognize these modes. > >Yes, as part of our systems identification process. > >> Because >> switching is unpredictable (ecosystems seem to exhibit self-organized >> critical modes), it makes no sense to try to 'calibrate' the whole system, >> in the same manner simulation engineers calibrate a model of a factory or a >> flexible workshop. Try this one. Perhaps the 'calibration' is in the environment. What I mean by this is that the configuration of the spatial environment inhabited by moving and stationary individuals in the ecosystem could in principle carry with it the 'model' (or several models) inherent in patterns of co-occupancy of space. Think, for example, about human systems like buildings or cities. Modern societies in Levi-Strauss 'statistical' sense, consist in a set of probabilities describing regularities - eg: within race marriages are more likely than inter racial marriages, but there is no rule system prohibiting them as there might have been in more primitive 'mechanical' social forms. How could such probabilistic regularities emerge and be maintained? One possibility is that they arise through the way that spatial configuration and movement of individuals bring different groups into contact. Most geographers treat space as 'map space' in that it is essentially open and homogeneous in all directions, but as soon as you start to think about the space through which we actually move, within and between buildings, then the probabilities of co-presence become significantly structured. If you then map onto this sort of spatial configuration the locations of particular individual's or social groups' facilities, the places they must visit regularly in their daily lives, then you can create a structured, but essentially probabilitic set of interfaces. You could envisage a single spatial configuration coupled to multiple sets of group's 'programmes' giving rise to just the sort of latent model with multiple realisations that complex things like societies require. A final point - perhaps it is the mapping of spatial configuration and cultural 'programmes' - which are actually constructed by people after all - that provides the locus for reproduction of the regularities that we call social?? Alan Penn ________________________________________________ Alan Penn The Bartlett School of Architecture and Planning Philips House (Room 335) University College London, Gower Street, London WC1E 6BT tel. (+44) (0)171 387 7050 ext 5919 fax. (+44) (0)171 916 1887 email. a.penn@ucl.ac.uk www. http://doric.bart.ucl.ac.uk/web/Pangea/index.html ________________________________________________ ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 15:31:11 1997 From: swarm-modelling@santafe.edu (Philippe LAVAL) Date: Tue, 22 Apr 1997 13:31:11 -0100 Subject: calibration Message-ID: <3.0.32.19970422133110.006d648c@poseidon.obs-vlfr.fr> At 16:17 21.04.97 +0000, Alan Penn wrote: >[...] >Try this one. Perhaps the 'calibration' is in the environment. What I mean >by this is that the configuration of the spatial environment inhabited by >moving and stationary individuals in the ecosystem could in principle carry >with it the 'model' (or several models) inherent in patterns of >co-occupancy of space. >[...] > >Most geographers treat space as 'map space' in that it is essentially open >and homogeneous in all directions, but as soon as you start to think about >the space through which we actually move, within and between buildings, >then the probabilities of co-presence become significantly structured. If >you then map onto this sort of spatial configuration the locations of >particular individual's or social groups' facilities, the places they must >visit regularly in their daily lives, then you can create a structured, but >essentially probabilitic set of interfaces. You could envisage a single >spatial configuration coupled to multiple sets of group's 'programmes' >giving rise to just the sort of latent model with multiple realisations >that complex things like societies require. > >Alan Penn > I like your idea of space being structured by the individuals. In " The representation of space in an object-oriented computational pelagic ecosystem", Ecol. Modelling, 88:113-124 (1996), I tried to model space as a shared resource accessed concurrently by individuals, in a first come, first served basis. Here it was food that structured space. This approach was probably adequate because the individuals were identical filter-feeders competing for a resource distributed unevenly in (geographical) space. But if there are different kinds of individuals, it makes sense to consider constraints in the accessibility of space: stronger individuals (or individuals with higher capabilities) have access to a larger range of geographic positions, so they may exploit more resources. Of course resource distribution still shapes the space, in the sense that where a resource is absent, even strong individuals get nothing. But the superposition of different 'ambits' in the same geographic space may better model the competition, or coexistence, of individuals in space. -------------------------------------------------------------------- Philippe Laval Station zoologique B.P. 28 - 06234 Villefranche-sur-Mer CEDEX (France) laval@ccrv.obs-vlfr.fr ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 19:06:03 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 22 Apr 1997 12:06:03 -0600 Subject: parallelism! In-Reply-To: <199704221748.KAA15917@linus.net-community.com> References: <199704221748.KAA15917@linus.net-community.com> Message-ID: <199704221806.MAA21765@grasshopper.santafe.edu> Scott Christley writes: > I'm trying but I haven't quite figured out what MPI stands for (multiple > process interface?). Message Passing Interface, I believe. > Its also unclear to me, probably because my understanding of Swarm is > minimal now, but what type of parallel models are you allowing/using? > > Local memory and message passing with no shared memory? > > What network topologies are you expecting to support? > > What type of memory models are you expecting to support? Ideally, we would use pure message passing with now shared memory. This should allow us to spread swarms over several independent machines in a transparent network. But, we will have to have one hub machine that should synchronize all the processes on the other machines. However, we may use some type of virtual shared memory. > Is the user expected to perform the problem decomposition and Swarm will > handle the distribution; is the user expected to specify both, or is Swarm > going to be able to perform both based upon common model "skeletons" or types. Again, ideally, the user won't have to think about distributing objects (in reality, of course, she will [grin]). The idea is to allow the user to program a model without thinking too much about programming the computer[s]. > I also see a different viewpoint when you discuss parallelism-1; which is > parallelism not of a single program, but of running many concurrent > simulations. Presumably this is important because a researcher may perform > 1000 runs of a simulation and perform some statistical analysis on the results. First off, none of this is true "parallelism". That's a complete misnomer. It's really concurrency (in the abstract) and distributed computing. So, if we just accept the usage of the word "parallel" to mean "doing two or more things at the same time", then we're ok. [grin] But, semantics aside, ||-1 is important to a small extent in allowing a user to run a bunch of well-specified independent Swarms at the same time. The only thing this really saves is time. (There's a continuum between ||-1 and ||-2 and somewhere in between the two, we could make use of the data being generated by any one indep. swarm and make decisions to cancel some swarms as dead-ends in a search or whatnot...) And it will help us work an implementation of MPI into Swarm gently. Other than that it's pretty trivial. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 19:34:10 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Tue, 22 Apr 97 14:34:10 EDT Subject: hex lattice2d In-Reply-To: <199704221806.MAA21765@grasshopper.santafe.edu>; from "glen e. p. ropella" at Apr 22, 97 12:06 (noon) Message-ID: <9704221843.AA20545@sfi.santafe.edu> Hi, gang, Did anyone ever do a 2d hex lattice implementation? Cheers, Ginger ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 20:19:39 1997 From: swarm-modelling@santafe.edu (Theodore C. Belding) Date: Tue, 22 Apr 1997 15:19:39 -0400 (EDT) Subject: parallelism! In-Reply-To: <199704221806.MAA21765@grasshopper.santafe.edu> Message-ID: I hope that none of this takes any resources from documentation and writing the manual; that's what's really needed right now. ||-Swarm is cool, but it can wait. -Ted -- Ted Belding University of Michigan Program for the Study of Complex Systems ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 20:39:04 1997 From: swarm-modelling@santafe.edu (Barry McMullin) Date: Tue, 22 Apr 1997 13:39:04 -0600 Subject: hex lattice2d In-Reply-To: <9704221843.AA20545@sfi.santafe.edu> References: <199704221806.MAA21765@grasshopper.santafe.edu> <9704221843.AA20545@sfi.santafe.edu> Message-ID: <199704221939.NAA03018@tsankawi.santafe.edu> Ginger Booth writes: > Did anyone ever do a 2d hex lattice implementation? Hmmm ... it all depends, of course. But my tipsybugs app is still in the archives at: ftp://ftp.santafe.edu/pub/swarm/users-contrib/anarchy/tipsybugs-0.05.tar.gz It doesn't quite have a hex "lattice" as such, but rather a generic lattice that supports "coord" objects; these coord objects have functionality for getting at "neighboring" coord objects, where the "neighboring" relationship can be flipped between von Neumann and Moore on a square tiling, and FHP (effectively hexagonal) on a triangular tiling. Enjoy, Barry. -- ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ | Barry McMullin, ALife Group, | McMullin@santafe.edu | | Santa Fe Institute, 1399 Hyde Park Road, | Voice: +1-505-984-8800 | | Santa Fe, NM 87501, USA. | FAX: +1-505-982-0565 | | http://www.eeng.dcu.ie/~mcmullin | | ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 13:43:42 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Wed, 23 Apr 1997 06:43:42 -0600 Subject: forwarded from Rosaria Conte (Soc. Sim. conf) Message-ID: <199704231243.GAA00205@seamus.dischordia.com> ------- start of forwarded message (RFC 934 encapsulation) ------- Return-Path: Received: from naga.mailbase.ac.uk by kc.trail.com with smtp (Linux Smail3.1.29.1 #3) id m0wJbfS-000JlQC; Tue, 22 Apr 97 03:15 MDT Received: by naga.mailbase.ac.uk id (8.7.x for naga.mailbase.ac.uk); Tue, 22 Apr 1997 09:33:31 +0100 (BST) Received: from pscs2.irmkant.rm.cnr.it by naga.mailbase.ac.uk id (8.7.x for naga.mailbase.ac.uk) with SMTP; Tue, 22 Apr 1997 09:32:48 +0100 (BST) Received: from [150.146.7.121] (conte.irmkant.rm.cnr.it) by pscs2.irmkant.rm.cnr.it (4.1/1.34) id AA07227; Tue, 22 Apr 97 10:27:40 +0200 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-List: simsoc@mailbase.ac.uk X-Unsub: To leave, send text 'leave simsoc' to mailbase@mailbase.ac.uk Reply-To: rosaria@pscs2.irmkant.rm.cnr.it (Rosaria Conte) Precedence: list From: rosaria@pscs2.irmkant.rm.cnr.it (Rosaria Conte) Sender: simsoc-request@mailbase.ac.uk To: simsoc@mailbase.ac.uk Subject: ICCS&SS Call for Participation Date: Tue, 22 Apr 1997 10:31:03 +0200 Dear Colleague, for information concerning the "International Conference on Computer Simulation and the Social Sciences" (ICCS&SS) to be held in Cortona (AR), Italy, 22-25 September 1997, please visit the following site http://pscs2.irmkant.rm.cnr.it/users/rosaria/CallPart.html Thanks for your attention rosaria conte Rosaria Conte, Division of AI, Cognitive and Interaction Modelling - PSS (Project on Social Simulation) IP/Cnr, V.LE Marx 15 - 00137 Roma voice: +39+6+86090210 fax: +39+6+86090214 email: rosaria@pscs2.irmkant.rm.cnr.it http://pscs2.irmkant.rm.cnr.it/users/rosaria/home.html ------- end ------- ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 05:49:06 1997 From: swarm-modelling@santafe.edu (Steve Emsley) Date: Wed, 23 Apr 1997 05:49:06 +0100 Subject: ABM/I(C?)BM In-Reply-To: (donalson@lifesci.lscf.ucsb.edu) Message-ID: <199704230449.FAA00801@archon> Perhaps I'm tagging this onto the wrong thread but Doug Donalson refers to a reference that I had in mind as I brought up the ODE / ABM dichotomy. Chapter 3 of DeAngelis' book on IBM (referred to by Doug) on IBM is by Caswell & Meredith John. In this they differentiate between the i-state configuration model (individual-based with individual differences), the p-state (a model of the population as a mean of the i-state configuration) and the i-state distribution ( a simplification of the p-state arising when all individual's experience the same environment). ODE's are justifiably used to model the p-state (including the special case of the i-state distribution). However, the authors suggest that the i-state configuration is important with "complicated i-states, small populations and local interactions". In addition, the p-state is derivable from the i-state whereas the i-state not derivable from the p-state. As a parting thought they suggest that "i-state configuration models may be useful ... to drive maximum likelihood parameter values [of i-state distribution models]". Personally, I view this distinction as having more mileage than an argument on the relative merits of ODE/PDE vrs. ABM(IBM) models. The latter being a set of tools to understand/describe/model the former. One posting suggested that ODEs are attractive due to the possibility of their analytical solution. IMHO if you have an ODE-based ecology capable of analytical solution (rather than numerical simulation) you are dealing with a mathematical abstraction (ecology as an excitable medium) NOT a system that propagates through time based on local interactions and continually varying stochastic or adaptive parameters i.e. a real ecosystem. It is not unusual for ODE/PDE models to be "tuned" to actual data though modifying the closure terms. Better, in my opinion, to turn one's back on parsimony and computational efficiency in order to model the i-state. If there's no significant difference between the i-state configuration (IBM) model and the p-state (ODE/PDE) model then efficiency favours the latter. In my field the ergonomics and economics of sampling skew the available data towards coarse resolution. To fit a multi-parameter mean-field model to such data could, unsympathetically I admit, be called tautologous. By my, naive, appraisal of current 'scientific method' I see modelling driving observation - a reversal of the traditional paradigm. To assume a mean-field solution from the onset establishes a priori length scales and time scales to the system under investigation which, a posteriori, can establish observational criteria. I suppose that the impact of IBMs on conventional mean-field modelling may be analogous to the impact of Kepler's First Law on the aesthetic bias of the Platonic circular orbit - as observations become more exact then no system of epicycles (or parameters) will exactly fit the facts. Regards -- -- Steve Emsley sme@oikos.warwick.ac.uk ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 18:15:16 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Wed, 23 Apr 1997 11:15:16 -0600 Subject: forwarded from Michael Lissack (boston CS conf.) Message-ID: <199704231715.LAA23265@grasshopper.santafe.edu> ------- start of forwarded message (RFC 934 encapsulation) ------- Reply-To: lissack@lissack.com Organization: Michael Lissack Comments: To: complex@lissack.com From: Michael Lissack Sender: Complexity and Management Mailing List To: COMPLEX@LISSACK.SPACELAB.NET Subject: [Fwd: Complex Systems Conference and Book] Date: Tue, 22 Apr 1997 17:56:35 -0400 ---------------------forward from Sean Pidgeon--------------------- Date: Tue, 22 Apr 1997 12:52:55 -0400 From: Sean Pidgeon Subject: Complex Systems Conference and Book To: lissack@lissack.com Message-id: MIME-version: 1.0 X-Mailer: Novell GroupWise 4.1 Content-type: text/plain Content-disposition: inline Content-transfer-encoding: 7BIT Dear Dr Lissack: I have appended below an announcement for a multidisciplinary conference on complex systems, to be chaired by Prof. Yaneer Bar-Yam of Boston University under the auspices of the New England Complex Systems Institute. The meeting has been organized in partnership with Oxford University Press. A book provisionally entitled "Complex Systems: A Multidisciplinary Sourcebook", to be edited by Dr. Bar-Yam and published by OUP, will draw upon the multidisciplinary themes and participants in the conference. The conference will, we hope, attract participation from many areas of scientific research. I would be most grateful if you could forward to me or to Dr. Bar-Yam (yaneer@bu.edu) any names (or mailing lists) of other researchers who might be interested in attending or participating in the meeting. We are particularly interested in knowing which individuals can speak or write about significant contributions to our understanding of the universal properties of complex systems -- e.g. as articulated in the themes section of the conference announcement -- and their application across disciplinary boundaries. Many thanks. - ---------------------------------------------- Sean Pidgeon Senior Editor Oxford University Press 198 Madison Avenue New York, NY 10016 phone: (212) 726-6134 fax: (212) 726-6445 sdp@oup-usa.org http://www.oup-usa.org/acadref/sdp.html phone: (212) 726-6134 fax: (212) 726-6445 sdp@oup-usa.org http://www.oup-usa.org/acadref/sdp.html - --------------------------------------------- First Announcement International Conference on COMPLEX SYSTEMS Boston Area September 21-26, 1997 Host: New England Complex Systems Institute http://necsi.org necsi@necsi.org With: Oxford University Press Conference Chairman: Yaneer Bar-Yam ORGANIZING COMMITTEE: Philip Anderson - Princeton University Kenneth J. Arrow - Stanford University Per Bak - Niels Bohr Institute Charles H. Bennett - IBM William A. Brock - University of Wisconsin Charles R. Cantor - Boston University Noam A. Chomsky - MIT Leon Cooper - Brown University Daniel Dennett - Tufts University Irving Epstein - Brandeis University Michael S. Gazzaniga - Dartmouth College William Gelbart - Harvard University Murray Gell-Mann - CalTech / Santa Fe Institute Pierre-Gilles de Gennes - ESPCI Stephen Grossberg - Boston University Michael Hammer - Hammer & Co John Holland - University of Michigan John Hopfield - Princeton University Jerome Kagan - Harvard University Stuart A. Kauffman - Santa Fe Institute Chris Langton - Santa Fe Institute Richard C. Lewontin - Harvard University Andrew W. Lo - MIT Marvin Minsky - MIT Alan Perelson - Los Alamos National Lab Herbert A. Simon - Carnegie Mellon University Temple F. Smith - Boston University H. Eugene Stanley - Boston University James H. Stock - Harvard University Gerald J. Sussman - MIT Edward O. Wilson - Harvard University SUBJECT AREAS: UNIFYING THEMES IN COMPLEX SYSTEMS Sessions will be structured around both themes and systems. The themes are: EMERGENCE, STRUCTURE AND FUNCTION: substructure; the relationship of component to collective behavior; the relationship of internal structure to external influence. INFORMATICS: structuring, storing, accessing, and distributing information describing complex systems. COMPLEXITY: characterizing the amount of information necessary to describe complex systems, and the dynamics of this information. DYNAMICS: time series analysis and prediction, chaos, temporal correlations, the time scale of dynamic processes. SELF-ORGANIZATION: Evolution, development and adaptation. The system categories are: FUNDAMENTALS OF COMPLEX SYSTEMS: Complexity, emergence, chaos, fractals, non-equilibrium processes, dynamic scaling, information and computation in physical systems. MOLECULAR SYSTEMS: Chemical dynamics, complex fluids, molecular self-organization, membranes, protein and DNA folding, bio-molecular informatics. CELLULAR SYSTEMS: Cellular response and communication, genetic regulation, gene-cytoplasm interactions, development, cellular differentiation, primitive multicellular organisms, the immune system. PHYSIOLOGICAL SYSTEMS: Nervous system, neuro-muscular control, neural network models of brain, cognition, psychofunction, pattern recognition, man-machine interactions. HUMAN SOCIAL AND ECONOMIC SYSTEMS: Corporate and social structures, markets, the global economy, the Internet. PEDAGOGICAL SESSIONS: The conference will include pedagogical sessions on Sunday, Sept. 21. ** A detailed announcement including instructions for submission of abstracts will follow.** If you want to receive future announcements about this conference, please e-mail us at necsi@necsi.org Include: Your name: ________________________________________ Preferred e-mail address: _________________________ If you want to be removed from this list please send us a note including the statement: "Please remove my e-mail address from this list." and include the e-mail address from which you received this announcement. - --------------03B6B28B4FD021B0B3FAF09B-- ------- end ------- ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 18:38:52 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Wed, 23 Apr 1997 18:38:52 +0100 (BST) Subject: Pathology of discrete diffusion Message-ID: Hi swarmers, some time ago there was a discussion of pathological cases of diffusion among the list. One such case was brought to light by David Sumpter: > Hello everyone, > > 1. I am using a technique similar to Diffuse 2d for the diffusion of > heat and am slightly confused by one aspect. Consider a portion of > lattice with temperatures and first order diffusion: > > 0 > 080 > 0 > > If the diffusion constant is 1.0 and evaporation rate is 1.0 then the > centre value will update to 0 while the outside values will update to > 2. i.e. > > 2 > 202 > 2 > > This appears somewhat unnatural for heat equations. If the value 8 came > from some source, you would not expect the source to be colder than > the surroundings on the next time step........ > > Am I right about this? Is there a theoretical explanation? > > 2. Has anyone got a reference they can give me discussing heat > diffusion in terms of lattices? I'd be very grateful. > > Thanks, > > David. > I wonder if someone knows of other pathological cases and if there is a way to avoid such cases within a discrete diffusion frame. Has the above problem been solved or ignored? Any comments welcome. Cheers, Jan. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 19:57:03 1997 From: swarm-modelling@santafe.edu (David Sumpter) Date: Wed, 23 Apr 1997 19:57:03 +0100 Subject: Pathology of discrete diffusion Message-ID: <199704231857.TAA07158@beehive.ma.umist.ac.uk> > 1. I am using a technique similar to Diffuse 2d for the diffusion of > heat and am slightly confused by one aspect. Consider a portion of > lattice with temperatures and first order diffusion: > > 0 > 080 > 0 > > If the diffusion constant is 1.0 and evaporation rate is 1.0 then the > centre value will update to 0 while the outside values will update to > 2. i.e. > > 2 > 202 > 2 > > This appears somewhat unnatural for heat equations. If the value 8 came > from some source, you would not expect the source to be colder than > the surroundings on the next time step........ The way to avoid such problems is to choose a correct value for the diffusion constant, we'll call D. This discretisation is the Laplacian Difference Equations for Diffusion. There are whole numerical analysis departments working on what the value of D should be for different problems. THe solution lies in the nature of the problem. It is smaller values of D which give more accurate results but obviously mean a larger diffusion time where diffusion time is proportional to (size of single lattice cell)^2 / D . The important thing is to view D as a constant determining the accruacy of your simulation and not to use it as an element of time. This is the trap I fell into at first. Instead of changing the diffusion constant to change the time for diffusion to take place, change the time steps at which agents update their position on the associated world lattice. Using this technique you can set the diffusion constant as low as you like, computing time allowing. In the end I just used D=0.5 so that I get, 1 141 1 Quite nice, I suppose. Good luck, David Sumpter ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 23:38:29 1997 From: swarm-modelling@santafe.edu (Chris Landauer) Date: Wed, 23 Apr 97 15:38:29 PDT Subject: correction Message-ID: <199704232238.PAA29770@antares.aero.org> the URL for the complex systems meeting is wrong on their announcement 8-) it should be http://www.necsi.org (they seem to have forgotten the www.) more later, cal ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 23:44:20 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Wed, 23 Apr 1997 15:44:20 -0700 Subject: ABM/I(C?)BM References: <199704230449.FAA00801@archon> Message-ID: <335E90C4.444A22CF@ix.netcom.com> Steve Emsley wrote: > > One posting suggested that ODEs are attractive due to the possibility > of their analytical solution. I don't remember anybody saying that, although I wrote that ODE's were invented in order to provide analytical solubility for otherwise intractable problems. I also wrote that no interesting problem formulated as ODE's or PDE's is likely to be analytically soluble, and that therefore their rationale as tools for analytical manipulation no longer exists. In saying that, I was trying to imply that, if we're going to have to attack our models numerically anyway, then we might as well go straight for an ABM representation instead of blissfully accepting the inappropriate assumptions (e.g. differentiability) of ODE's and PDE's. Those assumptions were the lesser evil when the problem at hand was otherwise intractable, period. That's no longer the case in domains where we can build useful ABM's. > IMHO if you have an ODE-based ecology > capable of analytical solution (rather than numerical simulation) you are > dealing with a mathematical abstraction (ecology as an excitable medium) > NOT a system that propagates through time based on local interactions and > continually varying stochastic or adaptive parameters i.e. a real > ecosystem. Precisely. It is unlikely that a useful [OP]DE model of an observed system will be capable of analytical solution. -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 01:14:31 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Wed, 23 Apr 1997 17:14:31 -0700 Subject: Pathology of discrete diffusion References: Message-ID: <335EA5E7.26DF6028@ix.netcom.com> Jan Kreft wrote: > > David Sumpter wrote: > > > > 1. I am using a technique similar to Diffuse 2d for the diffusion of > > heat and am slightly confused by one aspect. Consider a portion of > > lattice with temperatures and first order diffusion: > > > > 0 > > 080 > > 0 > > > > If the diffusion constant is 1.0 and evaporation rate is 1.0 then the > > centre value will update to 0 while the outside values will update to > > 2. i.e. > > > > 2 > > 202 > > 2 > > > > This appears somewhat unnatural for heat equations. If the value 8 came > > from some source, you would not expect the source to be colder than > > the surroundings on the next time step........ > > > > Am I right about this? Is there a theoretical explanation? I don't think I saw the original discussion here, so maybe there's something I'm missing. Fill me in if I'm not answering the question you're asking. Fourier's law says that the rate at which thermal energy flows between two points (or in the discrete case, between two adjacent lattice cells) is proportional (by a coefficient of thermal conductivity) to the negative gradient of temperature between the points (or cells). [Real diffusion involves both storage and flow, so you'll also have to deal with the material's thermal _capacity_, eventually.] For a large delta-t, I think you'll find that the gradient sometimes reverses sign and that temperatures damped-oscillate into the steady state. As you decrease delta-t, the model will equilibrate with fewer and more highly damped oscillations. In continuous time, thermal energy flows smoothly over the gradient until the steady state is reached (other things being equal). But even in continuous time, I wouldn't be surprised to find damped oscillations if you're starting from a thermally very heterogeneous body (hot or cold spots) and if the material has a high lambda (thermal conductivity). > > 2. Has anyone got a reference they can give me discussing heat > > diffusion in terms of lattices? I'd be very grateful. I haven't seen these for a while, but I think they have part of what you're looking for: Myrup, L.O. (1969) "A numerical model of the urban heat island", _Journal of Applied Meteorology_ 8: 908-918. Sellers, W.D. (1973) "A new global climate model", _Journal of Applied Meteorology_ 12: 241-254. Wierenga, P.J./de Wit, C.T. (1970) "Simulation of heat transfer in soils", _Proc. Soil Sci. Am._ 34: 845-848. But you can always go check the journal devoted to the topic: _Numerical Heat Transfer. Part A, Applications_ _Numerical Heat Transfer. Part B, Fundamentals_ and its precursor (same name, without the split). You could browse through all the article titles in the CARL database if you don't have easy access to the journals themselves. -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 03:19:57 1997 From: swarm-modelling@santafe.edu (Chris Landauer) Date: Wed, 23 Apr 97 19:19:57 PDT Subject: mathematical nonsense Message-ID: <199704240219.TAA02838@antares.aero.org> this one i can't pass up - the anti-intellectualism is stunning Mark P. Line wrote: I wrote that ODE's were invented in order to provide analytical solubility for otherwise intractable problems. I say: This is simply incorrect - ODE's were invented to provide analytic _models_ of complicated phenomena, usually in physics at first, and only some of them were ever analytically soluble. Mark P. Line wrote: I also wrote that no interesting problem formulated as ODE's or PDE's is likely to be analytically soluble, and that therefore their rationale as tools for analytical manipulation no longer exists. I say: The first part of this is correct, but the second part is not. There are many more analytic results that can be obtained from an ODE or PDE model than the solution. Indeed, many mathematicians have proved long term stability results for systems that have no hope of analytic solution. The power of an analytic formulation is _not_ about solutions; it is about understanding, and there are many different kinds of useful models. If you naively limit your understanding of mathematical analyses to "solutions", then you miss many of the most interesting and exciting results. Mark P. Line wrote: In saying that, I was trying to imply that, if we're going to have to attack our models numerically anyway, then we might as well go straight for an ABM representation instead of blissfully accepting the inappropriate assumptions (e.g. differentiability) of ODE's and PDE's. Those assumptions were the lesser evil when the problem at hand was otherwise intractable, period. That's no longer the case in domains where we can build useful ABM's. I say: There is a grain of truth in some of this, in that the mathematical assumptions often do not reflect the reality, but the same is true of the ABM models also, and anyone who forgets that is likely to get less than useful results. It also ignores the fact that many ODE and PDE models of complex phenomena are _much_ more easily analyzed (even without solutions) than the corresponding collections of individual actors. There are many people who think that modeling individual behavior in a complex system is somehow "less modeling; more reality", but that is too often just plain wrong, because the important behaviors are lost in the shuffle of complex interactions. The modeler must make choices either way; different choices make more sense for different purposes. For example, starting almost 100 years ago, Poincare and others studied long term stability of ODE's, such as those derived from the 3-body problem, and Fatou and Julia studied certain limiting sets of infinite processes in the complex plane, with no hint of an analytic "solution". My personal opinion is that the computational power available currently allows much modeling laziness, and while sometimes helpful (I have on many occasions used computer experimentation to give me insights about tricky problems), it actually very often inhibits clear and useful thinking about complex models. Don't throw away useful tools; we need all the help we can get. more later, cal Dr. Christopher Landauer National Systems Group, The Aerospace Corporation The Hallmark Building, Suite 187 13873 Park Center Road, Herndon, Virginia 20171 e-mail: cal@aero.org Phone: (703) 318-1666, FAX: (703) 318-5409 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 03:08:13 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Wed, 23 Apr 1997 19:08:13 -0700 Subject: mathematical nonsense References: <199704240219.TAA02838@antares.aero.org> Message-ID: <335EC08D.7B2DD48A@ix.netcom.com> Chris Landauer wrote: > > this one i can't pass up - the anti-intellectualism is stunning If I've succumbed to anti-intellectualism, then I guess I shouldn't be here. Bye. -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 05:45:38 1997 From: swarm-modelling@santafe.edu (Doug Donalson;) Date: Wed, 23 Apr 1997 21:45:38 -0700 (PDT) Subject: mathematical nonsense In-Reply-To: <335EC08D.7B2DD48A@ix.netcom.com> Message-ID: I want to thank Mark and Chris for holding a mirror up to my face. I can't read the last two articals without feeling somewhat shagrined as I have had a part in a piece of this discussion that should never have started. There is no place in this group for name calling and trashing of others viewpoints. This includes my comment that the truth hurts (or some such stupid thing.) At least in the area of theoretical population ecology, people who have attempted to introduce spatially-explict individual-based models have been met (by some) with scepticisum and sometimes disdain. It is very easy to fall into the same trap and find all the faults of the more mainstream ODE type models. Engaging in trash talk does nothing but set back the real purpose (I hope?) of better understanding complex interactions. There is no point in rejecting potentially useful tools (ODE or ABM) either because of "not invented here" or "you trash me so I'll trash you". At UCSB we are using analytical or numerical solutions of ODE type models as a first step in understanding and verifying the more complex models. Science involves the exchange of ideas in a conflict/resolution format. This can be done in either a postive or negative manor ( pun intended :-) ) This is something my advisor (Roger Nisbet) has gently tried to pound into my thick skull. The best justification for complex models is good structured analysis of their strengths and weakneses. Comparision between these and the simpiler models will allow us to understand when various assumptions are or are not valid and give us a better understanding of the level of detail necessary in a model to gain insite into a particular problem. There is a place in this email group for anyone who is working with SWARM and this is a very valuable thread. I hope it continues. Lets try to keep the debate to constructive comments. Cheers, Doug Donalson *************************************************************************** * Doug Donalson * Office: (805) 893-2962 * * Ecology, Evolution, and Marine Biology * Home: (805) 961-4447 * * UC Santa Barbara * email donalson@lifesci.ucsb.edu* * Santa Barbara Ca. 93106 * * *************************************************************************** * * * The most exciting phrase to hear in science, the one that * * hearlds new discoveries, is not "EUREKA" (I have found it) but * * "That's funny ...?" * * * * Isaac Asimov * * * *************************************************************************** ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 18:46:09 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: 24 Apr 1997 17:46:09 -0000 Subject: ABM/I(C?)BM and mathematical nonsense Message-ID: <19970424174609.6508.qmail@mango.tiem.utk.edu> Hi folks, I've been reluctant to jump into the recent discussions, and I'm not certain I've read all of them, but thought I might at least now state a couple of opinions. These are based on my experiences with individual- based models over the last 10 years or so. 1. There are different purposes for modeling, and different approaches appropriate based upon these purposes. Although I have long felt, as an educator, that there is vastly too much emphasis in our current curricula on analytic methods (e.g. ODE, PDE) relative to computational approaches (e.g. rule-based approaches, stochastic simulation), analytic methods allow us to address issues in a different way. The real difficulty, and one that I have the hardest time with in both the formal modeling courses I teach as well as in mentoring graduate students, is deciding which approach is most appropriate for the problems you wish to address. A group of us working on models for Everglades restoration (the ATLSS project - home page at http://www.tiem.utk.edu/~gross/atlss_www/atlss_frame.html) have dealt with these issues for at least one major environmental project. We have concluded that multiple approaches are required (a multimodel, in the sense that Paul Fishwick has defined it), based on the varying spatial and temporal resolutions and associated organismal detail needed to answer the questions of interest in restoration. So we mix ODE models, structured matrix population models, and individual-based models. 2. There are indeed very competent folks who disagree with my belief that individual-based approaches are the appropriate method to address many problems in ecology arising from practical concerns. For one opinion, see Levin et al. (1997) Science 275, 334. The authors of that paper and I disagree on several points, as in particular I cannot concur with their characterization that individual-based models as producing "cartoons that may look like nature but represent no real systems". These differences of opinion are based upon different views of what we can hope to attain by modeling, as well as more philosophical considerations regarding what can arise from reductionist approaches. 3. On nomenclature, an ecologist knows what an individual is (OK - so clonal organisms produce problems here, but that's another topic), and therefore intuitively understands what individual-based approaches mean with very little explanation. I have no desire to try to explain what an "agent" is - I will continue to use the term individual-based models when dealing with ecologists. Names are important. Hal and Meredith's definition of the i- and p-state and configuration models is highly useful, but these have not really entered the ecology literature - people find them too confusing. I have given up trying to explain them to audiences. Cheers, Lou Gross Professor of Ecology and Evolutionary Biology and Mathematics The Institute for Environmental Modeling University of Tennessee - Knoxville gross@tiem.utk.edu http://www.tiem.utk.edu/~gross/ http://archives.math.utk.edu/mathbio/ (Math Archives for Life Sciences) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 18:52:22 1997 From: swarm-modelling@santafe.edu (Sven N. Thommesen) Date: Thu, 24 Apr 1997 12:52:22 -0500 Subject: correction Message-ID: <3.0.32.19970424125218.0095a160@spidle2.humsci.auburn.edu> At 03:38 PM 4/23/97 PDT, Chris Landauer wrote: > >the URL for the complex systems meeting is wrong on their announcement 8-) > >it should be http://www.necsi.org >(they seem to have forgotten the www.) > >more later, >cal Sorry, wrong-o! The site can be accessed either way. (There is no requirement that a URL start with 'www.') Question: was this 'correction' based on empirical knowledge, or on a priori assumptions ... ? (I guess I should refrain from making comments about stunning anti-intellectualism, but the temptation is awesome ... ;-) Cheers, --Sven ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 25 01:06:14 1997 From: swarm-modelling@santafe.edu (Chris Landauer) Date: Thu, 24 Apr 97 17:06:14 PDT Subject: www missing - for me, if not for you Message-ID: <199704250006.RAA16458@antares.aero.org> >>Sven: me >>Sorry, wrong-o! The site can be accessed either way. >>(There is no requirement that a URL start with 'www.') yes, i know >>Question: was this 'correction' based on empirical knowledge, >>or on a priori assumptions ... ? empirical - i try not to correct things i only _think_ might be wrong my correction was based on 'unable to find DNS entry for necsi.org', and on successfully connecting to www.necsi.org, which i found quite interesting so i still don't know which one(s) may be available from other sites, but the one from the message was not available to me when i tried it 8-) more later, cal ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 25 22:50:11 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Fri, 25 Apr 1997 17:50:11 -0400 (EDT) Subject: invitation Message-ID: <199704252150.RAA14903@tide.cise.ufl.edu> This is an invitation to present a paper (or organize a session) on simulating complex systems (with multiple agents) using either current or future web technology. For details, please refer to: http://www.cise.ufl.edu/~fishwick/webconf.html -paul Paul A. Fishwick E-Mail: fishwick@cise.ufl.edu Dept. of Computer & Info Phone & FAX: (352) 392-1414 Science and Engineering WWW: http://www.cise.ufl.edu/~fishwick University of Florida (PGP Key available at above WWW address) P. O. Box 116120 332 Bldg. CSE, Gainesville, FL 32611-6120 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 18:30:50 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 11:30:50 -0700 Subject: swarm-modelling: testing Message-ID: <199704011830.LAA10170@grasshopper.santafe.edu> Testing the list. glen ================================== swarm-modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 18:37:22 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 11:37:22 -0700 Subject: swarm-modelling: testing Message-ID: <199704011837.LAA10181@grasshopper.santafe.edu> Just testing the list. glen ================================== swarm-modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 19:12:47 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 12:12:47 -0700 Subject: SMod: testing again Message-ID: <199704011912.MAA10209@grasshopper.santafe.edu> Just testing again. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 19:13:48 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 12:13:48 -0700 Subject: SMod: testing again In-Reply-To: <199704011912.MAA10209@grasshopper.santafe.edu> References: <199704011912.MAA10209@grasshopper.santafe.edu> Message-ID: <199704011913.MAA10213@grasshopper.santafe.edu> replying to the test glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 19:31:06 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 12:31:06 -0700 Subject: SMod: New List! Message-ID: <199704011931.MAA10232@grasshopper.santafe.edu> Hey guys, If you'll take a brief look up at the header "From:" field, you'll see that you've been subscribed to a new list, the Swarm-Modelling list. I've appended the info file (normally retrieved via a message to majordomo saying "info swarm-modelling" in the body. That has an attempted explanation of what this list if for. I'd appreciate comments telling me if you think it's worded correctly or if you think the target discussions are defined too narrowly. Let me know what you think. glen -------------------------------------------------------------------- [Last updated on: Tue Apr 1 11:59:27 1997] The Swarm Modelling list, , is an open mailing list intended to provide a forum for the discusion of high level modelling and simulation issues. Special attention should be devoted to using Swarm to do such modelling; but, posts to the list are not restricted to such. This list is not for installation or bug reports for the Swarm package. All posts should be understandable to those (esp. scientist and manager types) who have an interest in systems modelling but do not have a programming background. This list is archived: ask majordomo to "index swarm-modelling" There is a set of documents about Swarm on the World Wide Web at the URL http://www.santafe.edu/projects/swarm/. And there are two other lists of interest to Swarm programmers, the Swarm-Support list and the Swarm-GIS list. Use Majordomo to get the info files for those lists. The swarm developers at SFI can be reached as - this is a very small set of people, those who are currently actively developing the core of Swarm. Welcome aboard! -------------------------------------------------------------------- For list administration needs, please use the Majordomo server at the Santa Fe Institute. For help using Majordomo, send mail to: majordomo@santafe.edu with the following in the body: help -------------------------------------------------------------------- ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 19:48:40 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 12:48:40 -0700 Subject: SMod: List dynamics Message-ID: <199704011948.MAA10242@grasshopper.santafe.edu> With the addition of two new lists (swarm-modelling and swarm-gis), I've decided it's time to get a little more sophisticated with the delivery and content of the mailing lists. So, as usual, I'd like your input on a couple questions: 1) Since we now have 4 lists (announce, support, gis, modelling), simply subscribing the swarm-support list to the swarm-announce list doesn't cut it. It's conceivable that someone may want to be members of both swarm-announce and swarm-modelling but not swarm-support or swarm-announce and swarm-support but not swarm-modelling. (I'm presuming that most of the people on swarm-gis will be members of one of the other two lists, so it's not as much of a consideration, here.) Now, the *best* way I can see this happening is to subscribe anyone who is a member of any one of the other three lists to the announce list. That prevents duplicate posts to any list member and allows all the people on one of the other three lists to receive posts to swarm-announce, as well. Does this sound right? Of course, the problem is automating the subscription to swarm-announce and triggering that subscription from a user-initiated subscription to one of the other lists.... But, I presume I can deal with that. 2) I've added prefixes to the "Subject:" fields of messages from each of the support, gis, and modelling lists so that anybody out there using a mailer (or a mail filter) that is capable of sorting the mail can do so based on that prefix. a) Do you like this or dislike it? b) Would you prefer different prefixes? (currently, they're "SSup," "SGIS," and "SMod") 3) I've added a footer to each message with a short description of the lists purpose and a hint at how to use majordomo (as was suggested by a person who didn't know how to unsub). Again, let me know if this is ok and feel free to criticize my wording. 4) Digests are coming! We don't have them automated yet. But, it should be ok for now. The list is calm. I'll let you know when we have them up. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 20:36:43 1997 From: swarm-modelling@santafe.edu (Pietro Terna) Date: Tue, 01 Apr 1997 22:36:43 +0200 Subject: SMod: List dynamics Message-ID: <1.5.4.32.19970401203643.006897b8@alpcom.it> As 'an incoming true swarm user' I think that all four lists are useful and also necessary, without limitations. I'll work about economic models but ideas from GIS are also stimulating etc. ... So a little question (unfair?): Why not to keep a unique list? Pietro At 12.48 01/04/97 -0700, you wrote: > >With the addition of two new lists (swarm-modelling and >swarm-gis), I've decided it's time to get a little more >sophisticated with the delivery and content of the mailing >lists. So, as usual, I'd like your input on a couple >questions: > >1) Since we now have 4 lists (announce, support, gis, modelling), >simply subscribing the swarm-support list to the swarm-announce >list doesn't cut it. It's conceivable that someone may want to >be members of both swarm-announce and swarm-modelling but not >swarm-support or swarm-announce and swarm-support but not >swarm-modelling. (I'm presuming that most of the people on swarm-gis >will be members of one of the other two lists, so it's not as much >of a consideration, here.) > >Now, the *best* way I can see this happening is to subscribe anyone >who is a member of any one of the other three lists to the announce >list. That prevents duplicate posts to any list member and allows >all the people on one of the other three lists to receive posts to >swarm-announce, as well. > >Does this sound right? > >Of course, the problem is automating the subscription to swarm-announce >and triggering that subscription from a user-initiated subscription >to one of the other lists.... But, I presume I can deal with that. > >2) I've added prefixes to the "Subject:" fields of messages from each >of the support, gis, and modelling lists so that anybody out there >using a mailer (or a mail filter) that is capable of sorting the mail >can do so based on that prefix. > > a) Do you like this or dislike it? > b) Would you prefer different prefixes? (currently, they're > "SSup," "SGIS," and "SMod") > >3) I've added a footer to each message with a short description of >the lists purpose and a hint at how to use majordomo (as was >suggested by a person who didn't know how to unsub). Again, let >me know if this is ok and feel free to criticize my wording. > >4) Digests are coming! We don't have them automated yet. But, >it should be ok for now. The list is calm. I'll let you know when >we have them up. > >glen > > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 1 23:26:09 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 1 Apr 1997 16:26:09 -0700 Subject: SMod: List dynamics In-Reply-To: <1.5.4.32.19970401203643.006897b8@alpcom.it> References: <1.5.4.32.19970401203643.006897b8@alpcom.it> Message-ID: <199704012326.QAA10318@grasshopper.santafe.edu> Pietro Terna writes: > As 'an incoming true swarm user' I think that all four lists are useful and > also necessary, without limitations. I'll work about economic models but > ideas from GIS are also stimulating etc. ... > So a little question (unfair?): Why not to keep a unique list? > Pietro This is a bit of a controversy. Several people feel we don't need more than one list. But, more people seem to feel we do. So, I think it's probably a good idea. I believe their argument goes like this (if any proponents for the list split want to correct me, feel free). Some Swarm users have been using Swarm for awhile, now. They are not as interested in seeing questions they've seen over and over again or in questions relating to some obscure platform. Since alot of people on this list may get several (like...over 100?) email messages a day, it is more convenient to be able to subscribe to lists on which they're fairly sure only "interesting" messages are posted. Right now, this isn't really that much of a problem, because there hasn't been that much traffic. But, in high traffic periods (like after a new release), it can be difficult to sift through the posts to find which ones are "interesting" and which ones aren't. Ideally, with the scheme I'm setting up, you can subscribe to all the lists, if you'd like. In fact, this might be a good idea for a new Swarm user. But, after awhile, you might realize that you're not that interested in one or the other. It should also be amenable to those people who are not actually using Swarm, but who employ (or advise) people who are. Then you could subscribe to the modelling and GIS lists to interact on one level without having to pour through intricacies you're not concerned about. So, basically, it was decided at SwarmFest to split the list into a "code-slinger" type list and a "theory-slinger" type list. And, depending on whether the scheme works or not, we can keep or trash this scheme. If the traffic just continues to grow, then it might be justified to make another split later on. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 2 16:52:35 1997 From: swarm-modelling@santafe.edu (Stephen C. Upton) Date: Wed, 02 Apr 1997 09:52:35 -0700 Subject: Modeling of Crowd Behavior Message-ID: <3.0.32.19970402095234.007255b4@tsa-po.lanl.gov> As a 1.7 Swarm Lurker (between class 1 and 2 -- closer to 2 - I've finally got Linux loaded on a new machine! I've previously had Swarm up and running at another job, but, alas it's not part of my job description here -- yet!), I appreciate the separation of the lists. As of now, I am more interested in modeling aspects than specifics about Swarm code -- that will come :P My question is: has, or is, anyone attempting to look at modeling crowd (of humans) behavior, whether it be with Swarm or without? This has obvious applications for the justice department and the military, for example. They would like to disrupt a crowd using non-lethal technologies, but certain actions may be more provocative than others. I would appreciate any references any one might have also. thanx upton *** * **** * ***** ********* ** ****** ***** *** ***** ******** ********* Stephen C. Upton TSA-5, MS F602 Los Alamos National Laboratory Los Alamos, NM 87545 505-667-9435 FAX 505-665-2017 upton@lanl.gov ******* ********* *** ** *** ******** **** ****** ** ****** **** *** *** ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 2 18:17:26 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Wed, 2 Apr 1997 11:17:26 -0700 Subject: Modeling of Crowd Behavior In-Reply-To: <3.0.32.19970402095234.007255b4@tsa-po.lanl.gov> References: <3.0.32.19970402095234.007255b4@tsa-po.lanl.gov> Message-ID: <199704021817.LAA10633@grasshopper.santafe.edu> Stephen C. Upton writes: > My question is: has, or is, anyone attempting to look at modeling crowd (of > humans) behavior, whether it be with Swarm or without? This has obvious > applications for the justice department and the military, for example. > They would like to disrupt a crowd using non-lethal technologies, but > certain actions may be more provocative than others. > > I would appreciate any references any one might have also. I'm not sure if this is a good reference; but, it sure looks relevant. http://cbl.leeds.ac.uk/rodw/papers/tiemec-95/main.html Also, this seems perfect for a multi-agent system. I guess the problem lies in the agent motivation. If we could identify a common motivation and behavior that's purely local and that gives rise to a riot, we could ethically experiment with riot control. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 00:45:49 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Wed, 2 Apr 1997 17:45:49 -0700 Subject: Taking Snapshots Message-ID: <199704030045.RAA11710@grasshopper.santafe.edu> Hey! Well, I got fed up with Sven's needling [grin] and hacked xwd so that it's callable as a subroutine. It's relatively painless to compile it as a library, link it with Swarm, and put a method in, say, the observerSwarm that is called at whatever frequency you'd like. It's not in distributable form, yet. For instance, the xwd package still tries to do a link when you make it and you have to create your own libxwd.a with an ar rvs command. But, if anybody's got ants in their pants, I can send it to them. I doubt this is the way we should really go. XWD files are pretty large (105583 bytes for 80x80 default zoom heatbugs). Plus, we're trying to move *away* from X if possible, not towards it. So, I've started looking into HDF and AVS. Mind you, however, that this is still not high priority. The hackish stuff is so simple that it can never be considered high priority. And the long-term stuff ties in with our general data visualization problem and our portability problem, both of which are superceded by ||-Swarm, the Manual, and the Alpha port. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 01:08:59 1997 From: swarm-modelling@santafe.edu (Sven N. Thommesen) Date: Wed, 02 Apr 1997 19:08:59 -0600 Subject: Taking Snapshots Message-ID: <3.0.32.19970402190854.0099d700@spidle2.humsci.auburn.edu> At 05:45 PM 4/2/97 -0700, you wrote: > >Hey! > >Well, I got fed up with Sven's needling [grin] and hacked xwd so that Halleluja ! The squeaky wheel finally got a little grease! >it's callable as a subroutine. It's relatively painless to compile it >as a library, link it with Swarm, and put a method in, say, the >observerSwarm that is called at whatever frequency you'd like. > >It's not in distributable form, yet. For instance, the xwd package >still tries to do a link when you make it and you have to create your >own libxwd.a with an ar rvs command. But, if anybody's got ants in >their pants, I can send it to them. > Well, don't know 'bout them ants -- but I'd like some, anyhow ... >I doubt this is the way we should really go. XWD files are pretty >large (105583 bytes for 80x80 default zoom heatbugs). Plus, we're >trying to move *away* from X if possible, not towards it. So, I've >started looking into HDF and AVS. Mind you, however, that this is >still not high priority. The hackish stuff is so simple that it can >never be considered high priority. And the long-term stuff ties in >with our general data visualization problem and our portability >problem, both of which are superceded by ||-Swarm, the Manual, and the >Alpha port. > >glen > Well, we humble peons out here gratefully take whatever crumbs come our way ... :-) Thanks, Glen Sven ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 14:27:04 1997 From: swarm-modelling@santafe.edu (Brian Ruth (CBNED/SMSB) ) Date: Thu, 3 Apr 1997 09:27:04 -0500 (EST) Subject: Modeling of Crowd Behavior In-Reply-To: <3.0.32.19970402095234.007255b4@tsa-po.lanl.gov> Message-ID: On Wed, 2 Apr 1997, Stephen C. Upton wrote: > As a 1.7 Swarm Lurker (between class 1 and 2 -- closer to 2 - I've finally > got Linux loaded on a new machine! I've previously had Swarm up and running > at another job, but, alas it's not part of my job description here -- > yet!), I appreciate the separation of the lists. As of now, I am more > interested in modeling aspects than specifics about Swarm code -- that will > come :P > > My question is: has, or is, anyone attempting to look at modeling crowd (of > humans) behavior, whether it be with Swarm or without? This has obvious > applications for the justice department and the military, for example. > They would like to disrupt a crowd using non-lethal technologies, but > certain actions may be more provocative than others. > > I would appreciate any references any one might have also. > > thanx > upton > > > > *** * **** * ***** ********* ** ****** ***** *** ***** ******** ********* > Stephen C. Upton > TSA-5, MS F602 > Los Alamos National Laboratory > Los Alamos, NM 87545 > 505-667-9435 FAX 505-665-2017 > upton@lanl.gov > ******* ********* *** ** *** ******** **** ****** ** ****** **** *** *** I'm currently a Swarm user wannabe (I'm waiting for the source+binary release for SGIs), but I have looked a bit into modeling crowd behavior. Dana Eckart of Radford University and I are currently developing a cellular automata model, using his Cellular simulation system (available at http://rucs2.sunlab.cs.runet.edu/~dana/ca/cellular.html), which demonstrates the emergence of panic within a unit of soldiers when exposed to one or more battlefield threats. The model determines whether a particular soldier will panic by assigning a probability of panic conditional on the number of wounded and/or panicking neighbors within the soldier's extended Moore neighborhood (21 x 21 cells on a 2D lattice), and then performs a random draw to determine the soldier's panic state (panicking/not panicking). Flocking behavior is also considered within the unit (for non-panicking soldiers only) , where a soldier's speed and direction of travel is determined by that of his neighbors. This type of model could easily be extended to a civilian crowd being fired upon by a sniper or being broken up by law enforcement officers, where the probability of panic associated with a civilian would be somewhat higher than that for a soldier under similar circumstances, and flocking (or anti-flocking) might be governed by a hierarchy based on parameters such as uniform (national guard, policeman), visible weapon (gun, club), and so on. Also, check out the examples link on Prof. Eckart's Cellular page (referenced above), where he presents an implementation of a flocking algorithm developed by Tamas Vicsek and colleagues at Eotvos University in Budapest. Hope this helps. Brian Ruth *===================================================================* | Brian G. Ruth ** Voice: Comm. (410)612-8687 | | U.S. Army Research Laboratory ** FAX: Comm. (410)671-2375 | | ATTN: AMSRL-SL-CM ** | | Bldg. E3331 ** | | Aberdeen Proving Ground, MD ** | | 21010-5423 ** email: bruth@arl.mil | |===================================================================| | Army Research Laboratory | | Survivability/Lethality Analysis Directorate | | Chemical-Biological, Nuclear & Environmental Effects Division | | Survivability Modeling & Simulation Branch | *===================================================================* ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 14:16:06 1997 From: swarm-modelling@santafe.edu (Randy Gimblett) Date: Thu, 03 Apr 1997 08:16:06 -0600 Subject: Modeling of Crowd Behavior Message-ID: <1.5.4.32.19970403141606.006f3bf4@ag.arizona.edu> Glen and SWARM Users In response to modeling human behavior, we have developed some software that merges agents with GIS for simulating recreation behavior in complex wilderness settings. We examined SWARM as the modeling framework, but to get a prototype up and running developed our own software in Visual Basic 4.0 running under windows 95. A version of the software referred to as RBSim - Recreation Behaviour Simulator and is available with enought documenation from the web: http://www.dlsr.com.au/software/rbsim The prototype is currently being written in SWARM by Bohdan Durnota in Australia. For more detailed information on the software you can contact with Bob Itami at the above web site or Randy Gimblett (gimblett@ag.arizona.edu). Randy Gimblett ___________________________________________________________________ Randy Gimblett Associate Professor School of Renewable Natural Resources The University of Arizona Tucson, Arizona 85721 USA Email: gimblett@nexus.srnr.arizona.edu World Wide Web: http://nexus.srnr.arizona.edu/~gimblett Phone: (520) 621-6360 Fax: (520) 621-8801 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 21:47:39 1997 From: swarm-modelling@santafe.edu (Sven N. Thommesen) Date: Thu, 03 Apr 1997 15:47:39 -0600 Subject: Catalog of agents Message-ID: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> At 09:27 AM 4/3/97 -0500, Brian Ruth wrote: > >I'm currently a Swarm user wannabe (I'm waiting for the source+binary >release for SGIs), but I have looked a bit into modeling crowd behavior. >Dana Eckart of Radford University and I are currently developing a >cellular automata model, using his Cellular simulation system (available >at > > http://rucs2.sunlab.cs.runet.edu/~dana/ca/cellular.html), > Brian, thanks for that reference to Prof. Eckart's ca stuff. Very interesting. Worth looking at! Which leads me to a suggestion for 'someone' to do: the essence of Swarm modelling is, of course, the behavioral methods of our agents. I'd like to see a web site collect a catalog of different behavioral methods people have used, described in pseudo-code and/or source code. (Similar to Prof. Eckart's collection of CA models.) This would allow others to test out the posted methods, and to critique them. Over time, we might get an idea of which algorithms are useful and which not for given applications. Any takers ? (If none, I may do it myself after random is done ...) Sven ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 3 23:07:10 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Thu, 3 Apr 1997 16:07:10 -0700 Subject: What *is* Time? (was Re: re getCurrentTime(): solved) In-Reply-To: <3.0.32.19970403153149.009a2100@spidle2.humsci.auburn.edu> References: <3.0.32.19970403153149.009a2100@spidle2.humsci.auburn.edu> Message-ID: <199704032307.QAA14420@grasshopper.santafe.edu> I moved this to the modelling list because it's really a very large fuzzy issue versus an implementation issue. I realize that the boundaries between the two lists are not well defined, yet; but, I'm trying to help that de-fuzzification. Sven N. Thommesen writes: > Rick has accurately described my initial expectations, > given that (time-wise, at least) my app is not more > complex than Heatbugs. Then I suppose that means we should re-evaluate the docs and what they explain. Of course, we'll need y'all to help us! > I can now hit my button before the sim starts, while > it is running, and while it is stopped, without any > rude surprises. Good! The point of the whole exercise is to get you to where you can do what you want, eh? > PS: Perhaps Glen could further muddy the waters: > is it, in Swarm, possible for different (sub-)Swarms to > have their own sense of time get out of sync with > each other? [And how does the answer depend on > whether we have serial or parallell Swarm? :-) ] > If they do, what is the relationship between such > sub-swarm clocks and any master Global Time clock? > > If the answer is yes, they *can* get out of sync, > then perhaps the macro is a dangerous thing that > shouldn't be there? I can always count on Sven to just cavalierly pop open that can o' worms. First, the short form (correct me if I get anything wrong, Roger): There is a "Relative Time" for each Swarm. This is probably the time that should be used inside subswarms. For instance, the modelSwarm has a "modelTime" that the agents inside that swarm should access. These relative times will almost certainly get out of sync in a || Swarm. But, that concern should be handled by the constraints that will be specified for the given schedules inside each subswarm. Examples of these constraints are the ability to have a a schedule in a swarm containing a sequence of ConcurrentGroups with "DefaultOrders" of Concurrent, Sequential, or Randomized. The long form: There has been alot of consideration given to how these models are integrated. And while time may seem like an inherent feature of what is happening in Swarm, it isn't intended as such. What we're calling the "logical model of concurrency" in Swarm is based on partially ordered sets. (See *) This basis can be considered completely independent of time. Just because all processes implicitly contain a concept of the "passage of time," doesn't mean they have to refer to or be based upon time. So, it's completely reasonable to think about all processes in terms of the order of events. These events can be ordered by any constraint imaginable.... heat index, risk dependency, relationships, etc. Now that i've gone quite off the deep end, I'll return to the question: "Is it possible for different (sub-)Swarms to have their own sense of time get out of sync with each other?" Not only is it possible, but, it's preferable, since most real systems have subsystems with clocks out of sync with each other. They have sync points, obviously. One perfect example is a human-computer interface. The computer's polls on, say, the keyboard device are much much faster than the human's strikes of the keys. In fact, the keystrokes of the human are some neurally generated sequence with some (hopefully) random noise, which is certainly not "sync'ed" with the computer's high frequency poll. But, the two processes are forcibly synced because any keystroke that may actually happen in between polling events of the computer waits for the next cycle of the computer. This is an extremely simple synchronization. Much much more complicated sycing goes on in virtually any system we might want to model. So, on to the next point: "perhaps the macro is a dangerous thing that shouldn't be there?" Well, scissors are also dangerous in the hands of running children? [grin] Seriously, yes I think the macro is very dangerous. But, I can't imagine telling users that there's no way to get the absolute time of the simulation. And, it is a useful tool, especially if you have ever worked inside the paradigm where your simulation contains a "truth" model and a "system" model. In this type of simulation, "error" plays a large role... And one can only have "error" if one has "truth." So, the macro should stay. And it is dangerous. But, it's a useful tool. We could either let new users run with the scissors and learn the error of there ways after a couple of untoward accidents.... Or we could prescribe restrictions on their behaviour for reasons they don't understand and expect them to grow into that understanding. Both work; but, you come off as limited and pretentious in the latter. glen p.s. I'm aware that this was not well-written and may, as Sven put it, "muddy the waters;" but, since I'm in manual writing phase, I felt it was a good chance to just toss out some things that need to be said without being too self-critical of the way they're said. Any responses will help mold the way this subject is handled in the manual. * Sets upon which relationships are defined that don't have relevance to all members of the set. E.g. I can have a set of objects, some of which are cows and some of which are cars. I can have a partial order relation, say, "produces-better-milk-than". Then it makes complete sense to say "cow1 produces-better-milk-than cow2"; but, it makes no sense to say "car1 produces-better-milk-than car2". But, it certainly also makes sense to say that "cow2 produces-better-milk-than car1". ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 18:08:40 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Mon, 7 Apr 1997 18:08:40 +0100 (BST) Subject: Modeling of Crowd Behavior In-Reply-To: Message-ID: On Thu, 3 Apr 1997, Brian Ruth (CBNED/SMSB) wrote: > I'm currently a Swarm user wannabe (I'm waiting for the source+binary > release for SGIs), but I have looked a bit into modeling crowd behavior. > Dana Eckart of Radford University and I are currently developing a > cellular automata model, using his Cellular simulation system (available > at > > http://rucs2.sunlab.cs.runet.edu/~dana/ca/cellular.html), > > which demonstrates the emergence of panic within a unit of soldiers when > exposed to one or more battlefield threats. The model determines whether > a particular soldier will panic by assigning a probability of panic > conditional on the number of wounded and/or panicking neighbors within the > soldier's extended Moore neighborhood (21 x 21 cells on a 2D lattice), > and then performs a random draw to determine the soldier's panic state > (panicking/not panicking). Flocking behavior is also considered within > the unit (for non-panicking soldiers only) , where a soldier's speed and > direction of travel is determined by that of his neighbors. Why shouldn't panicking soldiers flock as well? Don't people who flock panic and people who panic flock? Just wondering, Jan. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 19:15:23 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Mon, 7 Apr 1997 12:15:23 -0600 Subject: Catalog of agents In-Reply-To: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> References: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> Message-ID: <199704071815.MAA16011@grasshopper.santafe.edu> Sven N. Thommesen writes: > Which leads me to a suggestion for 'someone' to do: > > the essence of Swarm modelling is, of course, the behavioral > methods of our agents. I'd like to see a web site collect > a catalog of different behavioral methods people have used, > described in pseudo-code and/or source code. (Similar to > Prof. Eckart's collection of CA models.) This would allow > others to test out the posted methods, and to critique them. > Over time, we might get an idea of which algorithms are > useful and which not for given applications. > > Any takers ? This is an excellent idea. Of course, more than just the agent methods should be described, I would guess. But, it might be reasonable to compress the "essence" of a model with descriptive pseudo-code for the agent types and their respective methods, and the environment design and it's respective methods. The one element left is the scheduling. Developing compact descriptions of these three elements might even help us with the Schedule-language specification when and if we start working seriously on that. A language needs well-specified data types (agents and environments) as well as well-specified operator types. As far as takers for the task.... Brad? glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 22:23:53 1997 From: swarm-modelling@santafe.edu (Jae Chan Oh) Date: Mon, 7 Apr 1997 17:23:53 -0400 (EDT) Subject: Which Machine for swarm: Pentium Pro, or SUN Ultra 140? In-Reply-To: <199704071815.MAA16011@grasshopper.santafe.edu> from "glen e. p. ropella" at Apr 7, 97 12:15:23 pm Message-ID: <199704072123.RAA00927@homer.cs.pitt.edu> Hi, We have choice of buying either Pentium Pro or Sun Ultra 140 for swarm application developement. Which machine do you think we should go with? (In terms of computing power, swarm compatibility/friendliness, etc.) Does anyone have any preference between the two? The Specs are: Pentium-Pro: 200 MHz, 64 M Main Memory, 3.8 G HD, SoundBlaster, CD Player, 17" Monitor, Cost around $3,000. (we can add extra 64Meg) Sun Ultra Sparc 140: 143 MHz, 128 Meg, 2.1 G HD, CD player, 17" monitor, Cost around $5000 We will use Linux for Pentum-Pro and Solaris for SUN, of course. Thanks, -Jae ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 23:00:35 1997 From: swarm-modelling@santafe.edu (Brian Haugh) Date: Mon, 7 Apr 1997 18:00:35 -0400 Subject: Which Machine for swarm: Pentium Pro, or SUN Ultra 140? In-Reply-To: Jae Chan Oh "Which Machine for swarm: Pentium Pro, or SUN Ultra 140?" (Apr 7, 5:23pm) References: <199704072123.RAA00927@homer.cs.pitt.edu> Message-ID: <9704071800.ZM1551@fasolt.csed.ida.org> We have found that our Swarm application runs faster on a Pentium Pro than on a Sun Ultra 1/140. Unfortunately, Swarm can't take any advantage of the Ultra's 64 bit architecture since the gcc Objective-C compiler does not compile for it. Of course you can compile with gcc on an UltraSparc with gcc, but the code is no different from a regular Sparc. Thus, the main difference in performance is just due to clock speed - since the Pro's clock is faster than an Ultra 1 at 140 Mhz, then the code run's faster. Our app runs at about the same speed on a Ultra Enterprise 2 200Mhz server as the Pentium Pro, but the Ultra 2 is a bit more expensive. So, if Swarm performance is the only consideration, I would recommend the Pentium Pro under Linux. If you are going to use the system for other applications, though, the Ultra might have some advantages. The Ultra is especially fast on multi-media, i.e., video compression/ decompression, from what I've read, due to special parallel execution for MPEG instructions. Brian -- Brian A. Haugh, Ph.D. Institute for Defense Analyses Computer & Software Engineering Division 1801 North Beauregard Street Alexandria, Virginia 22311-1772 phone: (703) 845-6678 fax: (703) 845-6788 email: bhaugh@ida.org ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 7 23:21:56 1997 From: swarm-modelling@santafe.edu (Randall Gray) Date: Tue, 8 Apr 1997 08:21:56 +1000 (EST) Subject: Which Machine for swarm: Pentium Pro, or SUN Ultra 140? In-Reply-To: <199704072123.RAA00927@homer.cs.pitt.edu> (message from Jae Chan Oh on Mon, 7 Apr 1997 17:23:53 -0400 (EDT)) Message-ID: <199704072221.IAA01474@njal.ml.csiro.au> > From: Jae Chan Oh > Date: Mon, 7 Apr 1997 17:23:53 -0400 (EDT) > We have choice of buying either Pentium Pro or Sun Ultra 140 for swarm > Does anyone have any preference between the two? In Australia (at least) the answer is fairly easy: go with the intel based machine. We recently purchased a dual processor machine with roughly 2/3 the grunt of our "central number crunching" machines at work for about 1/5 the price tag. Buy the extra memory for the PC -- it is even nicer. Linux can make huge disk buffers when there is no other demand for the memory and this is a *very* nice use which speeds things up quite a lot. > We will use Linux for Pentum-Pro and Solaris for SUN, of course. Linux is available on SPARCS: there is a port to Ultras which is running (though still in development). Apparently it flys. Dunno what the exchange rate is today, so I guess this is worth what ever I get for it ;-) -- Randall ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 02:41:51 1997 From: swarm-modelling@santafe.edu (Todd Allen) Date: Mon, 07 Apr 1997 21:41:51 -0400 Subject: Which Machine for swarm: Pentium Pro, or SUN Ultra 140? References: <199704072123.RAA00927@homer.cs.pitt.edu> Message-ID: <3349A25F.4E79@jhunix.hcf.jhu.edu> Jae, For the money the Pentium is the clear choice. I run a dual Pro-200 machine under Linux 2.0 at home and a Sun Sparc Ultra 1 at school and have found that the performance is nearly identical when running Swarm. With the huge drop in pentium pro 200 prices this week due to the release of the 233 and 266 processors, I'd recommend getting a dual pentium pro box. Swarm can't directly take advantage of the second processor, but performance is still enhanced due to the off-loading of much of the OS and X windows overhead onto the second processor. If you're going to go with a single processor you may want to look into one of the new faster processors. The Pentium II 233 is selling for about what the Pro-200 was a month ago, and the 266 is about a hundred more. Regards, Todd -- Todd Allen The Johns Hopkins University Department of Economics tallen@jhu.edu 410-516-7571 (office) "Rational expectations is rigorous deduction based upon faulty assumptions." - Brian Arthur ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 16:50:41 1997 From: swarm-modelling@santafe.edu (Brian Ruth (CBNED/SMSB) ) Date: Tue, 8 Apr 1997 11:50:41 -0400 (EDT) Subject: Modeling of Crowd Behavior In-Reply-To: Message-ID: On Mon, 7 Apr 1997, Jan Kreft wrote: > Why shouldn't panicking soldiers flock as well? Don't people who flock > panic and people who panic flock? > > Just wondering, > > Jan. Naturally they do (in a manner of speaking ;-)). The flocking of panicking soldiers (or people in general) should, however, probably reflect less centralization and more random variations relative to the flocking of non-panicking soldiers (people). Brian ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 17:43:13 1997 From: swarm-modelling@santafe.edu (John A. Lopez) Date: Tue, 08 Apr 1997 09:43:13 -0700 Subject: Catalog of agents References: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> <199704071815.MAA16011@grasshopper.santafe.edu> Message-ID: <334A759F.2A67@pop.calweb.com> glen e. p. ropella wrote: > > Sven N. Thommesen writes: > > Which leads me to a suggestion for 'someone' to do: > > > > the essence of Swarm modelling is, of course, the behavioral > > methods of our agents. I'd like to see a web site collect > > a catalog of different behavioral methods people have used, > > described in pseudo-code and/or source code. (Similar to > > Prof. Eckart's collection of CA models.) This would allow > > others to test out the posted methods, and to critique them. > > Over time, we might get an idea of which algorithms are > > useful and which not for given applications. > > > > Any takers ? > > This is an excellent idea. Of course, more than just the > agent methods should be described, I would guess. But, > it might be reasonable to compress the "essence" of a > model with descriptive pseudo-code for the agent types > and their respective methods, and the environment design > and it's respective methods. The one element left is the > scheduling. > > Developing compact descriptions of these three elements > might even help us with the Schedule-language specification > when and if we start working seriously on that. A language > needs well-specified data types (agents and environments) > as well as well-specified operator types. > > As far as takers for the task.... Brad? > > glen Sven and Glen, I think this is more than an excellent idea, since this list would necessarily evolve to discipline-specific lists. Such posted methods could be improved upon or varied in ways which could form the basis of a dialog (parallelog?) within a research discipline (mine is anthropology/archaeology). I'd be more than happy to post my kinship, mimicry, and the other behavioral methods I've developed or am working on. In light of the recent discussion regarding future funding and non-profit organization structure, we could take a lesson from Swarm. This idea of a methods listing could be a source for SFI income: users could pay some individual or institutional fee for specific downloads of methods (available free to contributers within a class of course). Perhaps the SFI folks could continue to make its server available and act as a GA and cull those discussions which have little or no interest. John ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 17:47:56 1997 From: swarm-modelling@santafe.edu (John A. Lopez) Date: Tue, 08 Apr 1997 09:47:56 -0700 Subject: Modeling of Crowd Behavior References: Message-ID: <334A76BA.5E72@pop.calweb.com> Brian Ruth (CBNED/SMSB) wrote: > > On Mon, 7 Apr 1997, Jan Kreft wrote: > > > Why shouldn't panicking soldiers flock as well? Don't people who flock > > panic and people who panic flock? > > > > Just wondering, > > > > Jan. > > Naturally they do (in a manner of speaking ;-)). The flocking of > panicking soldiers (or people in general) should, however, probably > reflect less centralization and more random variations relative to > the flocking of non-panicking soldiers (people). > > Brian Jan & Brian, I think paniced people give up their individual attributes for the global elements and consequences governing the crowd. Two cents worth. John ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 8 19:19:46 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 8 Apr 1997 12:19:46 -0600 Subject: Catalog of agents In-Reply-To: <334A759F.2A67@pop.calweb.com> References: <3.0.32.19970403154734.0099ceb0@spidle2.humsci.auburn.edu> <199704071815.MAA16011@grasshopper.santafe.edu> <334A759F.2A67@pop.calweb.com> Message-ID: <199704081819.MAA16505@grasshopper.santafe.edu> John A. Lopez writes: > I think this is more than an excellent idea, since this list would > necessarily evolve to discipline-specific lists. Such posted methods > could be improved upon or varied in ways which could form the basis of a > dialog (parallelog?) within a research discipline (mine is > anthropology/archaeology). I'd be more than happy to post my kinship, > mimicry, and the other behavioral methods I've developed or am working > on. > > In light of the recent discussion regarding future funding and > non-profit organization structure, we could take a lesson from Swarm. > This idea of a methods listing could be a source for SFI income: users > could pay some individual or institutional fee for specific downloads of > methods (available free to contributers within a class of course). > Perhaps the SFI folks could continue to make its server available and > act as a GA and cull those discussions which have little or no > interest. Hmmmm. This would complement the idea that someone had during SwarmFest of providing an archiving service of source code. The suggestion was that, since part of Swarm's purpose is to help with the documentation and repeatability aspects of doing research via simulation, it would help to have some repository in which simulants could deposit their source code for posterity. Along these lines, it would be reasonable to assume that the swarm.org organize and manage this repository as well as some kind of Agent-Reuse library/catalog, as mentioned above. Charging for this service wouldn't be a bad idea, at all. There are three ways that could be done: 1) charge the researcher for archiving his stuff (this wouldn't be effective for the agent catalog but might be ok for registering simulations), 2) charging for access to the database, or 3) grant style funding, where money is sought from institutions who have an interest in making simulation more mature. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 11 19:14:35 1997 From: swarm-modelling@santafe.edu (Phil Knight) Date: Fri, 11 Apr 1997 19:14:35 +0100 Subject: modelling CAs in swarm? Message-ID: I'm a delurking swarm user who rates somewhere just over 3 on Glen's user scale (3 - those who have successfully installed Swarm and played with the provided demos, but haven't made their own models, 4 - those who are actively using Swarm). I'm interested in using swarm to explore cellular automata, specifically to model physical phenomena (i.e. digital physics or Fredkin's universe as a CA). In order to get started I set up a sim of Conway's Life. This is literally a two minute job to wrap a model swarm around the pre- supplied life class. Since I'm likely to want to utilise relatively large lattice spaces (ultimately in 3d), and the life class is implemented with a bog standard 2d array, I set about cobbling together a sparse matrix version of life and I've now succeeded in getting this up and running. The initial world size is specified and seeded with the desired probability, the initial world area is constantly displayed on a raster, but the lattice iteslf extends to 32k x 32k (it should be possible to increase this even further). In actual fact the app will also run more than just the life rule, allowing any 2d, binary, totalistic, Moore neighborhood CA to be simulated by specifying the appropriate birth/survival rule before running. For anyone who's interested, both apps are available at: http://www.pknight.demon.co.uk/ I should emphasise though that they are very hacked together and not intended to be the best or most efficient means of construction. At this stage I'm just trying to explore methods of modelling CAs on large lattices rather than providing serious models. Any constructive criticism is more than welcome. The real purpose of this post is to ask firstly if anyone else here is using swarm for CAs, especially on large lattices and if so, the approaches being used in the models. Secondly, and perhaps more importantly, one area that I'm really going to want to work on is visualisation tools (esp. 3d). For example, the sims above only display the initial lattice size, but the "active" lattice in the sparse version quickly becomes much larger, so there is a requirement to be able to zoom in/out of different areas of the lattice. I haven't yet thought too much about 3d visualisation tools as yet, but hope this post might stimulate some discussion :-) -- Phil Knight ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sun Apr 13 17:12:11 1997 From: swarm-modelling@santafe.edu (Brad Leydorf) Date: Sun, 13 Apr 1997 12:12:11 -0400 Subject: Behavior modelling?? References: Message-ID: <335105DB.240E@one.net> Is anyone out there in Swarmland into behavior modelling? (of people). If so, I'd love to hear any suggestions or cautions you might have... Thanks, Brad bleydorf@one.net ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sun Apr 13 22:58:42 1997 From: swarm-modelling@santafe.edu (Pietro Terna) Date: Sun, 13 Apr 1997 23:58:42 +0200 Subject: modelling CAs in swarm? Message-ID: <1.5.4.32.19970413215842.0069f7bc@alpcom.it> Hi Phil, I have tested your helpful examples, many thanks. Only conwaylife2d has been creating some problem in my Linux box. I my directory (SWARMHOME)/include/graph/ I just had a ConwayLife2d.h file, which resulted imported in your application via #import so generating errors such as 'duplicate interface declarations' and, finally, no linking. My solution hacked solution has been that of modiying all instances of ConwayLife2d in ConwayLifeBis2d in your package and so ... all is running. Yours, Pietro At 19.14 11/04/97 +0100, you wrote: >I'm a delurking swarm user who rates somewhere just over 3 on Glen's >user scale (3 - those who have successfully installed Swarm and played >with the provided demos, but haven't made their own models, 4 - those >who are actively using Swarm). > >I'm interested in using swarm to explore cellular automata, specifically >to model physical phenomena (i.e. digital physics or Fredkin's universe >as a CA). In order to get started I set up a sim of Conway's Life. This >is literally a two minute job to wrap a model swarm around the pre- >supplied life class. > >Since I'm likely to want to utilise relatively large lattice spaces >(ultimately in 3d), and the life class is implemented with a bog >standard 2d array, I set about cobbling together a sparse matrix version >of life and I've now succeeded in getting this up and running. The >initial world size is specified and seeded with the desired probability, >the initial world area is constantly displayed on a raster, but the >lattice iteslf extends to 32k x 32k (it should be possible to increase >this even further). In actual fact the app will also run more than just >the life rule, allowing any 2d, binary, totalistic, Moore neighborhood >CA to be simulated by specifying the appropriate birth/survival rule >before running. > >For anyone who's interested, both apps are available at: >http://www.pknight.demon.co.uk/ >I should emphasise though that they are very hacked together and not >intended to be the best or most efficient means of construction. At this >stage I'm just trying to explore methods of modelling CAs on large >lattices rather than providing serious models. Any constructive >criticism is more than welcome. > >The real purpose of this post is to ask firstly if anyone else here is >using swarm for CAs, especially on large lattices and if so, the >approaches being used in the models. > >Secondly, and perhaps more importantly, one area that I'm really going >to want to work on is visualisation tools (esp. 3d). For example, the >sims above only display the initial lattice size, but the "active" >lattice in the sparse version quickly becomes much larger, so there is a >requirement to be able to zoom in/out of different areas of the lattice. >I haven't yet thought too much about 3d visualisation tools as yet, but >hope this post might stimulate some discussion :-) > >-- >Phil Knight > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 12 20:45:46 1997 From: swarm-modelling@santafe.edu (Brad Leydorf) Date: Sat, 12 Apr 1997 15:45:46 -0400 Subject: anyone behavior modelling? References: Message-ID: <334FE66A.1729@one.net> Is anyone out there in Swarmland into behavior modelling? (of people). If so, I'd love to hear any suggestions or cautions you might have... Thanks, Brad bleydorf@one.net ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 16 20:27:09 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Wed, 16 Apr 97 13:27:09 MDT Subject: Simulating Individual Behavior References: <199704161351.HAA00352@seamus.dischordia.com> Message-ID: <9704161927.AA12178@sfi.santafe.edu> I would be interested in knowing how the SWARM project might impact social simulation. We can start this off as a research discussion, and then perhaps, this will help to stop the flood of "unsubscriptions." The idea of simulating a system by going down to the atomic level is something that most disciplines have dealt with in one form or another. Consider, for example, ecosystem modeling. In ecosystem modeling, there is some controversy about individual-based modeling (IBM) and yet this modeling approach holds promise in predicting the evolution of a ecosystem. A key problem with IBM is not that it is necessarily computationally prohibitive, but that not enough data are available to calibrate the model, or individuals participating within the model. If this is a problem for modeling alligators or wood storks, I would venture that the problem for human systems would be manifold. What kind of model is used to model the human. A logical choice of model type for human objects in the simulation is one based on AI techniques: rule-based or operator based systems. On the other hand, what level of detail is required to model the human element? Perhaps, it need not be detailed. A good example of high level modeling of humans can be found in emergency planning simulation, which is a subject of interest in the Society for Computer Simulation since they have sponsored many conferences in this area. If you have a hazard or peril within an enclosed structure, how should the humans react? Modeling the human in this instance could be much easier than modeling for a less-streneous social goal. Thoughts? -paul -- Paul A. Fishwick E-Mail: fishwick@cise.ufl.edu Dept. of Computer & Info Phone & FAX: (352) 392-1414 Science and Engineering WWW: http://www.cise.ufl.edu/~fishwick University of Florida (PGP Key available at above WWW address) P. O. Box 116120 332 Bldg. CSE, Gainesville, FL 32611-6120 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 11:50:51 1997 From: swarm-modelling@santafe.edu (Philippe LAVAL) Date: Thu, 17 Apr 1997 09:50:51 -0100 Subject: Simulating Individual Behavior Message-ID: <3.0.32.19970417095049.006ebc8c@poseidon.obs-vlfr.fr> Paul Fishwick said : >A key problem with IBM is not that it is necessarily >computationally prohibitive, but that not enough data are >available to calibrate the model... Perhaps you are talking about ecosystems from your simulation-oriented point of view, as the word 'calibrate' indicates. I think that a complex system like an ecosystem cannot be modeled with some large FSM or even with a set of stochastic equations (except in the short-term): in an ecosystem there is probably no unique latent 'model' to be discovered, against which we could compare some data and make adjustments. Instead there probably exists many possible modes, with the system unpredictably switching from time to time to one or another. The first thing to do could be to try to recognize these modes. Because switching is unpredictable (ecosystems seem to exhibit self-organized critical modes), it makes no sense to try to 'calibrate' the whole system, in the same manner simulation engineers calibrate a model of a factory or a flexible workshop. -------------------------------------------------------------------- Philippe Laval Station zoologique B.P. 28 - 06234 Villefranche-sur-Mer CEDEX (France) laval@ccrv.obs-vlfr.fr ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 11:46:50 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: Thu, 17 Apr 1997 04:46:50 -0600 Subject: Simulating Individual Behavior Message-ID: <199704171046.EAA21778@santafe.santafe.edu> Paul Fishwick wrote: >A key problem with IBM is not that it is necessarily >computationally prohibitive, but that not enough data are >available to calibrate the model... Paul - why is data calibration more of a problem for IBM models than for other modelling technologies? What sorts of problem domains are you particularly concerned about? I can allow as how it would be difficult to calibrate an IBM model of a forest precisely so that each tree-agent is parameterized via data taken from its respective real-tree in the forest. Yet, all modelling technologies make do with some degree of approximation. A reservoir-flow model of tree-species interaction in a forest would simply treat all of the trees of each species as "one big tree" of that species with respect to some data (such as concentration, nutrient uptake, waste-production, and etc.) while ignoring other data (such as spatial distribution, variety within the species, and etc.) This will be justified for certain questions about forest dynamics, but not for others, and might make more sense for some problem domains than for others. Thus, there is data and there is data. All modeling technologies must pick and choose among the data, and one always has to focus on some reasonable subset of data. You seem to be suggesting that this is fundamentally more of a problem for IBM models than for other modelling technologies - can you elaborate? ...and, please!, it almost *hurts* to use the acronym IBM! could we use "ABM" for Agent-Based Models? I think it fits better anyway, as an agent in this class of models is not always an "individual" in the common sense of that term.... I know the term has some historical precedent for models in this class, but the acronym IBM induces a certain amount of, shall we say, cognitive dissonance, no? (not that the acronym "ABM" itself is inviolate with respect to prior cognitive content....but, still!.....) Chris Langton ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 18:43:53 1997 From: swarm-modelling@santafe.edu (Mike Brown) Date: 17 Apr 1997 09:43:53 -0800 Subject: Simulating Individual B Message-ID: I think Paul's question raises two important points - though they may be more about the organization of science than about the inherent difficulties of SWARM or any other ABM (Thanks, Chris. I like that acronym better too.) Ecologists are concerned with aggregations of individuals; they have not had to know as much about individual species as, for example, zoologists. Similarly, microeconomists focus on the behavior of individual firms and macroeconomists on the aggregate behavior of the economy, etc. ABM create two problems. First, the ecologist and macroeconomist have to "ratchet down" and study the behavior of individual entities. Moreover, they have to know enough about individual behavior to determine which specific behaviors might be relevant to the question under investigation. While this is not an insurmountable problem, it does demand a new focus for researchers. Second, there are some disciplines where data on the behavior of individual "agents" simply has not been studied. To take a bad example, look at economics. Macro studies the behavior of the aggregate, and micro the study of the firm -- but who has been looking at the behavior of the consumer? We have been able to make a "rational actor" assumption for so long that we have not bothered to collect data about the real-life behavior of induividual consumers. For these reasons, I think Paul is very right -- modeling and validating the behavior of individual entities can be very tricky. Mike ------------------------------ Date: 4/17/97 4:47 AM To: Brown, Mike From: swarm-modelling@santafe.edu Paul Fishwick wrote: >A key problem with IBM is not that it is necessarily >computationally prohibitive, but that not enough data are >available to calibrate the model... Paul - why is data calibration more of a problem for IBM models than for other modelling technologies? What sorts of problem domains are you particularly concerned about? I can allow as how it would be difficult to calibrate an IBM model of a forest precisely so that each tree-agent is parameterized via data taken from its respective real-tree in the forest. Yet, all modelling technologies make do with some degree of approximation. A reservoir-flow model of tree-species interaction in a forest would simply treat all of the trees of each species as "one big tree" of that species with respect to some data (such as concentration, nutrient uptake, waste-production, and etc.) while ignoring other data (such as spatial distribution, variety within the species, and etc.) This will be justified for certain questions about forest dynamics, but not for others, and might make more sense for some problem domains than for others. Thus, there is data and there is data. All modeling technologies must pick and choose among the data, and one always has to focus on some reasonable subset of data. You seem to be suggesting that this is fundamentally more of a problem for IBM models than for other modelling technologies - can you elaborate? ...and, please!, it almost *hurts* to use the acronym IBM! could we use "ABM" for Agent-Based Models? I think it fits better anyway, as an agent in this class of models is not always an "individual" in the common sense of that term.... I know the term has some historical precedent for models in this class, but the acronym IBM induces a certain amount of, shall we say, cognitive dissonance, no? (not that the acronym "ABM" itself is inviolate with respect to prior cognitive content....but, still!.....) Chris Langton ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== ------------------ RFC822 Header Follows ------------------ Received: by cpqm.saic.com with ADMIN;17 Apr 1997 04:44:33 -0800 Return-Path: Received: from sfi.santafe.edu by cpmx.mail.saic.com; Thu, 17 Apr 97 04:45:16 -0700 Received: by sfi.santafe.edu (4.1/SMI-4.1) id AA21897; Thu, 17 Apr 97 04:46:49 MDT Date: Thu, 17 Apr 1997 04:46:50 -0600 From: cgl@santafe.edu Message-Id: <199704171046.EAA21778@santafe.santafe.edu> To: swarm-modelling@santafe.edu Subject: Re: Simulating Individual Behavior Sender: owner-swarm-modelling@santafe.edu Precedence: bulk Reply-To: swarm-modelling@santafe.edu ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 16:12:57 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: Thu, 17 Apr 1997 09:12:57 -0600 Subject: Simulating Individual B Message-ID: <199704171512.JAA22183@santafe.santafe.edu> Mike Brown writes: > I think Paul's question raises two important points - though they may be more > about the organization of science than about the inherent difficulties of > SWARM or any other ABM (Thanks, Chris. I like that acronym better too.) > Ecologists are concerned with aggregations of individuals; they have not had > to know as much about individual species as, for example, zoologists. > Similarly, microeconomists focus on the behavior of individual firms and > macroeconomists on the aggregate behavior of the economy, etc. Excellent point, and I do think that it is more about the organization of science than inherent problems with models... In many scientific fields (but ecology and economics will serve nicely as examples), there are those (usually theorists) who study the macro behavior, and those (usually field workers) who study the micro behavior - and the two communities rarely interact with each other. The models built by each community often simply make enormously simplified assumptions about the other's results as the starting point and/or boundary conditions for their own models. One of the main reasons I am interested in these ABM models is that they provide an opportunity to close the loop and look at the way in which the behavior treated by the theorists at the macro level is grounded in the micro-level behaviors treated by the field workers - it all has to come together here - the macro behavior should emerge out of the micro behavior, as it does in the natural system. Thus these ABM models tend to be "meso"-scale, bigger than the small slices of systems typically treated by the field-workers, but necessarily smaller than the full scale macro-system: just big enough to study the way in which the macro emerges out of the micro, but not so big that it rivals the real system in all its unwieldy scale... It is precisely because these models can serve as a common venue for both the macro and micro communities in a discipline that they are of so much interest (and, I think, importance). > First, the ecologist and macroeconomist have to > "ratchet down" and study the behavior of individual entities. Moreover, they > have to know enough about individual behavior to determine which specific > behaviors might be relevant to the question under investigation. While this is > not an insurmountable problem, it does demand a new focus for researchers. > Second, there are some disciplines where data on the behavior of individual > "agents" simply has not been studied. To take a bad example, look at > economics. Macro studies the behavior of the aggregate, and micro the study of > the firm -- but who has been looking at the behavior of the consumer? We have > been able to make a "rational actor" assumption for so long that we have not > bothered to collect data about the real-life behavior of induividual > consumers. > For these reasons, I think Paul is very right -- modeling and validating the > behavior of individual entities can be very tricky. Right again - however I consider this as a feature rather than a bug! I think it gets at the essence of what "modelling" is all about... The term "model" means a lot of different things to different people. However, I find that the most useful characterization of a model is as an artifact that you construct that helps you think about the system under study - this can range from obviously over-simplified models that simply help you understand what data you still lack, and what parameters might be the interesting ones, to a full blown *theory* that claims to explain exactly how the real system works. The point is that in building these ABM models, often the first thing we learn from them is that we have to go back to the real system with new questions....but this is good! It is very useful to have a tool that provides you with a new way to look at your problem and which generates new questions about it. Mike (and Paul) may be right about the general lack of data, although I'm willing to bet that some of this perception is due to the lack of communication between the macro and micro scale research communities. But, I also think that this is because we haven't been thinking about some of these natural systems in the right way, I think ABM models provide the proper venue to help us think about the models in the right way, and if the result of that help in thinking is to force us to collect new data about the system, then that's a useful and necessary lesson.... So - again, I don't think of this as a problem, but a feature of these models......and I don't think of it as unique to ABM models, per se - I think new tools often offer us new "opportunities" to collect data... Chris Langton ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 16:51:50 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Thu, 17 Apr 1997 16:51:50 +0100 (BST) Subject: Simulating Individual Behavior In-Reply-To: Message-ID: OK, I agree with Paul and Mike that it can be very tricky to determine the parameters for individuals. Bacteria as individuals, e. g., are simply so small that only the latest high-tech methods could give estimates for the parameters you would have to use in a model (and that costs a lot of timemoney). Almost always microbiologists measure the population average and disregard the heterogeneity of the population. But I can see another way to solve the problem. If you start with a conceptual model with all the parameters and mechanics you need and only a rough idea of actual values (from population average measurements) you can run the sim and compare the output with the real system to be modeled. Then you optimize the parameters to achieve the desired output. Perhaps call it back-calibration. (Is there an official term for it?) For that purpose, you need a parameter manager (see previous discussion with "parameter" in the subject line) to allow a search through param space in an evolutionary fashion. As far as I can tell, Gecko's param manager is (the only one?) suited for that task. BTW, does IbM hurt also? If so, sorry for that ;-). Point being is that that's the term used in the literature and if I want to search for new literature, I rely on this keyword. Therefore I would want to use this keyword myself sometime. Cheers, Jan. Mike Brown wrote: > I think Paul's question raises two important points - though they may be more > about the organization of science than about the inherent difficulties of > SWARM or any other ABM (Thanks, Chris. I like that acronym better too.) > Ecologists are concerned with aggregations of individuals; they have not had > to know as much about individual species as, for example, zoologists. > Similarly, microeconomists focus on the behavior of individual firms and > macroeconomists on the aggregate behavior of the economy, etc. > > ABM create two problems. First, the ecologist and macroeconomist have to > "ratchet down" and study the behavior of individual entities. Moreover, they > have to know enough about individual behavior to determine which specific > behaviors might be relevant to the question under investigation. While this is > not an insurmountable problem, it does demand a new focus for researchers. > > Second, there are some disciplines where data on the behavior of individual > "agents" simply has not been studied. To take a bad example, look at > economics. Macro studies the behavior of the aggregate, and micro the study of > the firm -- but who has been looking at the behavior of the consumer? We have > been able to make a "rational actor" assumption for so long that we have not > bothered to collect data about the real-life behavior of induividual > consumers. > > For these reasons, I think Paul is very right -- modeling and validating the > behavior of individual entities can be very tricky. > > Mike > > Paul Fishwick wrote: > > >A key problem with IBM is not that it is necessarily > >computationally prohibitive, but that not enough data are > >available to calibrate the model... > > Paul - why is data calibration more of a problem for IBM > models than for other modelling technologies? What sorts > of problem domains are you particularly concerned about? > > I can allow as how it would be difficult to calibrate > an IBM model of a forest precisely so that each tree-agent > is parameterized via data taken from its respective > real-tree in the forest. Yet, all modelling technologies > make do with some degree of approximation. A reservoir-flow > model of tree-species interaction in a forest would simply > treat all of the trees of each species as "one big tree" > of that species with respect to some data (such as concentration, > nutrient uptake, waste-production, and etc.) while ignoring > other data (such as spatial distribution, variety within the > species, and etc.) This will be justified for certain questions > about forest dynamics, but not for others, and might make more > sense for some problem domains than for others. > > Thus, there is data and there is data. All modeling technologies > must pick and choose among the data, and one always has to > focus on some reasonable subset of data. You seem to > be suggesting that this is fundamentally more of a problem > for IBM models than for other modelling technologies > - can you elaborate? > > ...and, please!, it almost *hurts* to use the acronym IBM! > could we use "ABM" for Agent-Based Models? I think it fits > better anyway, as an agent in this class of models is not > always an "individual" in the common sense of that term.... > I know the term has some historical precedent for models > in this class, but the acronym IBM induces a certain amount > of, shall we say, cognitive dissonance, no? (not that the > acronym "ABM" itself is inviolate with respect to prior > cognitive content....but, still!.....) > > > Chris Langton ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 16:44:57 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Thu, 17 Apr 97 11:44:57 EDT Subject: Simulating Individual Behavior In-Reply-To: <199704171512.JAA22183@santafe.santafe.edu>; from "cgl@santafe.edu" at Apr 17, 97 9:12 am Message-ID: <9704171554.AA25869@sfi.santafe.edu> Paul and Phillippe and Mike and Chris, I've been enjoying this thread a great deal. From my ecology model, I have been somewhat frustrated by the lack of the kind of data I wanted, but in practice, taking the tiniest bit of relevant detail, instead of all possible detail, buys a lot.... You really get to see what the ramifications are of that seemingly little detail. And there are a lot of those little details "we all know", but haven't necessarily realized how important they are. As a trivial example, when deciding a herbivore's freedom of movement per round, I found that ~215 degrees (similar to a vision path) worked really well, and the "obvious" 360 degrees (random direction) was devastating to the plants. Gecko tells me these crazy things a lot, and I'm forced to think it through and hunt data. Another one I just ran into (predictably, while trying to do something else that Gecko balked at ;), was that my crazy program told me that terrestrial producers should be more productive than aquatic. "Dim-witted bug-generator!" I muttered.... However, to settle the argument, I got the data. Though I'd always "known" that oceanic plankton were "of course" the dominant producers on earth, turns out I was wrong. The "dim-witted bug-generator" was right. If it's right for the right reasons (always a danger, being right for the wrong reason ;), I accidentally found something out that really matters to ecological policy! For it suggests what kind of terrestrial ecosystems (rather safer and easier to manipulate than open ocean :) could conceivably offset CO2 emissions. Assuming I could prove it, of course.... In other words (Chris's :): > So - again, I don't think of this as a problem, but a feature > of these models......and I don't think of it as unique to > ABM models, per se - I think new tools often offer us new > "opportunities" to collect data... Ditto. Cheers, Ginger ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 17:01:03 1997 From: swarm-modelling@santafe.edu (Kevin Crowston) Date: Thu, 17 Apr 1997 12:01:03 -0400 Subject: Simulating Individual B Message-ID: > who has been looking at the behavior of the consumer? We have >been able to make a "rational actor" assumption for so long that we have not >bothered to collect data about the real-life behavior of induividual >consumers. Actually, consumer behaviour is an important research area in the field of marketing (which seems to be a mixture of applied microeconomic and psychology, at least to an outsider). Kevin ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 17:13:28 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: Thu, 17 Apr 1997 10:13:28 -0600 Subject: Simulating Individual B Message-ID: <199704171613.KAA22412@santafe.santafe.edu> > Actually, consumer behaviour is an important research area in the field of > marketing (which seems to be a mixture of applied microeconomic and > psychology, at least to an outsider). Right - I think that there is a tremendous amount of data on consumers that have been taken in market "demographic" studies, and of course there are Nielson ratings, the Billboard top-10 and etc... These data exist in corporations, and they may or may not be available to scientists for research use. One group working with such data in the context of ABMs is the Emergent Solutions group at Coopers and Lybrand, who have been building models of the propagation of fads and fashions throughout (initially music) markets - they used a demographic data base of 150,000 americans, provided by (I'm assuming) the entertainment industry.... Just one example of how there might, in fact, be data out there that has been collected by other groups for other purposes ... Chris ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 17:49:20 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Thu, 17 Apr 97 12:49:20 EDT Subject: Simulating Individual Behavior In-Reply-To: <199704171613.KAA22412@santafe.santafe.edu>; from "cgl@santafe.edu" at Apr 17, 97 10:13 am Message-ID: <9704171658.AA28614@sfi.santafe.edu> P.S., the analogue of consumer research <-> economics is possibly agriculture & fisheries & lumber <-> ecology. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 18:20:34 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Thu, 17 Apr 1997 13:20:34 -0400 (EDT) Subject: IBMs.. Message-ID: <199704171720.NAA22086@tide.cise.ufl.edu> Chris and all: Let me address your points: > why is data calibration more of a problem for IBM > models than for other modelling technologies? What sorts > of problem domains are you particularly concerned about? As a general statement, it is always more difficult to collect data from small scale phenomena than for large scale phenomena regardless of domain. Consider the ecological domain as an example. Improved sensing, radio and satellite methods are providing data for individuals, where this was not possible in the past; however, getting population data is generally easier than obtaining specific data from individuals in the population. The concepts of population and sample in statistics gives us some justification for this. What holds true for ecology most likely holds true for other fields---that is, procuring the data can be expensive, time consuming and difficult. > I can allow as how it would be difficult to calibrate > an IBM model of a forest precisely so that each tree-agent > is parameterized via data taken from its respective > real-tree in the forest. Yet, all modelling technologies > make do with some degree of approximation. A reservoir-flow > model of tree-species interaction in a forest would simply > treat all of the trees of each species as "one big tree" > of that species with respect to some data (such as concentration, > nutrient uptake, waste-production, and etc.) while ignoring > other data (such as spatial distribution, variety within the > species, and etc.) This will be justified for certain questions > about forest dynamics, but not for others, and might make more > sense for some problem domains than for others. Yes, agreed. > You seem to be suggesting that this is fundamentally more of a problem > for IBM models than for other modelling technologies. ...In the sense that it is generally less economic and more difficult to obtain individual data for basic reasons of time, cost and other factors. New sensing methods are helping but, yes, I would say that IBMs are more difficult to create and calibrate due to the sheer amount of data that are needed. I have copied by colleague in Miami to comment on this aspect since I am not a field person--i.e., not going out to tag gators :) This is not a theoretical issue; it is one of practicality relating to cost, time and availability of data. > ...and, please!, it almost *hurts* to use the acronym IBM! > could we use "ABM" for Agent-Based Models? I think it fits As you say, IBMs have precedence in the ecological community and so I see no need for changing this term when speaking of ecological matters. Regarding something more generic, I actually prefer the word "Object". "Agent" sounds like something human. Anyway, I realize we are discussing semantics at this point. -paul Paul A. Fishwick E-Mail: fishwick@cise.ufl.edu Dept. of Computer & Info Phone & FAX: (352) 392-1414 Science and Engineering WWW: http://www.cise.ufl.edu/~fishwick University of Florida (PGP Key available at above WWW address) P. O. Box 116120 332 Bldg. CSE, Gainesville, FL 32611-6120 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 18:32:34 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Thu, 17 Apr 1997 13:32:34 -0400 (EDT) Subject: calibration Message-ID: <199704171732.NAA22113@tide.cise.ufl.edu> Well, I am glad that this new mailing list called 'simsoc' is finally encountering some discussion and somehow we managed to bring in the SWARM mailing list. So much the better as this is appropriate. To respond to Philippe: > Perhaps you are talking about ecosystems from your simulation-oriented > point of view, as the word 'calibrate' indicates. Perhaps I am. > I think that a complex > system like an ecosystem cannot be modeled with some large FSM or even > with a set of stochastic equations (except in the short-term): in an > ecosystem there is probably no unique latent 'model' to be discovered, > against which we could compare some data and make adjustments. Instead > there probably exists many possible modes, with the system unpredictably > switching from time to time to one or another. I am not sure that we could argue this one way or the other. I see no reason why a model could not be developed. When you say "many possible modes", this can easily be modeled by stepping up a level of abstraction so that the modes are modeled as well. > The first thing to do could be to try to recognize these modes. Yes, as part of our systems identification process. > Because > switching is unpredictable (ecosystems seem to exhibit self-organized > critical modes), it makes no sense to try to 'calibrate' the whole system, > in the same manner simulation engineers calibrate a model of a factory or a > flexible workshop. I cannot see your logic. You say it makes no sense but do not provide any justification for your feeling. I think we'd need to get into specifics in order to continue along these lines. -paul ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 18:50:33 1997 From: swarm-modelling@santafe.edu (J.J. Merelo Guervos) Date: Thu, 17 Apr 97 19:50:33 +0200 Subject: Simulating Individual B In-Reply-To: <199704171613.KAA22412@santafe.santafe.edu> References: <199704171613.KAA22412@santafe.santafe.edu> Message-ID: <9704171750.AA09973@kal-el.ugr.es> >>>>> "cgl" == cgl writes: cgl> One group working with such data in the context of ABMs is cgl> the Emergent Solutions group at Coopers and Lybrand, who have cgl> been building models of the propagation of fads and fashions Anything published on that topic? Any reference? Of course, I could also push our own model, in which agents look at products, and decide to buy them or not; then they are punished or rewarded according to how much each product is publicized. Results show that it has an uncanny similitude with reality. But I dont have much quantitative data to support that model, only qualitative. I would like to have it, though. See ya! JJ -- JJ Merelo | http://kal-el.ugr.es/htbin/jj-plex Grupo Geneura ---- Univ. Granada | http://kal-el.ugr.es/ ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 20:34:15 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Thu, 17 Apr 1997 15:34:15 -0400 (EDT) Subject: ABMs Message-ID: <199704171934.PAA22512@tide.cise.ufl.edu> I think we are converging on an agreement: ABMs or IBMs (whatever your preference) will provide good models as long as 1) these particular models happen to answer the set of questions that you are asking of the model in the first place, and 2) enough quality data exist to use the ABMs effectively. In the ATLSS Project (across-trophic simulation of the Everglades), we are all pretty-much "pro-ABM/IBM" however, I was surprised to find out that IBMs were still controversial in the Ecology community, anyway. Maybe some of our Eco-experts could shed some light on this? -paul ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 20:46:35 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Thu, 17 Apr 97 15:46:35 EDT Subject: Simulating Individual Behavior In-Reply-To: <9704171658.AA28614@sfi.santafe.edu>; from "Ginger Booth" at Apr 17, 97 12:49 (noon) Message-ID: <9704171956.AA03098@sfi.santafe.edu> Paul, Re "controversy". Kinda depends on whether: a.) You believe you need to convince a middle-aged world-reknowned expert on doing things "the other way" (doesn't matter which way) that "your way" is better, or, b.) You believe that "your way" is validated (or not) by results and follow your own piper. I suspect a.) is impossible. If you pursue b.) and are really successful, you get to be the expert of a.) sometime. If you're really good, you may be dead by "sometime". :) My two cents. :) Ciao, Ginger ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 22:50:32 1997 From: swarm-modelling@santafe.edu (Doug Donalson;) Date: Thu, 17 Apr 1997 14:50:32 -0700 (PDT) Subject: ABM/I(C?)BM Message-ID: Ah, you have all jumped into the topic closest to my heart (well maby not the closest, but close). I am working on IBM's in ecology and am going through the wars of justification of a new technique. 1.) To Chris (ABM/IBM) You can call a gas station attendent a petroleum transfer engineer but he still smells like benzene! ;-) (Sorry Chris, I couldn't resist.) By the way, IBM in ecology doesn't just imply individual agents. It can also refer to differential equation based biomass change models where detailed physiology is included. See DeAngelis and Gross Individual Based Models. WARNING: PERSONAL OPINION NOW RUNS AMUCK!!!! 2.) To understand the attitude of mainstream theoretical ecology you have to start with it's emergence in the early '70s. Many of the builders were transfers from physics and were hoping to bring the idea of a nice clean set of paradigms (sorry for all the mis-spells) to ecology. There is a hugh litrature of stability boundries and equlibrium boundries for large groups of differential equation based models. Those who have put in so much hard work probably don't take kindly to being told that the assumptions behind their models are broken by almost every real ecological system! (The truth does hurt doesn't it?) The major arguments against complex models is that including parameters that are not precisly known can lead to large error propigation and misleading results. In particular, movement behavior is considered especially suspect. Unfortunatly for the detractors of complex models (see how easily I got out of the ABM/IBM discussion!) is that their argument points a lot of fingers at the simple models as well. If a model is highly sensitive to a single parameter or a group of parameters then what does that say about the results of a model that solves this problem by ignoring them! In a discussion with Alan Hastings (UC Davis) he stated that we should probably not yet study more than two or three species interactions because we (after 20+ years?) don't yet know enough about them. Maby this is because there is no such thing as a two or three species interaction in the real world. My work to date (pre-swarm) with a model I call the Heuristic Acynchronous Discrete Event Simulation or HADES, has shown that the assumptions of simple, differential equation based models can compleatly change the outcomes of a simulation. We used the original Lotka-Volterra predator-prey models as a base line and then successivly removed two of the main assumptions, that demographic stochasticity can be ignored and that the populations can be modeled as "well-mixed" at all times. The results showed that the equlibrium resullts predicted by the differential equation based model held fairly true over a large range of system sizes but that the time to extinction varied greatly with space playing both a stabilizing and destabilizing role at different system sizes. Bottom line. There is a lot of work that needs to go into understanding this new realm of complex models. One of the key areas is that of error propigation due to a combination of multiple parameters and inaccuracies in their measurements. Right or wrong, we have to prove the validity of the models to the main stream theoretical ecological community. Be of strong heart, remember, the young doctor who first proposed sterile surgical techniques was just about driven out of the profession by the greatest medical minds of his time! Doug Donalson *************************************************************************** * Doug Donalson * Office: (805) 893-2962 * * Ecology, Evolution, and Marine Biology * Home: (805) 961-4447 * * UC Santa Barbara * email donalson@lifesci.ucsb.edu* * Santa Barbara Ca. 93106 * * *************************************************************************** * * * The most exciting phrase to hear in science, the one that * * hearlds new discoveries, is not "EUREKA" (I have found it) but * * "That's funny ...?" * * * * Isaac Asimov * * * *************************************************************************** ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 23:14:03 1997 From: swarm-modelling@santafe.edu (Doug Donalson;) Date: Thu, 17 Apr 1997 15:14:03 -0700 (PDT) Subject: Simulating Individual Behavior In-Reply-To: Message-ID: Be careful of this method. There are at least two main problems you can run into. The first is that you can force fit a set of parameters to a model and get a result out that matches some experimental data but you have no guarentee that you have any more that a pretty picture. For a good example of what not to do see Gary Harrision differential equation based modeel of Luckinbill's Paramecium/Didinium experiments in the Jan 94 Ecology issue in the concepts section. Real systems are not deterministic. If you run a field experiment 20 times you will get 20 results and hope that there is enough commonality to get statistical power. Therefore, the result you are to which you are callibrating is just a sample from a statistical distribution. If that distribution has a wide varience then you may have fine-tuned your model to an point far from the expected behavior. A second trap is that if your assumption of the basic mechanisim is incorrect you may fine-tune to a different system that has the same result for that set of parameters. Unfortunatly, I don't has a good set of procedures to eleminate these hazzards. I am also not saying that this is a bad method, it happens to be one I saw abused in an attempt to justify the use of a differential equation based mode. Doug Donalson *************************************************************************** * Doug Donalson * Office: (805) 893-2962 * * Ecology, Evolution, and Marine Biology * Home: (805) 961-4447 * * UC Santa Barbara * email donalson@lifesci.ucsb.edu* * Santa Barbara Ca. 93106 * * *************************************************************************** * * * The most exciting phrase to hear in science, the one that * * hearlds new discoveries, is not "EUREKA" (I have found it) but * * "That's funny ...?" * * * * Isaac Asimov * * * *************************************************************************** On Thu, 17 Apr 1997, Jan Kreft wrote: > OK, I agree with Paul and Mike that it can be very tricky to determine the > parameters for individuals. Bacteria as individuals, e. g., are simply so > small that only the latest high-tech methods could give estimates for the > parameters you would have to use in a model (and that costs a lot of > timemoney). Almost always microbiologists measure the population average > and disregard the heterogeneity of the population. > > But I can see another way to solve the problem. If you start with a > conceptual model with all the parameters and mechanics you need and only a > rough idea of actual values (from population average measurements) you can > run the sim and compare the output with the real system to be modeled. > Then you optimize the parameters to achieve the desired output. Perhaps > call it back-calibration. (Is there an official term for it?) > > For that purpose, you need a parameter manager (see previous discussion > with "parameter" in the subject line) to allow a search through param > space in an evolutionary fashion. As far as I can tell, Gecko's param > manager is (the only one?) suited for that task. > > BTW, does IbM hurt also? If so, sorry for that ;-). Point being is that > that's the term used in the literature and if I want to search for new > literature, I rely on this keyword. Therefore I would want to use this > keyword myself sometime. > > Cheers, > > Jan. > > Mike Brown wrote: > > > I think Paul's question raises two important points - though they may be more > > about the organization of science than about the inherent difficulties of > > SWARM or any other ABM (Thanks, Chris. I like that acronym better too.) > > Ecologists are concerned with aggregations of individuals; they have not had > > to know as much about individual species as, for example, zoologists. > > Similarly, microeconomists focus on the behavior of individual firms and > > macroeconomists on the aggregate behavior of the economy, etc. > > > > ABM create two problems. First, the ecologist and macroeconomist have to > > "ratchet down" and study the behavior of individual entities. Moreover, they > > have to know enough about individual behavior to determine which specific > > behaviors might be relevant to the question under investigation. While this is > > not an insurmountable problem, it does demand a new focus for researchers. > > > > Second, there are some disciplines where data on the behavior of individual > > "agents" simply has not been studied. To take a bad example, look at > > economics. Macro studies the behavior of the aggregate, and micro the study of > > the firm -- but who has been looking at the behavior of the consumer? We have > > been able to make a "rational actor" assumption for so long that we have not > > bothered to collect data about the real-life behavior of induividual > > consumers. > > > > For these reasons, I think Paul is very right -- modeling and validating the > > behavior of individual entities can be very tricky. > > > > Mike > > > > Paul Fishwick wrote: > > > > >A key problem with IBM is not that it is necessarily > > >computationally prohibitive, but that not enough data are > > >available to calibrate the model... > > > > Paul - why is data calibration more of a problem for IBM > > models than for other modelling technologies? What sorts > > of problem domains are you particularly concerned about? > > > > I can allow as how it would be difficult to calibrate > > an IBM model of a forest precisely so that each tree-agent > > is parameterized via data taken from its respective > > real-tree in the forest. Yet, all modelling technologies > > make do with some degree of approximation. A reservoir-flow > > model of tree-species interaction in a forest would simply > > treat all of the trees of each species as "one big tree" > > of that species with respect to some data (such as concentration, > > nutrient uptake, waste-production, and etc.) while ignoring > > other data (such as spatial distribution, variety within the > > species, and etc.) This will be justified for certain questions > > about forest dynamics, but not for others, and might make more > > sense for some problem domains than for others. > > > > Thus, there is data and there is data. All modeling technologies > > must pick and choose among the data, and one always has to > > focus on some reasonable subset of data. You seem to > > be suggesting that this is fundamentally more of a problem > > for IBM models than for other modelling technologies > > - can you elaborate? > > > > ...and, please!, it almost *hurts* to use the acronym IBM! > > could we use "ABM" for Agent-Based Models? I think it fits > > better anyway, as an agent in this class of models is not > > always an "individual" in the common sense of that term.... > > I know the term has some historical precedent for models > > in this class, but the acronym IBM induces a certain amount > > of, shall we say, cognitive dissonance, no? (not that the > > acronym "ABM" itself is inviolate with respect to prior > > cognitive content....but, still!.....) > > > > > > Chris Langton > > > > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 23:33:17 1997 From: swarm-modelling@santafe.edu (Doug Donalson;) Date: Thu, 17 Apr 1997 15:33:17 -0700 (PDT) Subject: Simulating Individual Behavior In-Reply-To: <9704171956.AA03098@sfi.santafe.edu> Message-ID: Amen! But possibly not as bad as all that, there are top researchers such as Roger Nisbet and Bill Murdoch at UCSB who see relvance and need for this type of model. (Even if Roger and I don't always agree on all the details.) My present project is setting up an individual-based spatially-explicit model to begin to explore the dynamics of the Red Scale/Aphytis host-parasitoid system that Bill has been studying for some time. I was approched by Bill because he felt that this new method might be able to shed further light on the mechanisms that lead to the observed system dynamics. Now if only we could get the concurrent schedule part of the SWARM schedule working, oh well ... (Actually Roger Burkhart is being most helpful!) Sorry Ginger, I went and contaminated you modeling group with a tech statement. Doug Donalson *************************************************************************** * Doug Donalson * Office: (805) 893-2962 * * Ecology, Evolution, and Marine Biology * Home: (805) 961-4447 * * UC Santa Barbara * email donalson@lifesci.ucsb.edu* * Santa Barbara Ca. 93106 * * *************************************************************************** * * * The most exciting phrase to hear in science, the one that * * hearlds new discoveries, is not "EUREKA" (I have found it) but * * "That's funny ...?" * * * * Isaac Asimov * * * *************************************************************************** On Thu, 17 Apr 1997, Ginger Booth wrote: > Paul, > > Re "controversy". Kinda depends on whether: > > a.) You believe you need to convince a middle-aged world-reknowned expert > on doing things "the other way" (doesn't matter which way) that "your way" is > better, or, > > b.) You believe that "your way" is validated (or not) by results and > follow your own piper. > > I suspect a.) is impossible. If you pursue b.) and are really > successful, you get to be the expert of a.) sometime. If you're really good, > you may be dead by "sometime". :) > > My two cents. :) > > Ciao, > Ginger > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 17 23:44:40 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Thu, 17 Apr 1997 15:44:40 -0700 Subject: Simulating Individual Behavior References: Message-ID: <3356A7D8.15A2426B@ix.netcom.com> This is my first posting to any of the Swarm mailing lists, so maybe I'm supposed to introduce myself: Hello, I'm Mark, and I've been fooling around with microsimulation (and other approaches to what is now called ABM) off and on since about 1976. My main interest is in the ability of ABM to unify the scientific enterprise: different scientific communities talk about the same agents in different, sometimes contradictory, ways [this is _especially true in social and behavioral science], and the explicit realization that these _are_ the same agents can be the focal point for a deeper level of consistency among disparate theories and between theory and observation. That's why this particular thread has interested me a great deal. Jan Kreft wrote: > > But I can see another way to solve the problem. If you start with a > conceptual model with all the parameters and mechanics you need and only > a rough idea of actual values (from population average measurements) you > can run the sim and compare the output with the real system to be > modeled. Then you optimize the parameters to achieve the desired output. > Perhaps call it back-calibration. (Is there an official term for it?) There is a tradition of (social) microsimulation going back to G. Orcutt (1957) "A new type of socio-economic system", _Review of Economics and Statistics_ 58: 116-123 and then the classic Orcutt, G. et al. (1961) _Microanalysis of Socioeconomic Systems: A Simulation Study_ (NY/Evanston/London). There's also the more recent Orcutt, G. et al (eds.) (1984) _Microanalytic Simulation Models to Support Social and Financial Policy_ (Amsterdam). There's no doubt much more recent stuff, but I've been out of touch with this particular scene for a while. The Germans have been doing (social) microsimulation like this in grand style for about twenty years now, including official applications to federal policy. Most approaches that I'm aware of distribute parameters across agents based on dependent frequency distributions that have been derived from disparate aggregate data sources. I think this may be equivalent to the kind of "back-calibration" you suggest here, in that failure to validate a model must necessarily feed back into the assumptions used to estimate the initial population (and its agents' parameters). These models never (AFAIK) used anything more agent-specific than statistical cohorts derived from the dependent distributions. Although there is a mountain of literature in German, my favorite is a published dissertation: Helmut Vetterle (1986) _Konstruktion und Simulation mikroanalytischer Modelle. Die Methode der Mikrosimulation und ihre Anwendung_ ("Design and Simulation of Microanalytical Models: Method and Application of Microsimulation"). Augsburg: MaroVerlag. >150 refs. I suspect that this book is out of print, and that there is no English translation. If I can get Vetterle's consent, I'd be willing to put the dissertation on my website in the foreseeable future, and even translate part of it into English -- if there's any interest. (Maybe this is all old hat and I'm making a fool of myself -- somebody kick me if that's the case.) -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 00:41:54 1997 From: swarm-modelling@santafe.edu (Steve Emsley) Date: Fri, 18 Apr 1997 00:41:54 +0100 Subject: Simulating Individual Behavior In-Reply-To: <3.0.32.19970417095049.006ebc8c@poseidon.obs-vlfr.fr> (message from Philippe LAVAL on Thu, 17 Apr 1997 09:50:51 -0100) Message-ID: <199704172341.AAA00780@dial1199.csv.warwick.ac.uk> Just some thoughts: All of the individual-based models in ecology that I am aware of seek to model spatial resource competition. These range between 'artificial ecosystems", such as Gecko, and specific applications, such as Dave Green's CA study of 'Crown-of-Thorns' outbreaks. These models belong to the 'divide and conquer' school. Personally, my interest is in developing a model as a 'surrogate experimental system" (Olson 1995). Perhaps 'simulation' is a more accurate description. As such, I question Philippe Laval's statement that "in an ecosystem there is probably no unique latent 'model' to be discovered, against which we could compare some data and make adjustments." Although I concur that there is no 'whole ecosystem model' surely, in terms of emergent phenomena, one would not expect such a model to exist. However, there are latent models for physiological & behavioural processes within organisms which, by virtue of their interaction, determine one of many 'ecosystem' models. Mike Brown says "Ecologists are concerned with aggregations of individuals; they have not had to know as much about individual species as, for example, zoologists." In my field, plankton ecology, mean-field modellers have dominated the field for decades. Mathematically-endowed researchers will reduce any mean field coupled system of differential equations into an idealised excitable system. It is their goal in life. Forget individual differences or interaction between individuals - it just muddies the water. "First, the ecologists ... have to 'ratchet down' and study the behavior of individual entities ... it does demand a new focus for researchers." This is exactly what is required. In order to approach a plankton SWARM simulation this 'controversial (mad!)' researcher has had to enlist help from researchers into cell-orientated mechanistic models of physiology and behaviour. Even then we're only dealing with a drop in the ocean!! At least, it's not a multicellular drop! Jan Kreft says " If you start with a conceptual model with all the parameters and mechanics you need and only a rough idea of actual values (from population average measurements) you can run the sim and compare the output with the real system to be modeled. Then you optimize the parameters to achieve the desired output. Perhaps call it back-calibration." Excellent. I attempted this argument with a mean-field modeller PI on our "Testable Model" project by suggesting that my model will be intrinsically testable. I argue that for my simulation if I can adequately model the light field (easy) and turbulent environment (more difficult) then seed my 'ocean' with agents having a log normal spread of parameter values I can spin up and, assuming my mechanistic models are realistic, evolve a population that has parameter values optimised for fitness in their artificial environment (and comparable with organisms to be found in real oceanonographic settings). Sort of a spatially-extended genetic algorithm. Ginger says " For it suggests what kind of terrestrial ecosystems (rather safer and easier to manipulate than open ocean :) could conceivably offset CO2 emissions." Suggest you glance at Wally Broeker (1991) Keeping global change honest". Anti-intuitively, calcitic organisms PRODUCE CO2 when forming their skeletons. Although once it gets down into the deep-sea sediments it's there for a residence time c. 1 million years you have to find a way of increasing productivity first. Apart from dumping all our cars into the Pacific (Martin iron-feritilisation hypothesis) chances are that global warming transients furthur stabilise the mixed layer leading to REDUCTION in productivity. Lovelock is probably correct - grow more trees (and build timber houses, furniture or bury the crop!!). With that warming thought, I'll sign off! -- Steve Emsley Ecosystem Analysis & Management Group, University of Warwick, UK sme@oikos.warwick.ac.uk ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 00:45:00 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: Fri, 18 Apr 1997 10:45:00 +1100 (E ) Subject: Simulating Individual Behaviour In-Reply-To: <199704171743.LAA19514@grasshopper.santafe.edu> from "glen e. p. ropella" at Apr 17, 97 11:43:15 am Message-ID: <9704172345.AA06980@malthus> >From: cgl@sfi.santafe.edu >To: swarm-modelling@santafe.edu >Subject: Re: Simulating Individual B >Date: Thu, 17 Apr 1997 10:13:28 -0600 > >> Actually, consumer behaviour is an important research area in the field of >> marketing (which seems to be a mixture of applied microeconomic and >> psychology, at least to an outsider). > > Right - I think that there is a tremendous amount of data on >consumers that have been taken in market "demographic" studies, >and of course there are Nielson ratings, the Billboard top-10 >and etc... > > These data exist in corporations, and they may or may not be >available to scientists for research use. > > One group working with such data in the context of ABMs is >the Emergent Solutions group at Coopers and Lybrand, who have >been building models of the propagation of fads and fashions >throughout (initially music) markets - they used a >demographic data base of 150,000 americans, provided by >(I'm assuming) the entertainment industry.... > I'm pleased that the question of who studies consumer behaviour has been taken up, and answered. As well as the databases mentioned by Chris, there are also the databases from scanners in supermarkets, although it is not so easy to obtain access to these. (As a plug, see a recent paper by my coauthors and me which uses some scanner data in an ABM model: Midgley DF, Marks RE, and Cooper LG, "Breeding hybrid strategies" _Management Science_ 43(3): 257-275, March 1997.) Another area of data on individual consumer behaviour in economics comes from experimental economics: examining the choices of subjects in laboratory settings, when faced with varying settings. Robert Marks ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 02:07:29 1997 From: swarm-modelling@santafe.edu (Randall Gray) Date: Fri, 18 Apr 1997 11:07:29 +1000 (EST) Subject: Simulating Individual B In-Reply-To: (MICHAEL.L.BROWN@cpmx.saic.com) Message-ID: <199704180107.LAA11737@njal.ml.csiro.au> ABM's do indeed pose some problems w.r.t. "calibration". In a project of days gone by we skirted the issue somewhat by attempting to "bracket" reality in our range of assumptions of behaviour. We were looking at potential contamination of a marine trophic chain due to period (regular) dumping of a low level hazardous waste at sea. We ran a range of models which went from the most pessimistic to quite optimistic in terms of behaviour with respect to tainted waters and assessed the results. I am certain that what we did was not optimal, but collecting the sort of data we'd have required would have been (and still is) prohibitively expensive. I suppose the issue is "How close to reality do we need to get before the simulation exhibits the same sorts of strange attractors or ritical modes we see in Real Life?" Moreover, how do we correctly identify that we've arrived? I suppose that the problem with the ecosystem work is that you may only see a few critical modes in an ecosystem under study, but with your simulation you might actually come across dozens more. Some of the strange attractors arise from the implementation of the model, some are critical modes of the simulation, and some are also critical modes of the system being simulated. You always *hope* that the strange attractors of the system are strange enough to raise your suspicions, but differentiating the other two seems to be one of those harder problems. -- Randall ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 14:56:51 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Fri, 18 Apr 97 9:56:51 EDT Subject: Simulating Individual Behavior In-Reply-To: <199704172341.AAA00780@dial1199.csv.warwick.ac.uk>; from "Steve Emsley" at Apr 18, 97 12:41 (midnight) Message-ID: <9704181406.AA15206@sfi.santafe.edu> Hey, Steven! > Ginger says " For it suggests what kind of terrestrial ecosystems (rather > safer and easier to manipulate than open ocean :) could conceivably offset > CO2 emissions." Suggest you glance at Wally Broeker (1991) Keeping global > change honest". Anti-intuitively, calcitic organisms PRODUCE CO2 when > forming their skeletons. Although once it gets down into the deep-sea > sediments it's there for a residence time c. 1 million years you have to > find a way of increasing productivity first. Apart from dumping all our > cars into the Pacific (Martin iron-feritilisation hypothesis) chances are > that global warming transients furthur stabilise the mixed layer leading > to REDUCTION in productivity. Lovelock is probably correct - grow more > trees (and build timber houses, furniture or bury the crop!!). > > With that warming thought, I'll sign off! Another paper: "A large northern hemisphere terrestrial CO2 sink indicated by the 13C/12C ratio of atmospheric CO2", Science, Aug 25 1995. Since the title doesn't do much, from the abstract: "A strong terrestrial biospheric sink was found in the temperate lattitudes of the Northern Hemisphere in 1992 and 1993, the -magnitude of which is roughly half that of the global fossil fuel burning emissions for those years-." Later, Ginger ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 15:05:25 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Fri, 18 Apr 1997 15:05:25 +0100 (BST) Subject: Simulating Individual Behavior In-Reply-To: Message-ID: Doug, thanks for your comments. Doug Donalson wrote: > Be careful of this method. There are at least two > main problems you can run into. The first is that you can > force fit a set of parameters to a model and get a result out > that matches some experimental data but you have no guarentee > that you have any more that a pretty picture. The way to avoid such a force fit that I'm thinking about is optimizing/evolving the parameter set. Suppose that the parameter set is optimized with respect to the individuals' fitness (one or a few definitions of fitness or performance could be used). Then you compare the result of only those sims with optimized params with real data (in my case lab data, therefore the variation will be better known hopefully). (Real data also are the result of (evolutionary) optimized behavior). If you don't get a reasonable match the assumptions about the basic mechanism will be incorrect or lacking important points. Does this make sense to you? Do you see a better way to get around the lack of appropriate, not theoretically biased quality data? Simply setting parameters to experimental values beforehand could get you into serious problems also if a tiny deviation of a param has a large effect and it happens that that experimental value is just a bit wrong... Jan. > [...] > If you run a field experiment 20 times you > will get 20 results and hope that there is enough commonality > to get statistical power. Therefore, the result you are > to which you are callibrating is just a sample from a statistical > distribution. If that distribution has a wide varience then you > may have fine-tuned your model to an point far from the expected > behavior. > > A second trap is that if your assumption of the basic mechanisim > is incorrect you may fine-tune to a different system that has > the same result for that set of parameters. > > Unfortunatly, I don't has a good set of procedures to eleminate > these hazzards. I am also not saying that this is a bad method, > it happens to be one I saw abused in an attempt to justify > the use of a differential equation based mode. > > Doug Donalson ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 15:32:54 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Fri, 18 Apr 1997 15:32:54 +0100 (BST) Subject: Simulating Individual Behavior In-Reply-To: <3356A7D8.15A2426B@ix.netcom.com> Message-ID: Mark, thanks for the pointers. Mark P. Line wrote: > ...... > There is a tradition of (social) microsimulation going back to G. Orcutt > (1957) "A new type of socio-economic system", _Review of Economics and > Statistics_ 58: 116-123 and then the classic Orcutt, G. et al. (1961) > _Microanalysis of Socioeconomic Systems: A Simulation Study_ > (NY/Evanston/London). > > There's also the more recent Orcutt, G. et al (eds.) (1984) > _Microanalytic Simulation Models to Support Social and Financial Policy_ > (Amsterdam). > ...... > The Germans have been doing (social) microsimulation like this in grand > style for about twenty years now, including official applications to > federal policy. Most approaches that I'm aware of distribute parameters > across agents based on dependent frequency distributions that have been > derived from disparate aggregate data sources. I think this may be > equivalent to the kind of "back-calibration" you suggest here, in that > failure to validate a model must necessarily feed back into the > assumptions used to estimate the initial population (and its agents' > parameters). These models never (AFAIK) used anything more > agent-specific than statistical cohorts derived from the dependent > distributions. > > Although there is a mountain of literature in German, my favorite is a > published dissertation: > > Helmut Vetterle (1986) _Konstruktion und Simulation mikroanalytischer > Modelle. Die Methode der Mikrosimulation und ihre Anwendung_ ("Design > and Simulation of Microanalytical Models: Method and Application of > Microsimulation"). Augsburg: MaroVerlag. >150 refs. > > I suspect that this book is out of print, and that there is no English > translation. If I can get Vetterle's consent, I'd be willing to put the > dissertation on my website in the foreseeable future, and even translate > part of it into English -- if there's any interest. (Maybe this is all > old hat and I'm making a fool of myself -- somebody kick me if that's > the case.) > That would be fine! Would you make that diss available? I, personally, don't need a translation but most people will :-(. Jan. > ....... ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 17:00:09 1997 From: swarm-modelling@santafe.edu (Rick Riolo) Date: Fri, 18 Apr 1997 12:00:09 -0400 (EDT) Subject: Simulating Individual Behavior In-Reply-To: Message-ID: Jan, I think evolutionary algorithms (EAs) can contribute to the exploration of parameter value spaces, but I think one should use such techniques with care and with eyes open. Some of the issues include: 1. How do you make sure your definition(s) of fitness, or more generally, your implementation of selection pressures, will be the same as those that exist in the world you are trying to model? If they are not, then your model may evolve some parameter values but they won't reflect the underlying world "parameter" values. 2. EA's are notoriously touchy. You have to make choices about mechanisms and parameter values for the EA itself, you have to make representation choices, and so on. They definitely don't come with any kind of guarantee that they will find optimal (or even near optimal) "solutions" (ie sets of model parameter values in this case). 3. Part of the reason for 2 is that the spaces one searches with EA's are almost always "complex", often with many many "local hills" of pretty good solutions. Thus you might run the same algorithm from random starts 20 times and come up with 20 different "solutions" (parameter value settings). Then you have to choose amoung those, and of course there are all the other nearly equally good parameter settings you didn't ever see. How does one do that? Usually you go back to whatever empirical data, first principles, etc, you know about domain in question...that is, you are (almost) back to where you started from! Now I don't want to leave you with the impression that I have it out for EA's in general or EA's in this context....I use them all the time! They can help in the search for parameter values that at least lead to desired behavior...but you still have to make the case that the values found are plausible parameters values, given whatever other evidence you can bring to bear. I just think one has to be careful not ask or expect EAs to do more than they can do. One interesting use of EA's in connection to model parameter searches is exemplified by John Miller's ANT paper (available as an SFI working paper): 96-03-011 Active Nonlinear Tests (ANTs) of Complex Simulation Models. The basic idea is to use EA's to find weak or questionable aspects of a model's design. The important conceptual change is that rather than finding some "one optimal parameter settings" one is looking for *any* of the (perhaps many) parameter settings that can lead to really bad model behavior. Steven Bankes at RAND has also suggested similar uses of EA's for various parts of "exploratory modeling." - r Rick Riolo rlriolo@umich.edu Program for Study of Complex Systems (PSCS) 1061 Randall Lab University of Michigan Ann Arbor MI 48109-1120 http://pscs.physics.lsa.umich.edu/PEOPLE/rlr-home.html On Fri, 18 Apr 1997, Jan Kreft wrote: > Date: Fri, 18 Apr 1997 15:05:25 +0100 (BST) > From: Jan Kreft > To: swarm-modelling@santafe.edu > Subject: Re: Simulating Individual Behavior > > Doug, > > thanks for your comments. > > Doug Donalson wrote: > > > Be careful of this method. There are at least two > > main problems you can run into. The first is that you can > > force fit a set of parameters to a model and get a result out > > that matches some experimental data but you have no guarentee > > that you have any more that a pretty picture. > > The way to avoid such a force fit that I'm thinking about is > optimizing/evolving the parameter set. Suppose that the parameter set is > optimized with respect to the individuals' fitness (one or a few > definitions of fitness or performance could be used). Then you compare the > result of only those sims with optimized params with real data (in my case > lab data, therefore the variation will be better known hopefully). (Real > data also are the result of (evolutionary) optimized behavior). If you > don't get a reasonable match the assumptions about the basic mechanism > will be incorrect or lacking important points. > > Does this make sense to you? Do you see a better way to get around the > lack of appropriate, not theoretically biased quality data? > > Simply setting parameters to experimental values beforehand could get you > into serious problems also if a tiny deviation of a param has a large > effect and it happens that that experimental value is just a bit wrong... > > Jan. > > > [...] > > If you run a field experiment 20 times you > > will get 20 results and hope that there is enough commonality > > to get statistical power. Therefore, the result you are > > to which you are callibrating is just a sample from a statistical > > distribution. If that distribution has a wide varience then you > > may have fine-tuned your model to an point far from the expected > > behavior. > > > > A second trap is that if your assumption of the basic mechanisim > > is incorrect you may fine-tune to a different system that has > > the same result for that set of parameters. > > > > Unfortunatly, I don't has a good set of procedures to eleminate > > these hazzards. I am also not saying that this is a bad method, > > it happens to be one I saw abused in an attempt to justify > > the use of a differential equation based mode. > > > > Doug Donalson > > > > ================================== > Swarm-Modelling is for discussion of Simulation and Modelling techniques > esp. using Swarm. For list administration needs (esp. [un]subscribing), > please send a message to with "help" in the > body of the message. > ================================== > ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 17:04:36 1997 From: swarm-modelling@santafe.edu (Stephen C. Upton) Date: Fri, 18 Apr 1997 10:04:36 -0600 Subject: Simulating Individual Behavior Message-ID: <3.0.32.19970418100435.00754004@tsa-po.lanl.gov> This is a great thread!!! First, I'm reminded of a lyric: IBM, UBM, We all BM for IBM (I believe this was from the science fiction book "With a Finger in my I" by D. Hofstader -- if not, then from some other 1970's scifi book on AI :-) Second, I'll paraphrase Hamming (R. W. Hamming, Numerical Methods for Scientists and Engineers, 1962): The purpose of modeling is understanding, not numbers. Third, I'll add some of my own thoughts: I am currently attempting to understand warfare. (For those that find this offensive, consider the better we understand the origin and causes of warfare, the better we can prevent it) There are quite a few models and simulations, as you may imagine, of various aspects and levels of warfare, from the individual soldier to the campaign level, e.g., Desert Storm. Each abstracts certain features which are relevant to a particular analysis. There have been several good points made that seem to cross problem domain boundaries: 1. The resolution or the level of abstraction of the agent (actor, individual, entity) 2. The resolution or the level of abstraction of the interaction between agents 3. Collecting, gathering, identifying data for 1 and 2 4. Identifying patterns or structures, emergent or not (modes, or the resultant behavior of the system as a whole) 5. The difference in modeling philosophy between ABM and ODE's or PDE's (The Newtonian legacy) 6. The relationship of the model to the "real world" (Maybe this was obvious!) There have been several attempts to model the campaign level using ABM's. However, for computational reasons,these were arguably unsuccessful. But the attempts bring up some good questions: At what level do you model your agent or their interactions? Is a Battalion sufficient or do I need to model individual tanks? How much detail is required? (More detail, More detail is the current rallying cry) How does that level of detail relate to system characteristics, e.g., location of strange attractors, etc.? How do you know when you have the level of detail and interactions correct, especially if there isn't a whole lot of data, at either the agent level or the system level? Steve Emsley says, "Personally, my interest is in developing a model as a 'surrogate experimental system". I concur. I also don't have much of a choice, except that historical data provides some clues as to what the important parameters, plus some thinking about the processes in general. But then I'm not sure if those parameters were a function of some other interactions, e.g., the evolution of new tactics as a function of technology. I also sense some "My model is better than your model", i.e., ABM's are "better" than ODE models, or ODE's are "better" than ABM's. This is an easy trap to fall into. Each is abstracting system characteristics and behavior differently. Certainly, thermodynamics is as useful as statistical mechanics, within their respective regimes. However, I currently believe the ODE folks are more guilty of this, but we don't want the pendulum to swing all the way to other side either. Finally, we are all probably guilty of thinking our models represent the real world in some useful fashion. We should always maintain a certain amount of sceptism for any model and remember the paraphrase, modeling is for understanding. Love to hear your comments. thanx *** * **** * ***** ********* ** ****** ***** *** ***** ******** ********* Stephen C. Upton TSA-5, MS F602 Los Alamos National Laboratory Los Alamos, NM 87545 505-667-9435 FAX 505-665-2017 upton@lanl.gov ******* ********* *** ** *** ******** **** ****** ** ****** **** *** *** ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 18:30:45 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Fri, 18 Apr 1997 18:30:45 +0100 (BST) Subject: Simulating Individual Behavior In-Reply-To: <3.0.32.19970418100435.00754004@tsa-po.lanl.gov> Message-ID: Stephen C. Upton wrote: > ........ > There have been several attempts to model the campaign level using ABM's. > However, for computational reasons,these were arguably unsuccessful. But > the attempts bring up some good questions: > At what level do you model your agent or their interactions? Is a Battalion > sufficient or do I need to model individual tanks? > How much detail is required? (More detail, More detail is the current > rallying cry) > How does that level of detail relate to system characteristics, e.g., > location of strange attractors, etc.? > How do you know when you have the level of detail and interactions correct, > especially if there isn't a whole lot of data, at either the agent level or > the system level? You can only know that if you vary the level of detail and look at the results. Maybe there is some sort of optimum, then that's what you would have to search for. > ......... > Finally, we are all probably guilty of thinking our models represent the > real world in some useful fashion. We should always maintain a certain > amount of sceptism for any model and remember the paraphrase, modeling is > for understanding. Yes! Cheers, Jan. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 18:54:45 1997 From: swarm-modelling@santafe.edu (Daniel J Shapero) Date: Fri, 18 Apr 1997 13:54:45 -0400 Subject: Simulating Individual Behavior In-Reply-To: Message-ID: hi, this is my first message to the group. I have been reading the dialogue for about two weeks and feel that something that happened to me yesterday pertains to Dougs message about parameters and conditions of models. I am an undergraduate at Johns Hopkins working on a Buffer Stock Model for a professor with Todd Allen. After getting the program to work (in SWARM) and the the GA values to converge to specific values, we showed the program to the overseeeing prof. he said that our values were all wrong. Todd and I then looked at the model and realized that we took into account something that the professor did not. Our agents adopted rules that had good average utilities, but also the length of time over which that rule was successful was important... a good rule for 10 years was better than one for 2 years. This parameter definately mimics life. But me realized that we did not know which should be more important to the agent, length of duration or actual average utility. So how did the professor get an answer... he did not take length of time importance into account when deriving the analytic answer. once we eliminated that part of the adoption rule, the converged values were correct. The moral is that I thought that the most interesting part of the experience is that ABM(IBM, OBM, whatever) can be used to do parameter sweeps over the importances of length of rule v. average utility over that time. This sweep would derive a pretty picture of a fitness landscape in another variable... Then these new results could be compared to data(if there is any out there). On the other hand, this macroeconomics professor is not interested in these other mechanisms, the ones i feel are most interesting. It is going to be interesting how these results are going to be published (ABM along side MAcroeconomics) Macro using SWARM. I love it!!! hope you didn't mind my jump into the group, Daniel Shapero ps. doug, i like your quote at the end of your messages ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 18 21:03:21 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Fri, 18 Apr 1997 13:03:21 -0700 Subject: Simulating Individual Behavior References: Message-ID: <3357D389.610C2A23@ix.netcom.com> Jan Kreft wrote: > > Mark P. Line wrote: > > > Helmut Vetterle (1986) _Konstruktion und Simulation mikroanalytischer > > Modelle. Die Methode der Mikrosimulation und ihre Anwendung_ ("Design > > and Simulation of Microanalytical Models: Method and Application of > > Microsimulation"). Augsburg: MaroVerlag. >150 refs. > > That would be fine! Would you make that diss available? I, personally, > don't need a translation but most people will :-(. In the short term, I'm willing to loan the book out if it does happen to be out of print. Check your bookstore to see, and send me your snail address if you want me to loan you my copy. The address of the publisher given in the book is MaroVerlag, Benno Ka"smayr Riedingerstr. 24 [new zipcode?] Augsburg The copyright for the book is held by Beratungsgesellschaft fu"r angewandte Systemforschung mbH - Augsburg Haunstetterstr. 19 [new zipcode?] Augsburg (phone +49-821-571093) The ISBN is 3-87512-501-0. Please let me know what you find out about the book's print status. If it's out of print, I'll try to get permission from the outfit in Augsburg to put relevant excerpts on the Web (the parts about state-of-the-art (anno 1986) hardware and software technology are not very relevant, except historically) as soon as I can get to it. ==== I should have also mentioned Martin Clarke's work in spatial microsimulation, which he had been doing for quite a while by the mid-80's; I assume he's still doing it. See, for instance, Clarke, Martin (1986) "Demographic processes and household dynamics: a microsimulation approach", in Robert Woods and Philip Rees (eds.) _Population Structures and Models_. London: Allen & Unwin. ==== There was a whole Sonderforschungsbereich (SFB; a federally funded research focus) in Germany on microsimulation, especially with federal policy applications, which has probably ended by now. It was SFB 3, "Mikroanalytische Grundlagen der Gesellschaftspolitik" ['Microanalytical Foundations/Underpinnings/Bases of/for Social Policy'], in Mannheim or Frankfurt, I believe. The DFG (Deutsche Forschungsgemeinschaft, like NSF in this country) would certainly know how to access their mountain of grey literature. ==== After posting that message yesterday, I started wondering what had happened in microsim since I last took a look. I found a lot on the Web searching under the usual terms ('microsimulation', 'mikrosimulation', 'microanalytical', 'mikroanalytisch'). Here are a few items I liked: http://www.tc.cornell.edu/Edu/SPUR/SPUR96/Greta/report.html http://www-cpr.maxwell.syr.edu/demogctr/micropap/microlst.htm http://petty.econ.rochester.edu/nas.htm http://itkwww.kub.nl:2080/TUP/Fondslijst/Eco/9536-5.html http://www1.ifs.org.uk/research/personal/CzechModel.HTM and especially http://misic.soc.cornell.edu/ -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 19 02:56:57 1997 From: swarm-modelling@santafe.edu (Steve Emsley) Date: Sat, 19 Apr 1997 01:56:57 +-100 Subject: Simulating Individual Behavior Message-ID: <01BC4C65.3E212A20@p62.ascend3.is2.u-net.net> ------ =_NextPart_000_01BC4C65.3E212A20 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable ---------- From: Stephen C. Upton[SMTP:upton@lanl.gov] Sent: Friday, April 18, 1997 05:04 To: swarm-modelling@santafe.edu Subject: Re: Simulating Individual Behavior This is a great thread!!! Steve Upton's problem domain seems well suited to a hierarchical swarm approach. In my domain each bucket of seawater can contain more = particles than my computer has bytes of RAM and there is no obvious aggregation rule. However, it seems inherent in military modelling that there is a hierarchical structure and a chain of command. So, the agents are: soldiers -> platoon -> company -> regiment -> division -> army corresponding to a chain of command: lieutenant ->=20 captain -> major -> colonel -> general. The lieutenant polls his grunts = and passes the data up the line to the captain. The captain reports his = synopsis of the data from the lieutenants to the major and so on to the general = who, far from the conflict, passes directives down the chain of command.=20 (Apologies for my ignorance of military organisation, whether UK or = US!). "How much detail is required? (More detail, More detail is the current rallying cry)" My point is: As far as the general is concerned there are no individual = soldiers. However, individual soldiers could be modelled - but only a few, and = less=20 (than actually involved in a conflict) platoons, and even less regiments = etc.=20 Whereas modelling a general would require a context-dependent rule-based system I should imagine that, the furthur down the chain of command, = rules become more fuzzy until, at the level of the individual platoon, one may = as well use stochastic differential equations. Of course, one might argue = that engagements are won or lost not because of the 'normal' = distribution but due to the effect of the few on the many. Since heroism is probably = as elusive an AI concept as creativity I imagine that the aim of modelling = warfare is more an exercise in risk assessment than understanding. Anyway, the main point of this posting is: (1) SWARM has been designed with hierarchicies of subswarms in mind although, apart from a few posting, this feature isn't being exploited. = Perhaps the computational overheads of running individual swarms has been a disincentive. If that is the case, roll on parallel swarm! (2) Just because SWARM is suited for ABM does not preclude the inclusion of ODEs. Despite my last post, which may have suggested that I was denigrating mean-field modelling in respect of ABM, I use ODEs in my model. My ModelSwarm sends a step message to a ModelState class that polls my LightSpace, OceanSpace and PhytoPlankton objects for variables = and Runge-Kutta's them for 24 hours to produce the next day's state. More of = an Infinite State Machine than a FSM!=20 (3) Assuming hierarchical order then a "surrogate experimental system" = is=20 a WELL-DEFINED "descriptive model" of the level above. In addition, the = "surrogate system" requires a "descriptive model" of the level=20 below to satisfy the criteria of testability with real world data. = Before disappearing down the hole with diameter =3D Planck's length the hopeful = aim of the exercise is to define the ranges of the parameters used in the level = above ... if only to prevent mean-field modellers from tuning their systems = to their heart's conten... Sorry, [diatribeMode: OFF]. Excuse the musings of a lapsed Type 4 lurker but I couldn't resist this = thread - hail patch dynamics, there's noting like an invasion of territory (ie = the mailing-list) to bring out a spurt of evolution. Maybe it was an = experiment!? Steve Emsley, Ecosystems Analysis & Management Group, University of Warwick, UK. ------ =_NextPart_000_01BC4C65.3E212A20 Content-Type: application/ms-tnef Content-Transfer-Encoding: base64 eJ8+IjkAAQaQCAAEAAAAAAABAAEAAQeQBgAIAAAA5AQAAAAAAADoAAENgAQAAgAAAAIAAgABBJAG AEQBAAABAAAADAAAAAMAADADAAAACwAPDgAAAAACAf8PAQAAAFUAAAAAAAAAgSsfpL6jEBmdbgDd AQ9UAgAAAABzd2FybS1tb2RlbGxpbmdAc2FudGFmZS5lZHUAU01UUABzd2FybS1tb2RlbGxpbmdA c2FudGFmZS5lZHUAAAAAHgACMAEAAAAFAAAAU01UUAAAAAAeAAMwAQAAABwAAABzd2FybS1tb2Rl bGxpbmdAc2FudGFmZS5lZHUAAwAVDAEAAAADAP4PBgAAAB4AATABAAAAHgAAACdzd2FybS1tb2Rl bGxpbmdAc2FudGFmZS5lZHUnAAAAAgELMAEAAAAhAAAAU01UUDpTV0FSTS1NT0RFTExJTkdAU0FO VEFGRS5FRFUAAAAAAwAAOQAAAAALAEA6AQAAAAIB9g8BAAAABAAAAAAAAAMsQgEIgAcAGAAAAElQ TS5NaWNyb3NvZnQgTWFpbC5Ob3RlADEIAQSAAQAjAAAAUkU6IFNpbXVsYXRpbmcgSW5kaXZpZHVh bCBCZWhhdmlvcgCHDAEFgAMADgAAAM0HBAATAAEAOAA5AAYAYwEBIIADAA4AAADNBwQAEwABADcA CwAGADQBAQmAAQAhAAAAOEFGOEY5RDk0MEI4RDAxMUE2Q0Q0NDQ1NTM1NDAwMDAABwcBA5AGAAwM AAASAAAACwAjAAAAAAADACYAAAAAAAsAKQAAAAAAAwA2AAAAAABAADkAwB2zlFxMvAEeAHAAAQAA ACMAAABSRTogU2ltdWxhdGluZyBJbmRpdmlkdWFsIEJlaGF2aW9yAAACAXEAAQAAABYAAAABvExc lLPZ+fiLuEAR0KbNREVTVAAAAAAeAB4MAQAAAAUAAABTTVRQAAAAAB4AHwwBAAAAGAAAAHNtZUBv aWtvcy53YXJ3aWNrLmFjLnVrAAMABhDye3qMAwAHEIILAAAeAAgQAQAAAGUAAAAtLS0tLS0tLS0t RlJPTTpTVEVQSEVOQ1VQVE9OU01UUDpVUFRPTkBMQU5MR09WU0VOVDpGUklEQVksQVBSSUwxOCwx OTk3MDU6MDRUTzpTV0FSTS1NT0RFTExJTkdAU0FOVEFGAAAAAAIBCRABAAAAgwoAAH8KAAAKEgAA TFpGdXqe54r/AAoBDwIVAqgF6wKDAFAC8gkCAGNoCsBzZXQyNwYABsMCgzIDxQIAcHJCcRHic3Rl bQKDM3cC5AcTAoB9CoAIzwnZO/EWDzI1NQKACoENsQtg5G5nAdA1Nwr7FFEL8hpjAEAgCoUKi2xp MQQ4MALRaS0xNDTPDfAM0BzTC1kxNgqgA2D1E9BjBUAtHvcKhx2rDDD1HnZGA2E6H/4edgyCBgAJ E9BwaAnwIEMuIAZVBTACIFtTTVRQFDp1JHJAGPFsLmf4b3ZdH58grQZgAjAh34Mi6yGQaWRheSwU sA8TUAMRHFAqgDE5OTdiIBlAOjA0Jg8grVSebyhPIusD4ArAbS0EYmZsHDAZEEBzAHABkGY0ZS4J gHUr7ycedWJOah6xLg8i61JlNABTFQdwdQtgdDChIEluKGRpdipAdQdAIEI+ZRGANxAFsBr/HAMz NlcddxpFLWdoBAAgOyFhRiAJwTZwIHRoO7FkPiE8YDgPOR8aWCOxdmU9JFQnBCAecQJgE+AgZK8D cQuAL9AJ4G0EIHcwcakv0HVpE9BkO/BvO3HPOxAEkArAEXBpYzdRL+PZCoVhcB5xANBoJEA20Dgg bXlApjvAEXAgYoh1Y2sRwCBvZkER/mEv8BPQBcBDIAOgBaAxAb9A8QRgFhBAIArANoBjQHA+cwqF PAADkUTxBaBtcM51RtIRgAQgYnkT0AQg8UZRUkFNO3A24DvxBJC7P5A7IW5CUEBQN8F1O2G6ZwnB ZzZxAiAKhXI2ULExQCBIb0GAP4ByKoD/QfBBFQuAS5ICME9RROADEH9B8ArARQAwR0kSO+JLpWH7 CoVCnHRN4B7ACHA/kEtCPzuAEXFA8UZRSaEDgWQu+TycU28qgEuBTLEoETthexYQNABzBvA28ASQ BCAt9j5AIDZhbwIgWDJJogBw/0UAWEFM8QdxT8FYQAqFNvIvAJBYxDABSYJyFhBzcH8CIDbwUSJC UlSONAAcMGX9SeFuMPFYMgqFQyAFMEDiu1hBAMBqBbFY8xWgbjBw/1gyVwFCsSWgOvE/kF5ZXGDv MIAEIDsSCcB1VyM24AqFvwqwBBAHkVayKlABkCAlEP9WozCRP5BCQVayX6ViBF+mTxYQXGAAIGNU c3lMEHD/AJBIpkZRZTcDUmXlXnZlEf9mlGBUS0JXsEZAA6BmhWGV7UFwaFaACoVmbuJqxkdR+xjg QxB0KoBkxTbwFhAewH83AHDCTlBtUWbCVJ4a5ih/KqAVkVnwB5ECEAXARPFp5mdMEELAbmM/kEZR UCf/BbBNEAMAMOBNMiqAbmARwBtLkSRQS3YxJFBTISn9VY0iTkFE4EXwRcANsEeBNwMgOyEWEHFB 4BYRPyD9c7BNR+J51CqAezk7QmazzwhwT6JNdgdAbHk2kgUA6HkpIjycTUUAXGALgPVP0XM0AEF0 UUPyZRRt5v87IUdRdVAEoEISS5NXYUwC/wuANvdXtlWGTkiDv4ISNlDdQiBibFEwU0IRLUXRRjH/ JZBFAGqRB9AqgEtCSIEEIL9zVkkjANBT0H4iT1F2BvBfP4BCIEDxXRFv9SlYZnN/iUQ/cQOgiaNZ 1gQgEcBj/XM3V0uSSkFQuDuBbfaHM0d6dYwEE9B4dC0NsHDPCfANsE/BTeItYkpACYD/CoVpABPC NsAv0G5wh0IHcO9MwGZDUWFWlGYIcDwACHD/cZ9U9yqATeJIpoeASaGHkvNH8ZbAenpFAGPRe8JR dP+JkT+AAyBp9YXpWHUqgGERf2BBiNFIpkGDTJA/kBPAb/8RcRPAQxBa4Q3QT5MHMUWA53qQTSOE wCBPVPIIcBGw252ldOBoBUAKwGcKUFFE/wnwTRBXAI6UgzKRcFTBBcDfFaATwEwBBUCZsWGfImn1 Zid1AQDAbCda4VOBaf+IYU1IiGIxcGZnDcEesmn1/4kRbTNsQ1lwJEA2IHVCS5H9gBBzQJA7IUAy AaCIwkim/zBwTJBxUUsxFLCVEIIzBTHvSkEFADvBNwF0RQCVEJWqv1akB3B1c1DHL/FvEWUKhd87 IUfTA5GSsASQYwQAnGL7TdAEAGuBAWTxq9BPskkjv2PQBIETwEtBMKFVjUFZcP8v8CpxbDRA8YAE afOsAqURJzaSgGE8nCgxjLBTV/xBUksgSjMJ4XnBAJB08P9CEQPwPABCiXQyRlIzoC/j/09CUBFk NwdAPAAIYKKgiUH/SCJqpIjzuRZWkjshMTA2cP9T4gQAP/ClcjaSkrALUIAQ+0IBJEBQBJARgGlA SLdvwv9Jwk0jN1El0MPhPDFKs2PB/wMAuXKF+b4Uu1dSNqcRqzL/muE/gEShagI70XzGk/GY0X9j IW0yCrF+IWExL+M8jSj6MoywSkyQpXi69GjSQeT5dHJBQksgQLAHkaVSE1B/BZAKQA2wnDXRIVsy aYhP9ERFoWFEXEFB8UTiC2D/pSG5IncCQxBFwJ4CN6GfQf+/oFcAE8FLYjvRlRAv8Eimv5NBdOBC wDaDB4AAcC0coP8wcEIgULhoE1xQqWXQISqA75UQnyLS8k/jeQqFMEMkQPd/0XswMGFTL/JBETbg O2L/E8Fl0AeBMOBXAEI03KTFQX9m0dQBZRI2cGRWYyNE8UzdopJTCrB1UCqAT3VQAHCT4ZNLM1Bo SoBvUBjx3mskgUwxM8J0VHYKwAcwm0BhZApSY9BXAC1LSeCfAZBAAUuBQJB0cjI0QoD/odJCMh5x MXB1UVayYSCSwP1lYXlAAbXBE9DcUUfiRlF/AHAKhTbQHKADANOh30RNV0RxleWL8kYkwCEa7Sj+ M4ywgKBB0L6hNrBCmwWw9wSBS3KL8iJB0FwgdBDfYt/DIQZxWhJDMhOzIjtCQ7Yhj6BFTEwt0wBG SfhORUR5ILvhBQOtkjBD//JQafWbtAGgxdEkQESyPED/NvB2xFayCoXwavH2enU7Yn/z//UPmUcV oAfgQkF2onP+ZkUAZrMFEEbRBzBp49XhvwGgUDJFALxzO7GRUnKHUf9lciRAN4B0cbI2pxFEITvA fwUQNqGXR25wQHC8ZDbwYfsHgEbSPcPAGPFGAEABQHD/GRC8kQK0kxCWwPXxsNPESf+zaGvjDbGV 5D+QdSHV0Wnmf8wCA8NSEJ8hi8P1fTGVLv8MIDtARmCIo+fTjbIpsNgf/1gCarRj0FEUwtCUhmvm cQD7UkYB8XRAAZJzZ1AMMVZw231QKnFbA5GnQmXcojQAkE9GRl1VjUV4fTD/nzFsM61xGRBKszuA MiDEIGlCEVR5kxAg51DR8HL/RhCoJK4Sh0HCglwxpyG4xP88BB7gUkZ6AjHAjwB5sWkQ/wOwQxCN QUuDQAGlUTaSPdD/RhCtwotRSkBbQv5zfVBB8Pt30EUAKHQwS3LblnoBMKHeLT3QECCMsEJBYgIT 54D/osEv0EnQwCFGUT9wV8Cng3/cUSpgh4FO0daxsyTxRiH2P1WdP2NFQVBAcCpwFLa/XAAQBrcw zDAQEDshJuvRc16wo9QgRzMgZcBullXfxtBOgbwAr2FGUVcx0ANAf0YA2nB3oFWGOZ8zNDGVfQUx kAAtkAADABAQAAAAAAMAERABAAAAQAAHMKBpxlVcTLwBQAAIMKBpxlVcTLwBHgA9AAEAAAAFAAAA UkU6IAAAAAAbcQ== ------ =_NextPart_000_01BC4C65.3E212A20-- ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 19 05:23:10 1997 From: swarm-modelling@santafe.edu (Scott Christley) Date: Fri, 18 Apr 1997 21:23:10 -0700 (PDT) Subject: Simulating Individual Behavior In-Reply-To: <01BC4C65.3E212A20@p62.ascend3.is2.u-net.net> Message-ID: Fascinating! Now that its the weekend; I have the time to step out of my role of striving to make GNUstep a reality; I wish I had such a discussion group when I was in graduate school...anyways let me pose some questions. first a metaphysical one: I am tending to notice a separation between two modelling paradigms: the ABMers (I was an IBMer once! ;-) and the ODEers. Are there true differences between these two paradigms or is it only a perception? Meaning is not a Swarm program a symbolic (mathematical) description of a model, just not as concise as a differential equation? Steve Emsley was talking about military simulation... Has anybody read the recent Wired article about the Marine's use of DOOM for battle simultion? Any comments about the merits of using VR in simulations besides the obvious: scenario iteration (as in responding to a terrorist threat)? Does anybody(or know of anybody) have interest in seeing Swarm expand in VR areas (versus 3D visualization)? Continuing with military simulation; is such simulation geared at controlling and predicting our own country's troups? Or is it more related to training(as in DOOM above) to prepare our troups for possible unknowns? From my understanding of VR use in the military its towards the later, but concentration on the former seems to be more of the theoretical nature. And the iteration I didn't mention where simulation is to geared towards acting out the opposing force. Imagine this scenario: You have some terrorists who have taken over a building and are holding a bunch of hostages. You have a small number of marines who are going to attempt to overtake them. It would seem to me that being able to simulate the most realistic terrorist is going to be both the hardest and most beneficial part of the simulation. Its simulations like this where individual actions have a huge impact upon the outcome. Did I actually read someone mention Planck's constant?!! :-) I'm currently reading a biography of Einstein by Albrecht Folsing (good book!); it got me thinking about how as humans we attempt to find the "parameters" so that we can "model" ourselves after other humans that we idolize. It makes me wonder if the search for the fundamental laws of the universe have progressed from physics to the higher level sciences of biology, sociology, history. excuse my musings Scott ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 19 07:36:35 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Fri, 18 Apr 1997 23:36:35 -0700 Subject: Simulating Individual Behavior References: Message-ID: <335867F3.63414CEB@ix.netcom.com> Scott Christley wrote: > > I am tending to notice a separation between two modelling paradigms: the > ABMers (I was an IBMer once! ;-) and the ODEers. Are there true > differences between these two paradigms or is it only a perception? > Meaning is not a Swarm program a symbolic (mathematical) description of a > model, just not as concise as a differential equation? I don't see it that way at all. Obviously, both an ABM and an ODE model are abstractions of our observations of behavior on the ground. The abstractions made in the two paradigms are different, of course. In ABM's, the abstractions made are informed directly by observation. There's ants, so we model ants. They cut up leaves and carry them around, so we have leaves and pieces and leaf-cutting behavior in our model. The more we know, the more constraints we can build into our ABM. There is no ODE system (no matter how observationally vacuous) that cannot be couched in terms of an ABM, unless we don't really know what it is our model is doing (in which case its uses are limited to crunching out trajectories). In ODE-based and PDE-based models, the abstractions made are the ones that are imposed by the kind of symbolic manipulation (analytic solution) that is possible with simple ODE and very simple PDE systems. In other words: calculus tells us the function has to be continuous, so suddenly we are forced into an abstraction of our population of wildebeests such that population size is a real number and population growth is a continuous function. Biologically (or sociologically, or whatever), we tend to remain rather unconvinced that either one of these assumptions is particularly realistic. ODE's and PDE's were invented so that problems could be solved analytically. But few interesting problems that we'd model with these formalisms are soluable analytically anyway, so there's no longer any good reason to use them, and one very good reason not to use them: they force on us an abstraction that serves merely a by-gone purpose and which is usually not warranted in the biology (or sociology). [As long as you're just doing some simple thumbnail models of biomass and energy balances and what-not, you might be safe up to a point just doing your ODE's, of course. I don't want to deny that.] Now I can say what I think the answers are to the questions you pose above. Are there true differences between these two paradigms? Yes, because ABM's don't force abstractions on us that are artifacts of a method whose day is past and which are not desired otherwise in our models. Is a Swarm program just a less concise description of a the same thing a differential equation describes? No, certainly not. A Swarm program does not normally describe the effects of fractional wildebeests pairing up and producing fractional young. It describes real, whole wildebeests pairing up and producing real, whole young. If anything, a Swarm model is _more_ concise. Swarm: "How many parents did this baby gnu have? Exactly two." ODE: "How many parents does every baby gnu have? 1.998785622342179 +/- 5% at the instant of conception, with a first-derivative slope of +0.73." That's my two gnus' worth... > It makes me wonder if the search for the fundamental laws of > the universe have progressed from physics to the higher level sciences of > biology, sociology, history. I certainly hope not. I would have hoped that the experience of physicists over the last 100 years would have taught biologists, sociologists and historians not to look for fundamental laws of the universe, but rather for theories that help understand the universe. :) -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sat Apr 19 08:04:22 1997 From: swarm-modelling@santafe.edu (Theodore C. Belding) Date: Sat, 19 Apr 1997 03:04:22 -0400 Subject: Simulating Individual Behavior In-Reply-To: References: <01BC4C65.3E212A20@p62.ascend3.is2.u-net.net> Message-ID: Hi Scott- At 9:23 PM -0700 4/18/97, Scott Christley wrote: >I am tending to notice a separation between two modelling paradigms: the >ABMers (I was an IBMer once! ;-) and the ODEers. Are there true >differences between these two paradigms or is it only a perception? >Meaning is not a Swarm program a symbolic (mathematical) description of a >model, just not as concise as a differential equation? Using ODEs to model an agent-based model like Conway's Game of Life may be theoretically possible, but it's often a very bad representation. It's like using a Fourier series of sine waves to represent a square (pulse) waveform. The Fourier series representation needs a huge number of terms to even approach the accuracy that another type of representation would have (say step functions). On the other hand, using step functions to represent a sine wave is not the best representation. Doyne Farmer and Norman Packard, as well as John Holland, have written about the need for using an appropriate, efficient representation of a dynamical system. For something like the Game of Life, I think the most appropriate representation is in term of agents, not ODEs. If we agree that in some cases we want to use ODEs and in others we want to use agent-based models to model a system, then I think we also have to recognize that each modeling paradigm has different tools that are needed to work with them. A lot can be determined about ODEs using traditional mathematical techniques, as well as numerical simulation techniques. In agent-based models we don't have many mathematical tools at present and usually must rely much more on computer simulation of the agents' behavior. Using ODE tools to analyze an agent-based model (or vice-versa) is like using a hammer to pound a screw in. In this sense, an agent-based model is very different from an ODE, even though they're both mathematical formulations at some level. -Ted -- Ted Belding University of Michigan Program for the Study of Complex Systems ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Sun Apr 20 15:19:36 1997 From: swarm-modelling@santafe.edu (Martina Schretzenmayr) Date: Sun, 20 Apr 1997 16:19:36 +0200 Subject: Simulating Individual Behavior/ ZIP codes Message-ID: >Jan Kreft wrote: >> >> Mark P. Line wrote: >> >> > Helmut Vetterle (1986) _Konstruktion und Simulation mikroanalytischer >> > Modelle. Die Methode der Mikrosimulation und ihre Anwendung_ ("Design >> > and Simulation of Microanalytical Models: Method and Application of >> > Microsimulation"). Augsburg: MaroVerlag. >150 refs. >> >> That would be fine! Would you make that diss available? I, personally, >> don't need a translation but most people will :-(. > >In the short term, I'm willing to loan the book out if it does happen to >be out of print. Check your bookstore to see, and send me your snail >address if you want me to loan you my copy. > > >The address of the publisher given in the book is > > MaroVerlag, Benno Ka"smayr > Riedingerstr. 24 > [new zipcode?] Augsburg ----------- The new ZIP Code is 86153 ----------- phone ++49/821/416033 > >The copyright for the book is held by > > Beratungsgesellschaft fu"r angewandte Systemforschung mbH - Augsburg > Haunstetterstr. 19 > [new zipcode?] Augsburg > (phone +49-821-571093) ----------- The new ZIP Code is 86161 > >The ISBN is 3-87512-501-0. > Augsburg is my home town. Very nice and very old (about 2000 years) city. If you ever can visit ... Martina Schretzenmayr ------------------------------------------------------------------------------ Martina Schretzenmayr Dipl.-Geogr. Raumplanerin ETH/NDS Institut fuer Orts-, Regional und Landesplanung (ORL-Institut)/ Institute for Local, Regional and National Planning Eidgenoessische Technische Hochschule Zuerich / Swiss Federal Institute of Technology Zurich ETH Hoenggerberg HIL H 41.3 CH - 8093 Zuerich Tel.: ++41 - 1 - 633 29 47 Fax: ++41 - 1 - 633 10 98 e-mail: Schretzenmayr@orl.arch.ethz.ch HomePage: http://www.orl.arch.ethz.ch/~Schretzenmayr/index.html ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 21 15:24:22 1997 From: swarm-modelling@santafe.edu (Mattias V. Bertelsen) Date: Mon, 21 Apr 1997 09:24:22 -0500 (CDT) Subject: Simulating Individual Behavior In-Reply-To: <335867F3.63414CEB@ix.netcom.com> Message-ID: I have a question: Why would a model need to be either ABM/IBM or ODE/PDE? It would seem that a model which even approached natural complexity would be extremely difficult to construct or analyze if one held strictly to one paradigm or the other. I think that Swarm lends itself quite handily to constructing models where the framework is agent-based, but the behavior of the agents themselves can be modeled using differential equations (for cases where the continuity assumption is justified), statistical models, or explicit behavioral descriptions. As soon as finals are done with, I am hoping to get started on my "summer coding project": A Swarm object using multi-layer perceptron code (neural networks) that functions as a black-box statistical model. You give it an input file of data and train a network to fit a statistical distribution. The object can then be used along the lines of the current distribution objects in Swarm--hook up a pseudorandom number generator, and pull numbers out of a distribution which came from real data. I could see this being used, for example, as a "weather object": the intervals between rain events in a simulated prairie ecosystem are reasonably modeled without having to model butterflies flapping their wings. Any thoughts? By the way--I am looking for a graduate program in CS or Ecology to work with this stuff, starting around the fall of '98. Any suggestions? Mattias V. Bertelsen mattias@spaceship.com http://www.spaceship.com/~mattias On Fri, 18 Apr 1997, Mark P. Line wrote: > Scott Christley wrote: > > > > I am tending to notice a separation between two modelling paradigms: the > > ABMers (I was an IBMer once! ;-) and the ODEers. Are there true > > differences between these two paradigms or is it only a perception? > > Meaning is not a Swarm program a symbolic (mathematical) description of a > > model, just not as concise as a differential equation? > > In ODE-based and PDE-based models, the abstractions made are the ones > that are imposed by the kind of symbolic manipulation (analytic > solution) that is possible with simple ODE and very simple PDE systems. > In other words: calculus tells us the function has to be continuous, so > suddenly we are forced into an abstraction of our population of > wildebeests such that population size is a real number and population > growth is a continuous function. Biologically (or sociologically, or > whatever), we tend to remain rather unconvinced that either one of these > assumptions is particularly realistic. ODE's and PDE's were invented so > that problems could be solved analytically. But few interesting problems > that we'd model with these formalisms are soluable analytically anyway, > so there's no longer any good reason to use them, and one very good > reason not to use them: they force on us an abstraction that serves > merely a by-gone purpose and which is usually not warranted in the > biology (or sociology). > > [As long as you're just doing some simple thumbnail models of biomass > and energy balances and what-not, you might be safe up to a point just > doing your ODE's, of course. I don't want to deny that.] > > Now I can say what I think the answers are to the questions you pose > above. > > Are there true differences between these two paradigms? Yes, because > ABM's don't force abstractions on us that are artifacts of a method > whose day is past and which are not desired otherwise in our models. > > Is a Swarm program just a less concise description of a the same thing a > differential equation describes? No, certainly not. A Swarm program does > not normally describe the effects of fractional wildebeests pairing up > and producing fractional young. It describes real, whole wildebeests > pairing up and producing real, whole young. If anything, a Swarm model > is _more_ concise. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Mon Apr 21 17:17:51 1997 From: swarm-modelling@santafe.edu (alan penn) Date: Mon, 21 Apr 1997 16:17:51 +0000 Subject: calibration Message-ID: <43859.9704211611@bas-a.bcc.ac.uk> To throw in twopenny worth, >> Perhaps you are talking about ecosystems from your simulation-oriented >> point of view, as the word 'calibrate' indicates. > >Perhaps I am. > >> I think that a complex >> system like an ecosystem cannot be modeled with some large FSM or even >> with a set of stochastic equations (except in the short-term): in an >> ecosystem there is probably no unique latent 'model' to be discovered, >> against which we could compare some data and make adjustments. Instead >> there probably exists many possible modes, with the system unpredictably >> switching from time to time to one or another. > >I am not sure that we could argue this one way or the other. I see >no reason why a model could not be developed. When you say "many >possible modes", this can easily be modeled by stepping up a level >of abstraction so that the modes are modeled as well. > >> The first thing to do could be to try to recognize these modes. > >Yes, as part of our systems identification process. > >> Because >> switching is unpredictable (ecosystems seem to exhibit self-organized >> critical modes), it makes no sense to try to 'calibrate' the whole system, >> in the same manner simulation engineers calibrate a model of a factory or a >> flexible workshop. Try this one. Perhaps the 'calibration' is in the environment. What I mean by this is that the configuration of the spatial environment inhabited by moving and stationary individuals in the ecosystem could in principle carry with it the 'model' (or several models) inherent in patterns of co-occupancy of space. Think, for example, about human systems like buildings or cities. Modern societies in Levi-Strauss 'statistical' sense, consist in a set of probabilities describing regularities - eg: within race marriages are more likely than inter racial marriages, but there is no rule system prohibiting them as there might have been in more primitive 'mechanical' social forms. How could such probabilistic regularities emerge and be maintained? One possibility is that they arise through the way that spatial configuration and movement of individuals bring different groups into contact. Most geographers treat space as 'map space' in that it is essentially open and homogeneous in all directions, but as soon as you start to think about the space through which we actually move, within and between buildings, then the probabilities of co-presence become significantly structured. If you then map onto this sort of spatial configuration the locations of particular individual's or social groups' facilities, the places they must visit regularly in their daily lives, then you can create a structured, but essentially probabilitic set of interfaces. You could envisage a single spatial configuration coupled to multiple sets of group's 'programmes' giving rise to just the sort of latent model with multiple realisations that complex things like societies require. A final point - perhaps it is the mapping of spatial configuration and cultural 'programmes' - which are actually constructed by people after all - that provides the locus for reproduction of the regularities that we call social?? Alan Penn ________________________________________________ Alan Penn The Bartlett School of Architecture and Planning Philips House (Room 335) University College London, Gower Street, London WC1E 6BT tel. (+44) (0)171 387 7050 ext 5919 fax. (+44) (0)171 916 1887 email. a.penn@ucl.ac.uk www. http://doric.bart.ucl.ac.uk/web/Pangea/index.html ________________________________________________ ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 15:31:11 1997 From: swarm-modelling@santafe.edu (Philippe LAVAL) Date: Tue, 22 Apr 1997 13:31:11 -0100 Subject: calibration Message-ID: <3.0.32.19970422133110.006d648c@poseidon.obs-vlfr.fr> At 16:17 21.04.97 +0000, Alan Penn wrote: >[...] >Try this one. Perhaps the 'calibration' is in the environment. What I mean >by this is that the configuration of the spatial environment inhabited by >moving and stationary individuals in the ecosystem could in principle carry >with it the 'model' (or several models) inherent in patterns of >co-occupancy of space. >[...] > >Most geographers treat space as 'map space' in that it is essentially open >and homogeneous in all directions, but as soon as you start to think about >the space through which we actually move, within and between buildings, >then the probabilities of co-presence become significantly structured. If >you then map onto this sort of spatial configuration the locations of >particular individual's or social groups' facilities, the places they must >visit regularly in their daily lives, then you can create a structured, but >essentially probabilitic set of interfaces. You could envisage a single >spatial configuration coupled to multiple sets of group's 'programmes' >giving rise to just the sort of latent model with multiple realisations >that complex things like societies require. > >Alan Penn > I like your idea of space being structured by the individuals. In " The representation of space in an object-oriented computational pelagic ecosystem", Ecol. Modelling, 88:113-124 (1996), I tried to model space as a shared resource accessed concurrently by individuals, in a first come, first served basis. Here it was food that structured space. This approach was probably adequate because the individuals were identical filter-feeders competing for a resource distributed unevenly in (geographical) space. But if there are different kinds of individuals, it makes sense to consider constraints in the accessibility of space: stronger individuals (or individuals with higher capabilities) have access to a larger range of geographic positions, so they may exploit more resources. Of course resource distribution still shapes the space, in the sense that where a resource is absent, even strong individuals get nothing. But the superposition of different 'ambits' in the same geographic space may better model the competition, or coexistence, of individuals in space. -------------------------------------------------------------------- Philippe Laval Station zoologique B.P. 28 - 06234 Villefranche-sur-Mer CEDEX (France) laval@ccrv.obs-vlfr.fr ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 19:06:03 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Tue, 22 Apr 1997 12:06:03 -0600 Subject: parallelism! In-Reply-To: <199704221748.KAA15917@linus.net-community.com> References: <199704221748.KAA15917@linus.net-community.com> Message-ID: <199704221806.MAA21765@grasshopper.santafe.edu> Scott Christley writes: > I'm trying but I haven't quite figured out what MPI stands for (multiple > process interface?). Message Passing Interface, I believe. > Its also unclear to me, probably because my understanding of Swarm is > minimal now, but what type of parallel models are you allowing/using? > > Local memory and message passing with no shared memory? > > What network topologies are you expecting to support? > > What type of memory models are you expecting to support? Ideally, we would use pure message passing with now shared memory. This should allow us to spread swarms over several independent machines in a transparent network. But, we will have to have one hub machine that should synchronize all the processes on the other machines. However, we may use some type of virtual shared memory. > Is the user expected to perform the problem decomposition and Swarm will > handle the distribution; is the user expected to specify both, or is Swarm > going to be able to perform both based upon common model "skeletons" or types. Again, ideally, the user won't have to think about distributing objects (in reality, of course, she will [grin]). The idea is to allow the user to program a model without thinking too much about programming the computer[s]. > I also see a different viewpoint when you discuss parallelism-1; which is > parallelism not of a single program, but of running many concurrent > simulations. Presumably this is important because a researcher may perform > 1000 runs of a simulation and perform some statistical analysis on the results. First off, none of this is true "parallelism". That's a complete misnomer. It's really concurrency (in the abstract) and distributed computing. So, if we just accept the usage of the word "parallel" to mean "doing two or more things at the same time", then we're ok. [grin] But, semantics aside, ||-1 is important to a small extent in allowing a user to run a bunch of well-specified independent Swarms at the same time. The only thing this really saves is time. (There's a continuum between ||-1 and ||-2 and somewhere in between the two, we could make use of the data being generated by any one indep. swarm and make decisions to cancel some swarms as dead-ends in a search or whatnot...) And it will help us work an implementation of MPI into Swarm gently. Other than that it's pretty trivial. glen ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 19:34:10 1997 From: swarm-modelling@santafe.edu (Ginger Booth) Date: Tue, 22 Apr 97 14:34:10 EDT Subject: hex lattice2d In-Reply-To: <199704221806.MAA21765@grasshopper.santafe.edu>; from "glen e. p. ropella" at Apr 22, 97 12:06 (noon) Message-ID: <9704221843.AA20545@sfi.santafe.edu> Hi, gang, Did anyone ever do a 2d hex lattice implementation? Cheers, Ginger ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 20:19:39 1997 From: swarm-modelling@santafe.edu (Theodore C. Belding) Date: Tue, 22 Apr 1997 15:19:39 -0400 (EDT) Subject: parallelism! In-Reply-To: <199704221806.MAA21765@grasshopper.santafe.edu> Message-ID: I hope that none of this takes any resources from documentation and writing the manual; that's what's really needed right now. ||-Swarm is cool, but it can wait. -Ted -- Ted Belding University of Michigan Program for the Study of Complex Systems ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Tue Apr 22 20:39:04 1997 From: swarm-modelling@santafe.edu (Barry McMullin) Date: Tue, 22 Apr 1997 13:39:04 -0600 Subject: hex lattice2d In-Reply-To: <9704221843.AA20545@sfi.santafe.edu> References: <199704221806.MAA21765@grasshopper.santafe.edu> <9704221843.AA20545@sfi.santafe.edu> Message-ID: <199704221939.NAA03018@tsankawi.santafe.edu> Ginger Booth writes: > Did anyone ever do a 2d hex lattice implementation? Hmmm ... it all depends, of course. But my tipsybugs app is still in the archives at: ftp://ftp.santafe.edu/pub/swarm/users-contrib/anarchy/tipsybugs-0.05.tar.gz It doesn't quite have a hex "lattice" as such, but rather a generic lattice that supports "coord" objects; these coord objects have functionality for getting at "neighboring" coord objects, where the "neighboring" relationship can be flipped between von Neumann and Moore on a square tiling, and FHP (effectively hexagonal) on a triangular tiling. Enjoy, Barry. -- ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ | Barry McMullin, ALife Group, | McMullin@santafe.edu | | Santa Fe Institute, 1399 Hyde Park Road, | Voice: +1-505-984-8800 | | Santa Fe, NM 87501, USA. | FAX: +1-505-982-0565 | | http://www.eeng.dcu.ie/~mcmullin | | ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 13:43:42 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Wed, 23 Apr 1997 06:43:42 -0600 Subject: forwarded from Rosaria Conte (Soc. Sim. conf) Message-ID: <199704231243.GAA00205@seamus.dischordia.com> ------- start of forwarded message (RFC 934 encapsulation) ------- Return-Path: Received: from naga.mailbase.ac.uk by kc.trail.com with smtp (Linux Smail3.1.29.1 #3) id m0wJbfS-000JlQC; Tue, 22 Apr 97 03:15 MDT Received: by naga.mailbase.ac.uk id (8.7.x for naga.mailbase.ac.uk); Tue, 22 Apr 1997 09:33:31 +0100 (BST) Received: from pscs2.irmkant.rm.cnr.it by naga.mailbase.ac.uk id (8.7.x for naga.mailbase.ac.uk) with SMTP; Tue, 22 Apr 1997 09:32:48 +0100 (BST) Received: from [150.146.7.121] (conte.irmkant.rm.cnr.it) by pscs2.irmkant.rm.cnr.it (4.1/1.34) id AA07227; Tue, 22 Apr 97 10:27:40 +0200 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-List: simsoc@mailbase.ac.uk X-Unsub: To leave, send text 'leave simsoc' to mailbase@mailbase.ac.uk Reply-To: rosaria@pscs2.irmkant.rm.cnr.it (Rosaria Conte) Precedence: list From: rosaria@pscs2.irmkant.rm.cnr.it (Rosaria Conte) Sender: simsoc-request@mailbase.ac.uk To: simsoc@mailbase.ac.uk Subject: ICCS&SS Call for Participation Date: Tue, 22 Apr 1997 10:31:03 +0200 Dear Colleague, for information concerning the "International Conference on Computer Simulation and the Social Sciences" (ICCS&SS) to be held in Cortona (AR), Italy, 22-25 September 1997, please visit the following site http://pscs2.irmkant.rm.cnr.it/users/rosaria/CallPart.html Thanks for your attention rosaria conte Rosaria Conte, Division of AI, Cognitive and Interaction Modelling - PSS (Project on Social Simulation) IP/Cnr, V.LE Marx 15 - 00137 Roma voice: +39+6+86090210 fax: +39+6+86090214 email: rosaria@pscs2.irmkant.rm.cnr.it http://pscs2.irmkant.rm.cnr.it/users/rosaria/home.html ------- end ------- ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 05:49:06 1997 From: swarm-modelling@santafe.edu (Steve Emsley) Date: Wed, 23 Apr 1997 05:49:06 +0100 Subject: ABM/I(C?)BM In-Reply-To: (donalson@lifesci.lscf.ucsb.edu) Message-ID: <199704230449.FAA00801@archon> Perhaps I'm tagging this onto the wrong thread but Doug Donalson refers to a reference that I had in mind as I brought up the ODE / ABM dichotomy. Chapter 3 of DeAngelis' book on IBM (referred to by Doug) on IBM is by Caswell & Meredith John. In this they differentiate between the i-state configuration model (individual-based with individual differences), the p-state (a model of the population as a mean of the i-state configuration) and the i-state distribution ( a simplification of the p-state arising when all individual's experience the same environment). ODE's are justifiably used to model the p-state (including the special case of the i-state distribution). However, the authors suggest that the i-state configuration is important with "complicated i-states, small populations and local interactions". In addition, the p-state is derivable from the i-state whereas the i-state not derivable from the p-state. As a parting thought they suggest that "i-state configuration models may be useful ... to drive maximum likelihood parameter values [of i-state distribution models]". Personally, I view this distinction as having more mileage than an argument on the relative merits of ODE/PDE vrs. ABM(IBM) models. The latter being a set of tools to understand/describe/model the former. One posting suggested that ODEs are attractive due to the possibility of their analytical solution. IMHO if you have an ODE-based ecology capable of analytical solution (rather than numerical simulation) you are dealing with a mathematical abstraction (ecology as an excitable medium) NOT a system that propagates through time based on local interactions and continually varying stochastic or adaptive parameters i.e. a real ecosystem. It is not unusual for ODE/PDE models to be "tuned" to actual data though modifying the closure terms. Better, in my opinion, to turn one's back on parsimony and computational efficiency in order to model the i-state. If there's no significant difference between the i-state configuration (IBM) model and the p-state (ODE/PDE) model then efficiency favours the latter. In my field the ergonomics and economics of sampling skew the available data towards coarse resolution. To fit a multi-parameter mean-field model to such data could, unsympathetically I admit, be called tautologous. By my, naive, appraisal of current 'scientific method' I see modelling driving observation - a reversal of the traditional paradigm. To assume a mean-field solution from the onset establishes a priori length scales and time scales to the system under investigation which, a posteriori, can establish observational criteria. I suppose that the impact of IBMs on conventional mean-field modelling may be analogous to the impact of Kepler's First Law on the aesthetic bias of the Platonic circular orbit - as observations become more exact then no system of epicycles (or parameters) will exactly fit the facts. Regards -- -- Steve Emsley sme@oikos.warwick.ac.uk ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 18:15:16 1997 From: swarm-modelling@santafe.edu (glen e. p. ropella) Date: Wed, 23 Apr 1997 11:15:16 -0600 Subject: forwarded from Michael Lissack (boston CS conf.) Message-ID: <199704231715.LAA23265@grasshopper.santafe.edu> ------- start of forwarded message (RFC 934 encapsulation) ------- Reply-To: lissack@lissack.com Organization: Michael Lissack Comments: To: complex@lissack.com From: Michael Lissack Sender: Complexity and Management Mailing List To: COMPLEX@LISSACK.SPACELAB.NET Subject: [Fwd: Complex Systems Conference and Book] Date: Tue, 22 Apr 1997 17:56:35 -0400 ---------------------forward from Sean Pidgeon--------------------- Date: Tue, 22 Apr 1997 12:52:55 -0400 From: Sean Pidgeon Subject: Complex Systems Conference and Book To: lissack@lissack.com Message-id: MIME-version: 1.0 X-Mailer: Novell GroupWise 4.1 Content-type: text/plain Content-disposition: inline Content-transfer-encoding: 7BIT Dear Dr Lissack: I have appended below an announcement for a multidisciplinary conference on complex systems, to be chaired by Prof. Yaneer Bar-Yam of Boston University under the auspices of the New England Complex Systems Institute. The meeting has been organized in partnership with Oxford University Press. A book provisionally entitled "Complex Systems: A Multidisciplinary Sourcebook", to be edited by Dr. Bar-Yam and published by OUP, will draw upon the multidisciplinary themes and participants in the conference. The conference will, we hope, attract participation from many areas of scientific research. I would be most grateful if you could forward to me or to Dr. Bar-Yam (yaneer@bu.edu) any names (or mailing lists) of other researchers who might be interested in attending or participating in the meeting. We are particularly interested in knowing which individuals can speak or write about significant contributions to our understanding of the universal properties of complex systems -- e.g. as articulated in the themes section of the conference announcement -- and their application across disciplinary boundaries. Many thanks. - ---------------------------------------------- Sean Pidgeon Senior Editor Oxford University Press 198 Madison Avenue New York, NY 10016 phone: (212) 726-6134 fax: (212) 726-6445 sdp@oup-usa.org http://www.oup-usa.org/acadref/sdp.html phone: (212) 726-6134 fax: (212) 726-6445 sdp@oup-usa.org http://www.oup-usa.org/acadref/sdp.html - --------------------------------------------- First Announcement International Conference on COMPLEX SYSTEMS Boston Area September 21-26, 1997 Host: New England Complex Systems Institute http://necsi.org necsi@necsi.org With: Oxford University Press Conference Chairman: Yaneer Bar-Yam ORGANIZING COMMITTEE: Philip Anderson - Princeton University Kenneth J. Arrow - Stanford University Per Bak - Niels Bohr Institute Charles H. Bennett - IBM William A. Brock - University of Wisconsin Charles R. Cantor - Boston University Noam A. Chomsky - MIT Leon Cooper - Brown University Daniel Dennett - Tufts University Irving Epstein - Brandeis University Michael S. Gazzaniga - Dartmouth College William Gelbart - Harvard University Murray Gell-Mann - CalTech / Santa Fe Institute Pierre-Gilles de Gennes - ESPCI Stephen Grossberg - Boston University Michael Hammer - Hammer & Co John Holland - University of Michigan John Hopfield - Princeton University Jerome Kagan - Harvard University Stuart A. Kauffman - Santa Fe Institute Chris Langton - Santa Fe Institute Richard C. Lewontin - Harvard University Andrew W. Lo - MIT Marvin Minsky - MIT Alan Perelson - Los Alamos National Lab Herbert A. Simon - Carnegie Mellon University Temple F. Smith - Boston University H. Eugene Stanley - Boston University James H. Stock - Harvard University Gerald J. Sussman - MIT Edward O. Wilson - Harvard University SUBJECT AREAS: UNIFYING THEMES IN COMPLEX SYSTEMS Sessions will be structured around both themes and systems. The themes are: EMERGENCE, STRUCTURE AND FUNCTION: substructure; the relationship of component to collective behavior; the relationship of internal structure to external influence. INFORMATICS: structuring, storing, accessing, and distributing information describing complex systems. COMPLEXITY: characterizing the amount of information necessary to describe complex systems, and the dynamics of this information. DYNAMICS: time series analysis and prediction, chaos, temporal correlations, the time scale of dynamic processes. SELF-ORGANIZATION: Evolution, development and adaptation. The system categories are: FUNDAMENTALS OF COMPLEX SYSTEMS: Complexity, emergence, chaos, fractals, non-equilibrium processes, dynamic scaling, information and computation in physical systems. MOLECULAR SYSTEMS: Chemical dynamics, complex fluids, molecular self-organization, membranes, protein and DNA folding, bio-molecular informatics. CELLULAR SYSTEMS: Cellular response and communication, genetic regulation, gene-cytoplasm interactions, development, cellular differentiation, primitive multicellular organisms, the immune system. PHYSIOLOGICAL SYSTEMS: Nervous system, neuro-muscular control, neural network models of brain, cognition, psychofunction, pattern recognition, man-machine interactions. HUMAN SOCIAL AND ECONOMIC SYSTEMS: Corporate and social structures, markets, the global economy, the Internet. PEDAGOGICAL SESSIONS: The conference will include pedagogical sessions on Sunday, Sept. 21. ** A detailed announcement including instructions for submission of abstracts will follow.** If you want to receive future announcements about this conference, please e-mail us at necsi@necsi.org Include: Your name: ________________________________________ Preferred e-mail address: _________________________ If you want to be removed from this list please send us a note including the statement: "Please remove my e-mail address from this list." and include the e-mail address from which you received this announcement. - --------------03B6B28B4FD021B0B3FAF09B-- ------- end ------- ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 18:38:52 1997 From: swarm-modelling@santafe.edu (Jan Kreft) Date: Wed, 23 Apr 1997 18:38:52 +0100 (BST) Subject: Pathology of discrete diffusion Message-ID: Hi swarmers, some time ago there was a discussion of pathological cases of diffusion among the list. One such case was brought to light by David Sumpter: > Hello everyone, > > 1. I am using a technique similar to Diffuse 2d for the diffusion of > heat and am slightly confused by one aspect. Consider a portion of > lattice with temperatures and first order diffusion: > > 0 > 080 > 0 > > If the diffusion constant is 1.0 and evaporation rate is 1.0 then the > centre value will update to 0 while the outside values will update to > 2. i.e. > > 2 > 202 > 2 > > This appears somewhat unnatural for heat equations. If the value 8 came > from some source, you would not expect the source to be colder than > the surroundings on the next time step........ > > Am I right about this? Is there a theoretical explanation? > > 2. Has anyone got a reference they can give me discussing heat > diffusion in terms of lattices? I'd be very grateful. > > Thanks, > > David. > I wonder if someone knows of other pathological cases and if there is a way to avoid such cases within a discrete diffusion frame. Has the above problem been solved or ignored? Any comments welcome. Cheers, Jan. ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 19:57:03 1997 From: swarm-modelling@santafe.edu (David Sumpter) Date: Wed, 23 Apr 1997 19:57:03 +0100 Subject: Pathology of discrete diffusion Message-ID: <199704231857.TAA07158@beehive.ma.umist.ac.uk> > 1. I am using a technique similar to Diffuse 2d for the diffusion of > heat and am slightly confused by one aspect. Consider a portion of > lattice with temperatures and first order diffusion: > > 0 > 080 > 0 > > If the diffusion constant is 1.0 and evaporation rate is 1.0 then the > centre value will update to 0 while the outside values will update to > 2. i.e. > > 2 > 202 > 2 > > This appears somewhat unnatural for heat equations. If the value 8 came > from some source, you would not expect the source to be colder than > the surroundings on the next time step........ The way to avoid such problems is to choose a correct value for the diffusion constant, we'll call D. This discretisation is the Laplacian Difference Equations for Diffusion. There are whole numerical analysis departments working on what the value of D should be for different problems. THe solution lies in the nature of the problem. It is smaller values of D which give more accurate results but obviously mean a larger diffusion time where diffusion time is proportional to (size of single lattice cell)^2 / D . The important thing is to view D as a constant determining the accruacy of your simulation and not to use it as an element of time. This is the trap I fell into at first. Instead of changing the diffusion constant to change the time for diffusion to take place, change the time steps at which agents update their position on the associated world lattice. Using this technique you can set the diffusion constant as low as you like, computing time allowing. In the end I just used D=0.5 so that I get, 1 141 1 Quite nice, I suppose. Good luck, David Sumpter ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 23:38:29 1997 From: swarm-modelling@santafe.edu (Chris Landauer) Date: Wed, 23 Apr 97 15:38:29 PDT Subject: correction Message-ID: <199704232238.PAA29770@antares.aero.org> the URL for the complex systems meeting is wrong on their announcement 8-) it should be http://www.necsi.org (they seem to have forgotten the www.) more later, cal ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Wed Apr 23 23:44:20 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Wed, 23 Apr 1997 15:44:20 -0700 Subject: ABM/I(C?)BM References: <199704230449.FAA00801@archon> Message-ID: <335E90C4.444A22CF@ix.netcom.com> Steve Emsley wrote: > > One posting suggested that ODEs are attractive due to the possibility > of their analytical solution. I don't remember anybody saying that, although I wrote that ODE's were invented in order to provide analytical solubility for otherwise intractable problems. I also wrote that no interesting problem formulated as ODE's or PDE's is likely to be analytically soluble, and that therefore their rationale as tools for analytical manipulation no longer exists. In saying that, I was trying to imply that, if we're going to have to attack our models numerically anyway, then we might as well go straight for an ABM representation instead of blissfully accepting the inappropriate assumptions (e.g. differentiability) of ODE's and PDE's. Those assumptions were the lesser evil when the problem at hand was otherwise intractable, period. That's no longer the case in domains where we can build useful ABM's. > IMHO if you have an ODE-based ecology > capable of analytical solution (rather than numerical simulation) you are > dealing with a mathematical abstraction (ecology as an excitable medium) > NOT a system that propagates through time based on local interactions and > continually varying stochastic or adaptive parameters i.e. a real > ecosystem. Precisely. It is unlikely that a useful [OP]DE model of an observed system will be capable of analytical solution. -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 01:14:31 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Wed, 23 Apr 1997 17:14:31 -0700 Subject: Pathology of discrete diffusion References: Message-ID: <335EA5E7.26DF6028@ix.netcom.com> Jan Kreft wrote: > > David Sumpter wrote: > > > > 1. I am using a technique similar to Diffuse 2d for the diffusion of > > heat and am slightly confused by one aspect. Consider a portion of > > lattice with temperatures and first order diffusion: > > > > 0 > > 080 > > 0 > > > > If the diffusion constant is 1.0 and evaporation rate is 1.0 then the > > centre value will update to 0 while the outside values will update to > > 2. i.e. > > > > 2 > > 202 > > 2 > > > > This appears somewhat unnatural for heat equations. If the value 8 came > > from some source, you would not expect the source to be colder than > > the surroundings on the next time step........ > > > > Am I right about this? Is there a theoretical explanation? I don't think I saw the original discussion here, so maybe there's something I'm missing. Fill me in if I'm not answering the question you're asking. Fourier's law says that the rate at which thermal energy flows between two points (or in the discrete case, between two adjacent lattice cells) is proportional (by a coefficient of thermal conductivity) to the negative gradient of temperature between the points (or cells). [Real diffusion involves both storage and flow, so you'll also have to deal with the material's thermal _capacity_, eventually.] For a large delta-t, I think you'll find that the gradient sometimes reverses sign and that temperatures damped-oscillate into the steady state. As you decrease delta-t, the model will equilibrate with fewer and more highly damped oscillations. In continuous time, thermal energy flows smoothly over the gradient until the steady state is reached (other things being equal). But even in continuous time, I wouldn't be surprised to find damped oscillations if you're starting from a thermally very heterogeneous body (hot or cold spots) and if the material has a high lambda (thermal conductivity). > > 2. Has anyone got a reference they can give me discussing heat > > diffusion in terms of lattices? I'd be very grateful. I haven't seen these for a while, but I think they have part of what you're looking for: Myrup, L.O. (1969) "A numerical model of the urban heat island", _Journal of Applied Meteorology_ 8: 908-918. Sellers, W.D. (1973) "A new global climate model", _Journal of Applied Meteorology_ 12: 241-254. Wierenga, P.J./de Wit, C.T. (1970) "Simulation of heat transfer in soils", _Proc. Soil Sci. Am._ 34: 845-848. But you can always go check the journal devoted to the topic: _Numerical Heat Transfer. Part A, Applications_ _Numerical Heat Transfer. Part B, Fundamentals_ and its precursor (same name, without the split). You could browse through all the article titles in the CARL database if you don't have easy access to the journals themselves. -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 03:19:57 1997 From: swarm-modelling@santafe.edu (Chris Landauer) Date: Wed, 23 Apr 97 19:19:57 PDT Subject: mathematical nonsense Message-ID: <199704240219.TAA02838@antares.aero.org> this one i can't pass up - the anti-intellectualism is stunning Mark P. Line wrote: I wrote that ODE's were invented in order to provide analytical solubility for otherwise intractable problems. I say: This is simply incorrect - ODE's were invented to provide analytic _models_ of complicated phenomena, usually in physics at first, and only some of them were ever analytically soluble. Mark P. Line wrote: I also wrote that no interesting problem formulated as ODE's or PDE's is likely to be analytically soluble, and that therefore their rationale as tools for analytical manipulation no longer exists. I say: The first part of this is correct, but the second part is not. There are many more analytic results that can be obtained from an ODE or PDE model than the solution. Indeed, many mathematicians have proved long term stability results for systems that have no hope of analytic solution. The power of an analytic formulation is _not_ about solutions; it is about understanding, and there are many different kinds of useful models. If you naively limit your understanding of mathematical analyses to "solutions", then you miss many of the most interesting and exciting results. Mark P. Line wrote: In saying that, I was trying to imply that, if we're going to have to attack our models numerically anyway, then we might as well go straight for an ABM representation instead of blissfully accepting the inappropriate assumptions (e.g. differentiability) of ODE's and PDE's. Those assumptions were the lesser evil when the problem at hand was otherwise intractable, period. That's no longer the case in domains where we can build useful ABM's. I say: There is a grain of truth in some of this, in that the mathematical assumptions often do not reflect the reality, but the same is true of the ABM models also, and anyone who forgets that is likely to get less than useful results. It also ignores the fact that many ODE and PDE models of complex phenomena are _much_ more easily analyzed (even without solutions) than the corresponding collections of individual actors. There are many people who think that modeling individual behavior in a complex system is somehow "less modeling; more reality", but that is too often just plain wrong, because the important behaviors are lost in the shuffle of complex interactions. The modeler must make choices either way; different choices make more sense for different purposes. For example, starting almost 100 years ago, Poincare and others studied long term stability of ODE's, such as those derived from the 3-body problem, and Fatou and Julia studied certain limiting sets of infinite processes in the complex plane, with no hint of an analytic "solution". My personal opinion is that the computational power available currently allows much modeling laziness, and while sometimes helpful (I have on many occasions used computer experimentation to give me insights about tricky problems), it actually very often inhibits clear and useful thinking about complex models. Don't throw away useful tools; we need all the help we can get. more later, cal Dr. Christopher Landauer National Systems Group, The Aerospace Corporation The Hallmark Building, Suite 187 13873 Park Center Road, Herndon, Virginia 20171 e-mail: cal@aero.org Phone: (703) 318-1666, FAX: (703) 318-5409 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 03:08:13 1997 From: swarm-modelling@santafe.edu (Mark P. Line) Date: Wed, 23 Apr 1997 19:08:13 -0700 Subject: mathematical nonsense References: <199704240219.TAA02838@antares.aero.org> Message-ID: <335EC08D.7B2DD48A@ix.netcom.com> Chris Landauer wrote: > > this one i can't pass up - the anti-intellectualism is stunning If I've succumbed to anti-intellectualism, then I guess I shouldn't be here. Bye. -- Mark (Mark P. Line -- Bellevue, Washington -- ) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 05:45:38 1997 From: swarm-modelling@santafe.edu (Doug Donalson;) Date: Wed, 23 Apr 1997 21:45:38 -0700 (PDT) Subject: mathematical nonsense In-Reply-To: <335EC08D.7B2DD48A@ix.netcom.com> Message-ID: I want to thank Mark and Chris for holding a mirror up to my face. I can't read the last two articals without feeling somewhat shagrined as I have had a part in a piece of this discussion that should never have started. There is no place in this group for name calling and trashing of others viewpoints. This includes my comment that the truth hurts (or some such stupid thing.) At least in the area of theoretical population ecology, people who have attempted to introduce spatially-explict individual-based models have been met (by some) with scepticisum and sometimes disdain. It is very easy to fall into the same trap and find all the faults of the more mainstream ODE type models. Engaging in trash talk does nothing but set back the real purpose (I hope?) of better understanding complex interactions. There is no point in rejecting potentially useful tools (ODE or ABM) either because of "not invented here" or "you trash me so I'll trash you". At UCSB we are using analytical or numerical solutions of ODE type models as a first step in understanding and verifying the more complex models. Science involves the exchange of ideas in a conflict/resolution format. This can be done in either a postive or negative manor ( pun intended :-) ) This is something my advisor (Roger Nisbet) has gently tried to pound into my thick skull. The best justification for complex models is good structured analysis of their strengths and weakneses. Comparision between these and the simpiler models will allow us to understand when various assumptions are or are not valid and give us a better understanding of the level of detail necessary in a model to gain insite into a particular problem. There is a place in this email group for anyone who is working with SWARM and this is a very valuable thread. I hope it continues. Lets try to keep the debate to constructive comments. Cheers, Doug Donalson *************************************************************************** * Doug Donalson * Office: (805) 893-2962 * * Ecology, Evolution, and Marine Biology * Home: (805) 961-4447 * * UC Santa Barbara * email donalson@lifesci.ucsb.edu* * Santa Barbara Ca. 93106 * * *************************************************************************** * * * The most exciting phrase to hear in science, the one that * * hearlds new discoveries, is not "EUREKA" (I have found it) but * * "That's funny ...?" * * * * Isaac Asimov * * * *************************************************************************** ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 18:46:09 1997 From: swarm-modelling@santafe.edu (swarm-modelling@santafe.edu) Date: 24 Apr 1997 17:46:09 -0000 Subject: ABM/I(C?)BM and mathematical nonsense Message-ID: <19970424174609.6508.qmail@mango.tiem.utk.edu> Hi folks, I've been reluctant to jump into the recent discussions, and I'm not certain I've read all of them, but thought I might at least now state a couple of opinions. These are based on my experiences with individual- based models over the last 10 years or so. 1. There are different purposes for modeling, and different approaches appropriate based upon these purposes. Although I have long felt, as an educator, that there is vastly too much emphasis in our current curricula on analytic methods (e.g. ODE, PDE) relative to computational approaches (e.g. rule-based approaches, stochastic simulation), analytic methods allow us to address issues in a different way. The real difficulty, and one that I have the hardest time with in both the formal modeling courses I teach as well as in mentoring graduate students, is deciding which approach is most appropriate for the problems you wish to address. A group of us working on models for Everglades restoration (the ATLSS project - home page at http://www.tiem.utk.edu/~gross/atlss_www/atlss_frame.html) have dealt with these issues for at least one major environmental project. We have concluded that multiple approaches are required (a multimodel, in the sense that Paul Fishwick has defined it), based on the varying spatial and temporal resolutions and associated organismal detail needed to answer the questions of interest in restoration. So we mix ODE models, structured matrix population models, and individual-based models. 2. There are indeed very competent folks who disagree with my belief that individual-based approaches are the appropriate method to address many problems in ecology arising from practical concerns. For one opinion, see Levin et al. (1997) Science 275, 334. The authors of that paper and I disagree on several points, as in particular I cannot concur with their characterization that individual-based models as producing "cartoons that may look like nature but represent no real systems". These differences of opinion are based upon different views of what we can hope to attain by modeling, as well as more philosophical considerations regarding what can arise from reductionist approaches. 3. On nomenclature, an ecologist knows what an individual is (OK - so clonal organisms produce problems here, but that's another topic), and therefore intuitively understands what individual-based approaches mean with very little explanation. I have no desire to try to explain what an "agent" is - I will continue to use the term individual-based models when dealing with ecologists. Names are important. Hal and Meredith's definition of the i- and p-state and configuration models is highly useful, but these have not really entered the ecology literature - people find them too confusing. I have given up trying to explain them to audiences. Cheers, Lou Gross Professor of Ecology and Evolutionary Biology and Mathematics The Institute for Environmental Modeling University of Tennessee - Knoxville gross@tiem.utk.edu http://www.tiem.utk.edu/~gross/ http://archives.math.utk.edu/mathbio/ (Math Archives for Life Sciences) ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Thu Apr 24 18:52:22 1997 From: swarm-modelling@santafe.edu (Sven N. Thommesen) Date: Thu, 24 Apr 1997 12:52:22 -0500 Subject: correction Message-ID: <3.0.32.19970424125218.0095a160@spidle2.humsci.auburn.edu> At 03:38 PM 4/23/97 PDT, Chris Landauer wrote: > >the URL for the complex systems meeting is wrong on their announcement 8-) > >it should be http://www.necsi.org >(they seem to have forgotten the www.) > >more later, >cal Sorry, wrong-o! The site can be accessed either way. (There is no requirement that a URL start with 'www.') Question: was this 'correction' based on empirical knowledge, or on a priori assumptions ... ? (I guess I should refrain from making comments about stunning anti-intellectualism, but the temptation is awesome ... ;-) Cheers, --Sven ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 25 01:06:14 1997 From: swarm-modelling@santafe.edu (Chris Landauer) Date: Thu, 24 Apr 97 17:06:14 PDT Subject: www missing - for me, if not for you Message-ID: <199704250006.RAA16458@antares.aero.org> >>Sven: me >>Sorry, wrong-o! The site can be accessed either way. >>(There is no requirement that a URL start with 'www.') yes, i know >>Question: was this 'correction' based on empirical knowledge, >>or on a priori assumptions ... ? empirical - i try not to correct things i only _think_ might be wrong my correction was based on 'unable to find DNS entry for necsi.org', and on successfully connecting to www.necsi.org, which i found quite interesting so i still don't know which one(s) may be available from other sites, but the one from the message was not available to me when i tried it 8-) more later, cal ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ================================== From swarm-modelling@santafe.edu Fri Apr 25 22:50:11 1997 From: swarm-modelling@santafe.edu (Paul Fishwick) Date: Fri, 25 Apr 1997 17:50:11 -0400 (EDT) Subject: invitation Message-ID: <199704252150.RAA14903@tide.cise.ufl.edu> This is an invitation to present a paper (or organize a session) on simulating complex systems (with multiple agents) using either current or future web technology. For details, please refer to: http://www.cise.ufl.edu/~fishwick/webconf.html -paul Paul A. Fishwick E-Mail: fishwick@cise.ufl.edu Dept. of Computer & Info Phone & FAX: (352) 392-1414 Science and Engineering WWW: http://www.cise.ufl.edu/~fishwick University of Florida (PGP Key available at above WWW address) P. O. Box 116120 332 Bldg. CSE, Gainesville, FL 32611-6120 ================================== Swarm-Modelling is for discussion of Simulation and Modelling techniques esp. using Swarm. For list administration needs (esp. [un]subscribing), please send a message to with "help" in the body of the message. ==================================