[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Swarm-Modelling] theroy behind the swarm simulation toolbox
From: |
Alex Lancaster |
Subject: |
Re: [Swarm-Modelling] theroy behind the swarm simulation toolbox |
Date: |
Wed, 17 Dec 2008 02:09:10 -0700 |
>>>>> "JK" == Jörg Kühne writes:
JK> Dear Swarm group Im looking for a multi-agent strategy for
JK> economic simulation like global financial market with banks as
JK> agents. On what kind of algorithm based swarm? Exists there a
JK> scientific paper which describes the mechanics behind. Exists
JK> there a simulation in form of an animation, for example collisions
JK> in a 2D space, which I could use for a presentation?
I think you are perhaps confused about what Swarm (the software
platform) actually is. It's not a specific model, it's a framework
for building discrete agent-based simulations, thus there is no one
set of mechanics that a Swarm simulation uses: that's up to the user
of Swarm to implement.
That said, there are some canonical types of simulation, such as the
2D lattice of heatbugs and the dynamic scheduling of mousetrap:
http://www.swarm.org/index.php/Examples_of_Swarm_applications
that users often use as a springboard for their own applications.
For the global stockmarket example, there is the Artificial Stock
Market model that Marcus already mentioned:
http://artstkmkt.sourceforge.net
http://www.phy.duke.edu/~palmer/papers/fuji93.pdf
but remember that's only a specific instance of a Swarm model.
When you say "physical nature" or "theory" you might be conflating
"Swarm", the software, with the more specific algorithm known as
"particle swarm optimization":
http://en.wikipedia.org/wiki/Particle_swarm_optimization
Not all Swarm models are instances of this kind of optimization.
There has been some work to formalize agent-based modeling in general
(i.e. not specifically Swarm, the software) by developing a more
formal theory. There is a series of papers that I'm aware of by
Barrett, Mortveit & Reidys: "Elements of a Theory of Simulation", that
develops something called "sequential dynamic systems", the full list
is here:
http://www.ccs.lanl.gov/ccs5/math/pubs.shtml
(I'm sure there are other similar papers as well, I haven't read the
computer science literature on this closely for a while).
Hope this helps,
Alex