Agent-Based Model

From CasGroup

Jump to: navigation, search
m (Reverted edits by Eboxytezi (Talk) to last version by Jfromm)
(→Definition)
 
Line 1: Line 1:
-
Agent-based modelling is the modelling of phenomena as dynamical systems of
+
Agent-based modelling (ABM) is the modelling of phenomena as dynamical systems of interacting agents. An '''agent-based model''' is a specific individual based computational model for the computer simulation of [[Complex_System|complex systems]]. ABM can be used as abbreviation for both, Agent-Based Modelling or Agent-Based Model.
-
interacting agents. An '''agent-based model''' (ABM) is a specific individual based computational model  
+
-
for the computer simulation of [[Complex_System|complex systems]].  
+
== Definition ==
== Definition ==
-
An agent-based model is a computational model in which every participant in a system  
+
An agent-based model is a computational model in which every participant in a system or process is modelled indivdually. A system is modeled as a collection of autonomous decision-making entities called [[Agent|agents]]. Each agent acts autonomously, individually assesses its situation, and makes decisions on the basis of a set of internal rules. In agent-based modeling the word 'model' often means to specify 'the rules of the game': to specify exactly what kind of participants exist (the agents and their states) and how they interact (the rules). As Bonabeau says in his paper about [http://www.pnas.org/cgi/content/full/99/suppl_3/7280 agent-based modeling] "at the simplest level, an agent-based model consists of a system of agents and the relationships between them". ABMs are the equivalent to games in [http://en.wikipedia.org/wiki/Game_theory game theory]. A strategic game in game theory consists of a set of players, a set of moves (or strategies) available to those players, and a specification of payoffs for each combination of strategies. An ABM consists of a set of (simplified) agents, a set of possible basic behaviors or  elementary interactions, and a set of rules which specifies the evolution or development of the system.
-
or process is modelled indivdually.
+
-
In agent-based modeling the word 'model' often means to specify 'the rules of the game':  
+
-
to specify exactly what kind of participants exist (the agents and their states) and how  
+
-
they interact (the rules). As Bonabeau says in his paper about  
+
-
[http://www.pnas.org/cgi/content/full/99/suppl_3/7280 agent-based modeling]
+
-
"at the simplest level, an agent-based model consists of a system of agents and the relationships  
+
-
between them". ABMs are the equivalent to games in [http://en.wikipedia.org/wiki/Game_theory game theory].
+
-
A strategic game in game theory consists of a set of players, a set of moves (or strategies) available  
+
-
to those players, and a specification of payoffs for each combination of strategies.
+
-
An ABM consists of a set of (simplified) agents, a set of possible basic behaviors or   
+
-
elementary interactions, and a set of rules which specifies the evolution or development  
+
-
of the system.
+
-
Contrary to general [[Multi-Agent System|multi agent systems]],
+
Contrary to general [[Multi-Agent System|multi agent systems]], an ABM describes the interactions among individual [[Agent|agents]] and their environment in a specific situation which leads to particular organizational patterns and [[Emergence|emergent properties]]. Model in 'agent-based model' is used in the sense of an abstract representation of a concrete system from a particular viewpoint. ABMs can be used to explain collective human behavior in the social sciences, to understand [[Complex System|complex systems]], and they enable the user to run large scale virtual experiments without altering the corresponding real [[System|system]].
-
an ABM describes the interactions among individual [[Agent|agents]] and their environment
+
-
in a specific situation which leads to particular organizational patterns and  
+
-
[[Emergence|emergent properties]]. Model in 'agent-based model' is used in the  
+
-
sense of an abstract representation of a concrete system from a particular viewpoint.
+
-
ABMs can be used to explain collective human behavior in the social sciences, to  
+
-
understand [[Complex System|complex systems]], and they enable the user to run  
+
-
large scale virtual experiments without altering the corresponding real [[System|system]].
+
== Ingredients ==
== Ingredients ==
Line 48: Line 27:
* [[Adaptation|Adaptive behavior]], which means agents grow smarter over time in response to the specific conditions and requirements of their environment
* [[Adaptation|Adaptive behavior]], which means agents grow smarter over time in response to the specific conditions and requirements of their environment
-
Agent-based simulations can be used to model
+
ABMs can simulate almost any interactive system: a stockmarket, a habitat, a pandemics, any social or political system, or a business supply-chain. Agent-based simulations can be used to model
* animal societies, ecological systems and food webs
* animal societies, ecological systems and food webs
* physiological systems and organisms
* physiological systems and organisms
-
* social systems
+
* social or politcal systems  
-
* economic systems
+
* financial systems and stockmarkets
 +
* economic systems and business supply-chains
* robotic systems
* robotic systems
* traffic and transportation systems, pedestrian dynamics
* traffic and transportation systems, pedestrian dynamics
* P2P and other computational systems  
* P2P and other computational systems  
 +
* pandemics
* combats and wars
* combats and wars
Line 112: Line 93:
* Axelrod and Bennett's (1993) model of competing bimodal coalitions
* Axelrod and Bennett's (1993) model of competing bimodal coalitions
* Joyce et al.'s (2006) model of conditional association
* Joyce et al.'s (2006) model of conditional association
 +
 +
== Models and Plays ==
 +
 +
Can agent based models and traditional plays
 +
(for example Shakespeare's plays) be considered as two
 +
extremes on one scale? In both we witness the outcome of a
 +
small number of agents or actors interacting
 +
with each other in a particular environment
 +
according to certain rules and intentions.
 +
Shakespeares' plays are basic "models" for the
 +
complexity which arises through "love in hate"
 +
(Romeo and Juliet), "hesitation in action" (Hamlet)
 +
or "striving against destiny" (Macbeth). In
 +
Shakespeares' words "Though this be madness,
 +
yet there is method in't."
 +
 +
Can plays be considered as a kind of model for systems
 +
with complex BDI agents and abstract rules?
 +
If a developer programs an object-oriented system,
 +
he determines in detail what every object should do,
 +
step by step, in every possible situation.
 +
What happens if we write advanced programs for agent-
 +
oriented systems, would they look like plays from
 +
a playwright? Or would we have to program
 +
beliefs, desires, emotions and intentions?
 +
 +
In plays, the actors are driven by basic emotions
 +
and intentions. One theme is "love in hate" or
 +
"desire in vengeance" which appears in "Romeo and
 +
Juliet". Among others, the "desire in vengeance"
 +
theme can also be found in Hamlet: Ophelia and
 +
Hamlet love each other, but Hamlet kills her
 +
father accidentally, she kills herself, then her
 +
brother Laertes takes revenge, and in the end
 +
they are all dead.
 +
 +
The playwright usually describes one of many
 +
possible plots or outcomes for a certain theme,
 +
here for instance what could happen if love and
 +
hate collide. The basic motives are as simple as
 +
the rules in basic ABMs. This is why I said
 +
that ABMs and plays can be considered as two
 +
extremes on one scale:
 +
 +
ABMs:
 +
* simple agents, driven by basic rules
 +
* outcome (the sequence of actions) is not determined
 +
 +
Plays:
 +
* complex actors, driven by basic emotions and intentions
 +
* outcome (the sequence of actions) is determined
 +
== Literature ==
== Literature ==

Latest revision as of 22:39, 16 March 2011

Retrieved from "https://wiki.cas-group.net/index.php?title=Agent-Based_Model"
Personal tools