Scribd
Upload
14 views

Lecture 4-Agent-Based Model

Uploaded by

Sana Khan
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
14 views

Lecture 4-Agent-Based Model

Uploaded by

Sana Khan
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 15

Simulation & Modeling

Agent-Based

Modeling
Prepared by Sania Azhar 1
Agent-Based Model

 Before we discuss why Agent-Based Modeling is important,


we briefly describe what agent-based modeling is.

 An agent is an autonomous computational individual or


object with particular properties and actions and possibly
goals.

 Agent-based modeling is a form of computational modeling


whereby a phenomenon is modeled in terms of agent and
their interactions.
2
Agent-Based Model

 In the past two decades, scientists have increasingly used


agent-based modeling methods to conduct their research.

 ABM (Agent-Based Modeling) is a species of computation,


growing up alongside the maturation of computer
technology.

 The combination of large data, cheap computation, high


connectivity allows agent-based models to be constructed
with million of individual agents.
3
Agent-Based Model

 Agent-based representations are easier to understand than


mathematical representations of the same phenomenon.

 This is because agent-based models are constructed out of


individual objects.
 In thinking about individual agents, we can make sense of
them by projecting our bodily experience onto the agents.

 Thus, the language and concepts we use in ABM is much


closer to natural language and our natural thinking.
4
What is the environment?

 The environment is the landscape on which agents interact


and can be geometric, network-based, or drawn from real
data.
 The interactions that occur between these agents or with
the environment can be quite complex.
 Agents can interact with other agents or with the
environment.
 Not only can the agent’s interaction behaviors change in
time, but so can the strategies used to decide what action
to employ at a particular time.
5
ABM vs. Other Modeling Forms

 Continuous model do not always map well onto real-world


situations.
 For instance, Equation-based Models of population
dynamics treat populations as if they are continuous
quantities.
 In fact they are population of discrete individuals.

 When simulating population dynamics it is very important


to know if you have a sustainable population.

6
ABM vs. Other Modeling Forms

 The most common form of scientific models is the equation


form.
 One distinction is that because ABM models individuals it
can model heterogeneous population.

 Equational models typically must make assumptions of


homogeneity.
 Furthermore, when you model individuals , the interactions
and results are typically discrete and not continuous.

7
Randomness vs. Determinism

 One important feature of agent-based modeling, and of


computational modeling in general, is that it is easy to
incorporate randomness into your models.

 Many equation-based models and other modeling forms


require that each decision in the model be made
deterministically.
 In agent-based models this is not the case.
 Instead, the decisions can be made based on a probability.

8
Randomness vs. Determinism

 Finally, there are often times when we simply do not know


enough about how a complex system works in order to
build a completely deterministic model.

 In many of these cases the only type of model that we can


build is a model with some random elements.

 ABM or other modeling forms that allow you to incorporate


random features are essential to studying these kinds of
systems.
9
ABM Benefits

 ABM has some benefits over other modeling techniques,


but, as with any tool, there are context in which it is more
useful than others.
 ABM can be used to model just about any natural
phenomenon
.
 However, there are some contexts for which the cost of
building an ABM exceeds the benefits, and there are other
times when the benefits are extraordinary given the costs.

10
General Rules

 On the other hand, if a problem has only a handful of


interacting agents, then you usually do not need to bring to
bear the full power of ABM and instead can write detailed
equations described the interaction like two billiard balls
colliding for example, does not require ABM.

 As a rule of thumb, agent-based models are most useful


when there are a medium number of interacting agents.
[Casti,1995]

11
ABM Tradeoffs

 Sometimes ABM is not the right tool for the job.


 For example, ABM can be computationally intensive.

 Simulating thousands or millions of individual can require


great computing power.
 Equation-based models, by contrast, are often very simple
to run and essentially just require repetitive mathematical
calculations.
 This is true only for simple equation-based models.

12
ABM Tradeoffs

 Numerically solving complicated equation-based models


may take as much computational time as agent-based
models.
 The computational expense of running an ABM is a price
one pays for having the benefits of rich individual-level data.

 The additional computational power needed for running


AMBs is the same power that allows the tracking and
development of rich histories of individuals.

13
Simulation and Modeling

Thank You!

14

You might also like