Introduction to RePast and Tutorial I

1
Introduction to RePast and Tutorial I
2
Todays agenda

• Introduction to Repast
• IPD Model
• Tutorial sequence

3
What is RePast?

• Recursive Porous Agent Simulation
  Toolkithttp//repast.sourceforge.net
• Repast is an open-source software framework for
  creating agent-based simulations using Java
• Initially developed by the Social Science
  Research Computing at the University of Chicago
• Will be further developed by the RePast
  Organization for Architecture and Development
  (ROAD) and Argonne National Laboratory

4
Why RePast?

• Alternatives Swarm, Ascape, NetLogo...
• RePast is at the moment the most suitable
  simulation framework for the applied modeling of
  social interventions based on theories and data
  (2004)http//jasss.soc.surrey.ac.uk/7/1/6.html
• Modeled on Swarm but easier to use and better
  documented
• Important criteria
• abstraction, ease of use and user-friendliness
• flexibility and extensibility
• performance and scalability
• support for modeling, simulation
  experimentation
• Interoperability (GIS, statistical packages, )

5
What does RePast offer?

• Skeletons of agents and their environment
• Graphical user interface
• Scheduling of simulations
• Parameters management
• Behavior display
• Charting
• Data collection
• Batch and parallel runs
• Utilities and sample models

6
Iterated Prisoners Dilemma

• Cohen, Riolo, and Axelrod. 1999. The Emergence
  of Social Organization in the Prisoner's Dilemma
  (SFI Working Paper 99-01-002)
• http//www.santafe.edu/sfi/publications/99wplist.
  html
• In The Evolution of Cooperation, Robert Axelrod
  (1984) created a computer tournament of IPD
• cooperation sometimes emerges
• Tit For Tat a particularly effective strategy

7
Prisoners Dilemma Game

• Column
• C D
• C 3,3 0,5
• Row
• D 5,0 1,1

8
One-Step Memory Strategies
Strategy (i, p, q)
i prob. of cooperating at t 0 p prob. of
cooperating if opponent cooperated q prob. of
cooperating if opponent defected
C
p
Memory
C
D
q
C
D
D
t
t-1
9
The Four Strategies
10
TFT meets ALLD
Cumulated Payoff
p1 q0
0
1
1
1
3



Row (TFT)
i1
C
D
C
Column (ALLD)
D
i0
1
5
1
1
8



p0 q0
t
0
1
2
3
4
11
Payoffs for 4 x 4 Strategies
12
Three crucial questions

• 1. Variation What are the actors
  characteristics?
• 2. Interaction Who interacts with whom, when and
  where?
• 3. Selection Which agents or strategies are
  retained, and which are destroyed?
• (see Axelrod and Cohen. 1999. Harnessing
  Complexity)

13
Experimental dimensions

• 2 strategy spaces B, C
• 6 interaction processes RWR, 2DK, FRN, FRNE,
  2DS, Tag
• 3 adaptive processes Imit, BMGA, 1FGA

14
Soup-like topology RWR
In each time period, a player interacts with
four other random players.
ALLC
ATFT
ALLD
ALLD
TFT
ALLC
TFT
15
2D-Grid Topology 2DK
The players are arranged on a fixed torus and
interact with four neighbors in the
von-Neumann neighborhood.
16
Fixed Random Network FRN
The players have four random neighbors in a
fixed random network. The relations do not have
to be symmetric.
17
Adaptation through imitation
Imitation
ATFT
ALLC
ALLD
ALLD
TFT
TFT?
ALLC
Neighbors at t
18
Adaptation with BMGAComparison error (prob. 0.1)
Genetic adaptation
6.0
Fixed spatial neighborhood
2.8
9.0
2.2
0.8
19
BMGA continued Copy error (prob. 0.04 per bit)
Genetic adaptation
6.0
Fixed spatial neighborhood
p0 q0 gt p1 q0
2.8
9.0
6.0
0.8
20
Tutorial Sequence

• Today SimpleIPD strategy space
• May 11 EvolIPD RWR
• May 18 GraphIPD charts and GUI
• May 25 GridIPD 2DK
• June 1st ExperIPD batch runs and parameter
  sweeps