Local
Interaction Models with Mathematica
and Java
The LocalInteractions
package extends Mathematica
with over 50 functions for the construction of local interaction models
of evolutionary games. It uses a special Java engine so that models run
quickly once they have been configured. Graphs specifying the
underlying topology are created using Combinatorica,
a standard package included with Mathematica.
This combination of Java and Mathematica
provides for maximum flexibility and speed in the design and analysis
of local interaction models of evolutionary games.
Features
Support for arbitrary network topologies, including lattices up
to 150x150.
Support for arbitrary two-player, asymmetric games of N strategies.
Payoff matrices can be agent-dependent, so that the entire
population plays the same game or each agent can play a slightly
different game.
Payoff matrices can be entered using standard Mathematica two-dimensional
notation.
Learning rules determining how agents update strategies can be
programmed directly within Mathematica.
Multiple runs under varying initial conditions can be performed
with statistics and graphics extracted periodically during the run and
saved to disk.
Complete set (every included
in the basic set, plus all source)
After the download has completed, unpack the zip file and open the file
named "Installation Instructions.nb" in Mathematica. You will need to
copy a few files to a special directory on your computer in order to be
able to load the package using ordinary <<Package`SubPackage`
notation within Mathematica.
Acknowledgements
Many thanks to Lloyd Levon and Steven Skiena (SUNY Stony Brook) for
allowing me to incorporate a modified version of their Java Graph
Editor in this software.
Please send bug reports to jalex@lse.ac.uk.
Dr J. McKenzie Alexander Department of Philosophy, Logic
and Scientific Method London School of Economics and
Political Science
