An introduction to M.A.B.E.

(Modular Agent Based Evolver)

and Markov Network Brains

To see MABE code and documentation, visit:github.com/ahnt/MABE..

Markov Brains have proven to be a powerful tool for studying theoretical evolution and for modeling biological systems. In this workshop you will be introduced to Markov Brains and MABE.

MABE is an evolution system used to generate and evolve populations of digital organisms. Organisms contain genomes which define brains, which in turn control the actions of the organisms. Once created, organisms are placed into a digital world. Worlds can be designed to generate scores which can be used to determine which organisms will sire the next generation or worlds can be designed as open ended systems with implicate mating and death. In either case, both asexual and sexual reproduction are supported, and various forms of mutation and crossover can be applied during reproduction. MABE supports multi-chromosome multi-ploidy genomes.

While MABE was originally developed to study Markov Network Brains, one of the design philosophies was to make it entirely modular. MABE therefore allows for other types of genomes, brains, organisms, and worlds; making it an ideal system to compare different evolving computational systems.

This tutorial will include (rough outline).

• An Introduction to MABE - Why was it written and why you should use it.
• An Introduction to Markov Network Brains
• Using MABE:
  • configuration files
  • output files
  • graphing output
• Advanced features:
  • other genomes and other brains
  • worlds
  • writing brains and brain units
  • writing genomes

The MABE code base and user guide will be added to this page soon (expected date:4/1). Attendies who wish to follow along during the tutorial will be asked to download, compile and go though a basic setup before the tutorial.