readme

来自「非常好的进化算法EC 实现平台 可以实现多种算法 GA GP」· 代码 · 共 42 行

+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+

Iterated Prisoner's Dilemma (ipd): Co-evolution GA example with Open BEAGLE

Copyright (C) 2003
by  Christian Gagne <[e-mail]>
and Marc Parizeau <[e-mail]>

+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+


Getting started
===============

  Example is compiled in binary 'ipd'. Configuration file for first
  thread/population is in file 'ipd-thread1.conf', while configuration file
  for second thread/population is in file 'ipd-thread2.conf'.

Objective
=========

  Find solutions to the iterated prisoner's dilemma that get the highest reward for
  its actions.

Representation
==============

  Bit strings made of 3 bits. First bit is the action to take at the first turn
  (0=defect, 1=cooperate), the second bit is the action to make knowing that
  the companion has defected at the last turn, and the third bit is the action to
  do knowing that the companion has cooperate the last turn.

Fitness
=======

  For each turn, the individual receives points for its action. Each evaluation is done
  on a total of 8 turns (iterations). The points given for each turn are the following:
  - 0 points if the player cooperate and the companion defects.
  - 1 points if the player defects and the companion defects.
  - 3 points if the player defects and the companion cooperate.
  - 5 points if the player cooperate and the companion cooperate.