henryzord / bmda Goto Github PK

This repository implements a Python code for the paper

Pelikan, Martin, and Heinz Mühlenbein. "The bivariate marginal distribution algorithm." Advances in Soft Computing. Springer, London, 1999. 521-535.

As a educational code, its implementation is not aimed at high-end performance applications. Instead, legibility in the was kept as clear as possible.

Feel free to use this code as you seem more appropriate, citing the authors:

@incollection{pelikan1999bivariate,
  title={The bivariate marginal distribution algorithm},
  author={Pelikan, Martin and M{\"u}hlenbein, Heinz},
  booktitle={Advances in Soft Computing},
  pages={521--535},
  year={1999},
  publisher={Springer}
}

and adding a foot note to this repository =)

This code has two examples: the graph coloring problem and the worldmap coloring problem. Both problems try to minimize conflicts among nodes/countries (that is, two neighboring nodes/countries must not have the same color).

Evidently, quality of solutions is directly affected by the palette size.

Output of worldmap problem:

Output of graph coloring problem:

Problem

Relationship between variables found by BMDA

Best individual