sds 2.0.1

Stochastic Diffusion Search

A library which implements the main variants of Stochastic Diffusion Search (SDS), and provides a convenient front end.

Stochastic Diffusion Search (SDS) is a generic population-based search method. SDS agents perform cheap, partial evaluations of a hypothesis (a candidate solution to the search problem). Hypotheses with the potential to be strong solutions are then diffused through the swarm through direct one-to-one communication. As a result of the diffusion mechanism, high-quality solutions can be identified from clusters of agents with the same hypothesis.

This is a library used during the writing of my PhD thesis, full documentation and code are both published online.

SDS has a Scholarpedia page: http://www.scholarpedia.org/article/Stochastic_diffusion_search

A list of papers written on SDS can be found in the Stochastic Diffusion Search paper repository, maintained by the author of this module: http://aomartin.ddns.net/sds-repository/publications.html