The Peer-To-Peer Genetic Engine Package This special Genetic Engine unleashes the parallelization potential of Genetic Algorithms by distributing them over a peer-to-peer network employing the "island hopping"-meachnism.

Package Specification

Using the Distributed Genetic Programming Framework, one can easily distribute a genetic evolution over a network of computers or even the internet. This is done by letting each node run its own evolution, exchanging individuals using a model called "Island Hopping".

"Island Hopping" means that some individual living on an island (the local Genetic Engine) might get to another island (another node in the peer-to-peer network) and live on there. In nature, this will happen by a monkey sitting on a tree trunk swimming on the water, for example. Here it is done by connecting to a random node in the network and transmitting data to it. The indviduals sent away from the local node are called "emigrants", those newly arriving are called "immigrants". Only the best individuals of our node will emigrate.

Also, this package contains the peer-to-peer / client-server hybrid distribution genetic algorithms engine. That allows you to create peer-to-peer networks of interoperating client-server genetic engines.