I'm looking for an easy-to-use Genetic Framework in Java, which has to have the following features:
- Builtin multithreading support (possibly using islands and without the need to merge them manually, it will run on an 8-core server)
- Builtin integer chromosome (I need to work with integer arrays and I don't wat to write the chromosome code from scratch)
- Good amount of tutorials (Javadoc is not enough)
- Ability to serialize, pause (or shutdown) and restart the computation from the saved point (the JGAP sample code for serialization fails with an exception during loading)
The goal is to build a software that optimizes an array of
int for a given function (which I already built), and since the array is 128 elements long I need it to work with multiple threads to speed things up.
For now I'm using JGAP which seems to be the best choice, but unfortunately the tutorials are quite "weak" if your goal goes beyond the contents of the Getting Started section. It is easy to use tho, since I didn't have to deal with writing code for crossover or chromosome but I could use the builtin types.
Just to clarify, by easy-to-use I mean something which allows me to start without having to manually write chromosomes, genetic operators or natural selector, so something packaged with basic types and operations. I need to write the less amount of code possible, basically I'd only like to write the evaluation function and plug it to the framework.