Independant Research Update

I've got some work done on the evolutionary algorithm framework, which I'll informally call fea (functional evolutionary algorithms) (informally because I don't feel the need to give names to my projects).
The use of functional programming concepts such as composition and the basic data structures (lists and functions) the framework is much nicer, cleaner, and more expressive than it was before. Making an experiment consists of composing the right units into the right sequence, not filling out a lot of subclasses. I am spending a lot of time fighting java's generics and its syntax, as well as wrapping code in what are essentially lambdas (this would be so much nicer if java would just add closures).
I am just now starting to express gep, and I am glad to find that even in this infant form, my program can express them pretty much like a ga (which is the sample case I used to build the framework for). The only change is exactly where it should be, which is fitness evaluation.
Ideally I would have fitness evaluation consist of providing a translator from list to list and then an evaluator someobject->double. This would isolate the parts of fitness evaluation. There is a great deal of generalizing and abstracting work of this sort to do.

On another note, while programming this, I discovered and discarded self-bounded types. They seem interesting and potentially useful, but they can't seem to introduce generics in subclasses without breaking (I could write the classes but not instantiate them). Maybe I will post some code explaining them, cause they seem to have some cool properties and may be useful someday.