Friday, July 23, 2010

Independent Research Again

I am much further along in my attempt to write a Java program with some among of pure functional programming mixed in. I will be mutating the occasional state, but only because I don't currently have the time or interest in building enough infrastructure to avoid it. Right now I am using almost no new types. Populations are lists, individuals are lists, loci are symbols (which is a class I defined). Almost every part of the system is a higher order function, and no complex type hierarchies are getting in my way. I am really enjoying it, even if it is just Java, and I still have to fight the language and accept messy constructs for some things. Soon enough I will have a binary prefix gene expression program doing some simple symbolic regression using the fitness function I found in a paper on constant creation techniques in gene expression programming.

No comments:

Post a Comment