Computer chess is a long time hobby of mine. It's also a very part time hobby, as you may have gathered by the amount of effort I've put into this site.
I've created a Java based move generator. It is not very fast. Java isn't very fast to begin with, but I think the real issue is object creation. Chess isn't complex enough that you have to create a complicated object model, but I've become a fan of object oriented programming in recent years. Anyway, the source code is included; feel free to use it any way you'd like.
jmovegen-0.1.jarUsage: java -jar jmovegen-0.1jar -depth=D -fen="FEN"
e.g. : java -jar jmovegen-0.1.jar -depth=5 -fen="r3k2r/p1ppqpb1/bn2pnp1/3PN3/1p2P3/2N2Q1p/PPPBBPPP/R3K2R w KQkq -"
I did a master's project on parallel alpha beta search. The files below are in Open Office format.
The files below are in Microsoft Office format, converted from Open Office - so the formatting may be a bit off.
As an undergraduate I did a class project on temporal difference learning. East Carolina University was generous enough to send me to Alberta to present a student abstract, which was published in the AAI-02 journal. At the conference I got to meet a number of interesting people, including Mr. Machine Learning himself, Richard Sutton. (Here's the proof!) I no longer hold out much hope for TD learning in computer chess, but I do think there should be some way to do automated parameter tuning. Anyway, this was a fun experiment.
