hi, My name is Pawel Koziol, I live in Warsaw, Poland, working as a freelance translator, editor or proofreader, thus wasting my PhD in mediaeval literature. I also am a self-taught, amateur programmer, and a co-author of a chess-playing entity called Glass (http://www.marittima.pl/glass).
The go engine, called somwhat unimaginatively "Hopeless" (just like my first chess program) is available at: www.koziol.home.pl/hopeless.zip The machine is based on a random playing engine called Brown (see README). I just used it to play random MonteCarlo playouts, then tried to exclude some more moves, then to direct random playouts a bit. At the root, a score is modified using some heuristics. Hopeless has just achieved its first goal, beating igowin a couple of times on an empty 9x9 board, no komi, in under 5 minutes. It requires about 2000 simulations for each tentative move, so something like 150.000 per early game board position. My next goal is to add "an implausible move rejector" (seems to require less knowledge than a plausible move generator) in order to cut down the playing time. regards, Pawel Koziol [email protected]
_______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
