On Nov 13, 2007 2:48 PM, Petr Baudis <[EMAIL PROTECTED]> wrote: > I'm now somewhat torn. The speedup from using pseudo-liberty counts > could be huge, estimating from my profiling. On the other hand, it would > be very useful to still be able to quickly check if a group is in atari > - it looks like if atari stones would get special attention during the > random games, that could make the bot a lot stronger. > > Is there any known way to get the best of the both worlds? :-)
Maybe... I've previously used what I call local liberties. By examining the 3x3 area around the stone played, it's possible to avoid most over counting by pseudo liberties. It's guaranteed accurate for chains of 7 stones or less. Above that, there's a bounded counting error 1/2*(chain size-6). I know the counting error can be bounded better based on what 3x3 shapes are observed, but I have not gotten fancy enough to find a coding scheme for it. Local liberties essentially boils down to counting pseudo liberties, subtracting the number of open triangles, and adding in the number of 3x3 squares with a hole in the middle. It's easy to use incrementally, but large chains with low liberties will require resorting to another liberty tracking method (such as pseudo liberties). It's reasonable to experiment with, but I have not tried it in an MC engine.
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
