[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: computer-go: Group Strength 101




First, local searches for groups with small liberty counts. Perhaps start with
1 or 2 liberties.

First, connections or influence or some other way to collect strings into groups.

Then count eyes, count potential eyes, count ability to run or extend along edge.

Finally, compare liberty counts with nearby enemy groups that can't run to evaluate
semeai.

Counting eyes and potential is quite hard, so it will keep you busy for a while :)

I don't know any fast way to evaluate group strength that is even a little bit accurate :)


At 01:52 PM 5/30/2001 -0400, you wrote:
Okay, my program can play Go. It understands the rules and can do minimax
lookahead. I think the next step is for me to add some heuristics for group
strength. Should I be looking for eyes, doing local searches, counting
liberties? What is the logical first step? Right now, I would like a fast way
to indicate a string or group's strength on a scale of 1-10 or something.
Doesn't have to be 100% accurate, but at this point, I would like it to be fast.
Any suggestions?
David Fotland