Two very cool examples that you've built. Congrats. That much have been
a lot of work.
The bots will have something to chew on here. I think a little too much
to chew on, actually. :-)
It looks to me like a special exercise in humility for go programmers.
Shouldn't especially designed 9*9 problems be bad enough at this point?
Stefan
Am 17.01.2011 12:02, schrieb Aja:
Dear all,
I designed another example using winning rate. The correct winning
rate of Black should be 100%. This position is explained in the
following.
The correct result for this position is Black wins 1.5 points by
Chinese scoring.
[LEFT, all BLACK's WIN]
Bottom-left: inspired from the game of pachi. Black has to "connect" 3
times to atari white group, to win the semeai.
Top-left: a basic "connect to atari" inspired from Magnus' example.
Center-left: a basic 3-liberties semeai.
[RIGHT, all WHITE's WIN]
Bottom-right : inspired by Mogo's example, a basic 4-liberties semeai.
Top-right: the most difficult one for computer Go, in my view.
Basically, to win the semeai, White has to self-atari many stones and
several times.
Center-right: White has to connect to avoid being captured.
Trick:
I design a trick for the N3 group. If Black plays at N4, White must
capture it immediately, or White will fail in the semeai of the right
side.
They are all very difficult semeai. But to reach high dan, I think we
have general solutions.
Aja
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
