[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] Did we miss something in strategy?
|
From: |
Gunnar Farneback |
|
Subject: |
Re: [gnugo-devel] Did we miss something in strategy? |
|
Date: |
Wed, 11 Feb 2004 11:47:32 +0100 |
|
User-agent: |
EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode) |
SP Lee wrote:
> However, for the reason I don't know so well, gnugo wants to attack
> the bigger group L8 and the semeai of L5 and L8 seems to be not
> checked. I think this semeai must be checked. Is this because
> something went wrong, or should we add something to make sure gnugo
> will check the semeai?
It is checked, but misread, possibly because it runs out of nodes.
With the flags '-w -t -d0x101840' you get lots of informative
messages, including
analyze_semeai L8 vs. L5, result 5 5 F1 (508, 95843 nodes, 1.19 seconds)
results if white moves first: WIN WIN, F1 (uncertain)
analyze_semeai L5 vs. L8, result 5 5 J6 (506, 80013 nodes, 1.10 seconds)
results if black moves first: WIN WIN, J6 (uncertain)
It might be possible to reduce this problem by improving the handling
of uncertain semeai results. Getting the semeai code to read it
correctly would of course also be good.
/Gunnar