gnugo-devel
[Top][All Lists]
Advanced

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

Re: [gnugo-devel] endgame tuning


From: Gunnar Farneback
Subject: Re: [gnugo-devel] endgame tuning
Date: Wed, 12 Mar 2003 22:57:31 +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)

Dan wrote:
> Perhaps the standard can allow reg_genmove to be passed an
> optional integer parameter which can be silently discarded
> or used as a random seed by the engine.

The point of having a standardized reg_genmove command is to promote
sharing of test cases between programs. It's unrealistic that two
programs would interpret a random seed sufficiently similarly for it
to be a useful parameter. Hence I prefer to leave it out and keep the
specification simpler.

If we really need the random seed I think it's better to keep those
test cases as gg_genmove.

> grep gg_genmove *.tst | awk '$4 != "" {print $0}' | grep -v '#'
13x13.tst:47 gg_genmove black 8
13x13.tst:88 gg_genmove black 3
arend.tst:15 gg_genmove black 1
arend.tst:24 gg_genmove white 4
arend.tst:39 gg_genmove black 3
arion.tst:1 gg_genmove black 1
nngs.tst:1040 gg_genmove white 2
nngs.tst:1930 gg_genmove white 4
trevor.tst:240 gg_genmove black 1
trevorc.tst:570 gg_genmove white 2
trevorc.tst:1420 gg_genmove white 6

Currently the random seed is used in 11 test cases.

/Gunnar




reply via email to

[Prev in Thread] Current Thread [Next in Thread]