gnugo-devel
[Top][All Lists]
Advanced

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

[gnugo-devel] pattern F410c


From: Evan Berggren Daniel
Subject: [gnugo-devel] pattern F410c
Date: Tue, 19 Aug 2003 10:23:58 -0400 (EDT)

This patch changes pattern F410c (enclose the corner with a keima from
hoshi) to class t instead of j.  This solves nngs4:660.  From a partial
replay of the game GoFuN-gnugo-3.4-200308030626.sgf (the one in the
testcase) we get:

Move 4 (black): GNU Go plays O16 (21.87) - Game move R6 (16.08)
Move 6 (black): GNU Go plays O16 (21.21) - Game move F17 (16.08)
Move 42 (black): GNU Go plays O5 (22.46) - Game move F3 (16.08)

all three of these are improvements.

I think that this move is played far too often at inoportune moments by
gnugo in high handicap games.

Evan Daniel

Index: patterns/fuseki.db
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/fuseki.db,v
retrieving revision 1.32
diff -u -d -u -8 -r1.32 fuseki.db
--- patterns/fuseki.db  11 Aug 2003 13:29:02 -0000      1.32
+++ patterns/fuseki.db  19 Aug 2003 14:09:37 -0000
@@ -1112,27 +1112,29 @@
 x.........?
 x.........?
 -----------

 ; x_alive_somewhere(a,b)


 Pattern F410c
+# evand changed from j to t class. (3.5.1)
+# It was causing too many slow moves early.  See nngs4:660

 x.........xx
 x.........xx
 x..........?
 x...O.....O.
 x.......*...
 x...........
 x...........
 ------------

-:8,FEdj
+:8,FEdt

 x.........cd
 x.........ef
 x..........?
 a...O.....O.
 b.......*...
 x...........
 x...........




reply via email to

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