[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] 3.3.12 vs 3.2
From: |
Gunnar Farneback |
Subject: |
Re: [gnugo-devel] 3.3.12 vs 3.2 |
Date: |
Sun, 01 Dec 2002 13:05:56 +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) |
I wrote:
> I'll try to split this patch up into smaller pieces which are easier
> to evaluate, but this will obviously take a while.
Here's the first piece, mostly including tuning. The regression
results compared to CVS are:
filllib:42 PASS N19 [N19]
nngs1:31 PASS N17 [N17]
lazarus:15 PASS Q15 [Q15|T5|H5|H3]
nngs:880 FAIL H6 [E6]
13x13:85 PASS D2 [D2]
owl1:265 PASS 1 S16 [1 S16]
auto04:1 PASS 1 Q15 [!0]
auto_handtalk:9 PASS 1 T6 [1 (T6|T7)]
gunnar:11 pass K9 [K9]
gunnar:12 pass B13 [B13]
gunnar:18 pass P15 [P15]
(The last three fail in CVS.)
- tuning
- owl tuning
- influence tuning
- connection tuning
- new autohelpers omoyo_opposite and xmoyo_opposite
- CLASS_I added to CLASS_MOVE_REASONS in patterns.h
The last item is a bugfix causing I patterns in patterns.db to no
longer be ignored.
/Gunnar
Index: patterns/barriers.db
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/barriers.db,v
retrieving revision 1.41
diff -u -r1.41 barriers.db
--- patterns/barriers.db 10 Oct 2002 21:19:27 -0000 1.41
+++ patterns/barriers.db 1 Dec 2002 12:03:18 -0000
@@ -953,6 +953,19 @@
;oplay_attack(a,b,C)
+Pattern Barrier66
+# gf New pattern. (3.3.13)
+
+|..X...
+|......
+|..XO..
+|..,,..
+|..,...
+|......
++------
+
+:8,D
+
Pattern Intrusion1
Index: patterns/conn.db
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/conn.db,v
retrieving revision 1.31
diff -u -r1.31 conn.db
--- patterns/conn.db 23 Oct 2002 18:31:35 -0000 1.31
+++ patterns/conn.db 1 Dec 2002 12:03:20 -0000
@@ -811,6 +811,7 @@
Pattern CB16
+# gf Revised constraint. (3.3.13)
?O
x* fragile double connection
@@ -824,7 +825,7 @@
XO
b!
-;!xplay_connect(*,a,b)
+;xplay_connect(a,b) && !xplay_connect(*,a,b)
Pattern CB17
Index: patterns/connections.c
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/connections.c,v
retrieving revision 1.29
diff -u -r1.29 connections.c
--- patterns/connections.c 22 Nov 2002 07:02:14 -0000 1.29
+++ patterns/connections.c 1 Dec 2002 12:03:21 -0000
@@ -219,10 +219,14 @@
if (pattern->class & CLASS_B) {
if (pattern->patn[k].att != ATT_not)
break; /* The inhibition points are guaranteed to come first. */
- if (color == WHITE && white_eye[pos].color == WHITE_BORDER)
+ if (color == WHITE && white_eye[pos].color == WHITE_BORDER) {
white_eye[pos].type |= INHIBIT_CONNECTION;
- else if (color == BLACK && black_eye[pos].color == BLACK_BORDER)
+ DEBUG(DEBUG_DRAGONS, "inhibiting connection at %1m\n", pos);
+ }
+ else if (color == BLACK && black_eye[pos].color == BLACK_BORDER) {
black_eye[pos].type |= INHIBIT_CONNECTION;
+ DEBUG(DEBUG_DRAGONS, "inhibiting connection at %1m\n", pos);
+ }
}
} /* loop over elements */
}
Index: patterns/endgame.db
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/endgame.db,v
retrieving revision 1.39
diff -u -r1.39 endgame.db
--- patterns/endgame.db 15 Nov 2002 11:15:42 -0000 1.39
+++ patterns/endgame.db 1 Dec 2002 12:03:22 -0000
@@ -1452,11 +1452,17 @@
Pattern CE29
# tm New Pattern (3.1.20) (see trevord:950)
# FIXME: this pattern better in patterns.db?
+# gf Added constraint. (3.3.13) (see gunnar:18)
X*X Wedge
O.X
:8,OXe
+
+B*X
+OaX
+
+;oplay_attack_either(*,a,a,B)
Pattern CE30
Index: patterns/mkpat.c
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/mkpat.c,v
retrieving revision 1.92
diff -u -r1.92 mkpat.c
--- patterns/mkpat.c 18 Nov 2002 06:03:52 -0000 1.92
+++ patterns/mkpat.c 1 Dec 2002 12:03:29 -0000
@@ -212,6 +212,10 @@
{"x_somewhere", -1, 0.01,
"somewhere(OTHER_COLOR(color), 0, 1+%d"},
{"o_somewhere", -1, 0.01, "somewhere(color, 0, 1+%d"},
+ {"xmoyo_opposite", 1, 0.01,
+ "(influence_moyo_color_opposite(%s) == OTHER_COLOR(color))"},
+ {"omoyo_opposite", 1, 0.01,
+ "(influence_moyo_color_opposite(%s) == color)"},
{"xmoyo", 1, 0.01,
"(influence_moyo_color(%s) == OTHER_COLOR(color))"},
{"omoyo", 1, 0.01,
Index: patterns/owl_attackpats.db
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/owl_attackpats.db,v
retrieving revision 1.66
diff -u -r1.66 owl_attackpats.db
--- patterns/owl_attackpats.db 12 Nov 2002 13:41:55 -0000 1.66
+++ patterns/owl_attackpats.db 1 Dec 2002 12:03:33 -0000
@@ -3017,6 +3017,22 @@
;owl_escape_value(C)>0 && xplay_attack_either(*,a,a,b)
+Pattern A1005b
+# gf Need to consider both directions. (3.3.13)
+
+X? cut!
+*O
+.Y
+
+:8,-,value(90)
+
+C?
+*b
+aY
+
+;owl_escape_value(C)>0 && xplay_attack_either(*,a,a,b)
+
+
Pattern A1006
XOY push out
Index: patterns/patterns.db
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/patterns.db,v
retrieving revision 1.89
diff -u -r1.89 patterns.db
--- patterns/patterns.db 25 Nov 2002 19:37:12 -0000 1.89
+++ patterns/patterns.db 1 Dec 2002 12:03:42 -0000
@@ -2470,6 +2470,7 @@
# see trevora:610
#FIXME: There's still a problem here, in that this sacrifice
# is deemed unsafe, and not played....
+# gf Is this really a cut pattern? (3.3.13)
OOO block or cut into black's territory.
...
@@ -2489,6 +2490,17 @@
; || !oplay_attack(*,a,b,b)
+Pattern EC79
+# gf New Pattern (3.3.13)
+
+X.O connect
+X*.
+O..
+---
+
+:8,C
+
+
##############################################
# First line, one space, cuts and connections.
##############################################
@@ -4507,6 +4519,24 @@
:8,OXe
+Pattern EB615b
+# gf New pattern. (3.3.13)
+
+.O? Prevent monkey jump or other incursion
+.*X
+...
+---
+
+:8,OXed
+
+.O?
+a*X
+...
+---
+
+;omoyo_opposite(a)
+
+
Pattern EB616
?o.O?
@@ -6869,6 +6899,7 @@
Pattern CB1
+# gf Removed d classification (pattern too general). (3.3.13)
???????? jump into empty space
.......?
@@ -6876,10 +6907,11 @@
.......?
????????
-:-,OEd,jump_out_helper
+:-,OE,jump_out_helper
Pattern CB2
+# gf Removed d classification (pattern too general). (3.3.13)
?????? jump into empty space
?....x
@@ -6887,7 +6919,7 @@
?....x
??????
-:8,OEd,jump_out_helper
+:8,OE,jump_out_helper
Pattern CB3
@@ -6903,6 +6935,7 @@
Pattern CB4
+# gf Removed d classification (pattern too general). (3.3.13)
????? Jump out if starting to get surrounded
?...?
@@ -6910,10 +6943,11 @@
?...?
?.OO?
-:8,OEd,jump_out_helper
+:8,Oe,jump_out_helper
Pattern CB5
+# gf Removed d classification (pattern too general). (3.3.13)
????? jump out
?...O
@@ -6921,10 +6955,11 @@
??..?
?????
-:8,OEd,jump_out_helper
+:8,Oe,jump_out_helper
Pattern CB6
+# gf Removed d classification (pattern too general). (3.3.13)
?????? jump out
?...Oo
@@ -6932,7 +6967,7 @@
??..Oo
??????
-:8,OEd,jump_out_helper
+:8,Oe,jump_out_helper
Pattern CB7
@@ -8178,6 +8213,7 @@
Pattern CB258
+# gf Added constraint. (3.3.13)
?????? jump! (But not down to second line)
?....?
@@ -8187,15 +8223,24 @@
:8,OXd
+??????
+?....?
+oa.*.?
+?....?
+?X...?
+
+;is_surrounded(a)
+
Pattern CB259
+# gf Removed d classification (pattern too general). (3.3.13)
?.... jump!
oO.*.
?....
?X.X.
-:8,OXd
+:8,OX
Pattern CB261
@@ -10587,6 +10632,24 @@
> replace(a,*)
+Pattern ED94
+# gf New pattern. (3.3.13)
+
+oo.O? Significant improvement of eyespace.
+.*.OX Often sufficient for local life.
+..OXX
+-----
+
+:8,dX
+
+oo.b?
+.*abD
+..cDD
+-----
+
+;!attack(D) && !xplay_defend_both(a,b,c)
+
+
#####################################################################
#
# Center defend/attack patterns
@@ -11495,6 +11558,28 @@
-----
;!xmoyo(a)
+
+
+Pattern EJ13b
+# gf New pattern. (3.3.13)
+
+o...... invade
+o......
+O.X.*.X
+o......
+o......
+-------
+
+:8,Ia
+
+o......
+o......
+a.X.*.X
+o......
+o......
+-------
+
+;!weak(a)
Pattern EJ14
Index: patterns/patterns.h
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/patterns.h,v
retrieving revision 1.42
diff -u -r1.42 patterns.h
--- patterns/patterns.h 16 Nov 2002 00:30:25 -0000 1.42
+++ patterns/patterns.h 1 Dec 2002 12:03:42 -0000
@@ -169,7 +169,7 @@
/* Collection of the classes inducing move reasons. */
#define CLASS_MOVE_REASONS (CLASS_C | CLASS_B | CLASS_b | \
- CLASS_e | CLASS_E | CLASS_a | CLASS_d | \
+ CLASS_e | CLASS_E | CLASS_I | CLASS_a | CLASS_d | \
CLASS_J | CLASS_j | CLASS_U | CLASS_T | CLASS_t | \
CLASS_W | CLASS_c | CLASS_F)
Index: patterns/patterns2.db
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/patterns2.db,v
retrieving revision 1.48
diff -u -r1.48 patterns2.db
--- patterns/patterns2.db 15 Nov 2002 11:15:42 -0000 1.48
+++ patterns/patterns2.db 1 Dec 2002 12:03:45 -0000
@@ -2977,6 +2977,25 @@
;alive(d) && !xplay_defend_both(*,a,b,?,c,a,d)
+Pattern Sente18
+# gf New pattern. (3.3.13)
+# The reverse followup value is probably often bigger than 5 points.
+
+X..O..
+XOX*..
+......
+------
+
+:8,X,reverse_followup(5)
+
+C..b..
+COX*.a
+......
+------
+
+;omoyo_opposite(a) && alive(b) && alive(C)
+
+
########################################################
#
# Double threat moves and defense against double threats
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- Re: [gnugo-devel] 3.3.12 vs 3.2,
Gunnar Farneback <=