gnugo-devel
[Top][All Lists]
Advanced

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

Re: [gnugo-devel] CosmicGG , more good results


From: Arend Bayer
Subject: Re: [gnugo-devel] CosmicGG , more good results
Date: Thu, 21 Aug 2003 14:43:58 +0200 (CEST)

On Tue, 19 Aug 2003, max-d wrote:

> cosmic_gnugo level 11 =20
> GO++ level 5 = 18
>
> it's probably not enough to have an exact estimation , but at least that
> gives less flawed results than Gifu challenge.
>
> time  cosmic_gnugo: 14397
> time  GO++: 25276
>
> (these results must be compared to those i got :
>  = gnugo3321cvs lv 12= 35
>  = GO++ lv 5 = 62)

SP LEE wrote:

> While it's easy for me to win GG with 9 stone komi before cosmic patch, I
> find it's very difficult for me to win cosmic gnugo with 9 stone komi. I
> wonder how others feel about cosmic gnugo and I'm really interested what the
> results will be on nngs games after the cosmic option is used on nngs.

Thanks a lot to you for both of your testing! Comments like this to the
list are helpful and extremely welcome.

Max' observation that --cosmic-gnugo leads to more duplicated games
is also interesting, as this might tell that it leads to less
indeterminacy.

At the moment I'd like to know about two things:

1. Find out whether the patch below, which enables part of the
--cosmic-gnugo stuff unconditionally, can be shown to be beneficial. E.g.
if someone could run a twogtp match of CVS vs the patched version (both
without --cosmic-gnugo), that would be helpful.

2. I'd like to understand why the rengo approach is helpful.

Arend

Index: engine/value_moves.c
===================================================================
RCS file: /cvsroot/gnugo/gnugo/engine/value_moves.c,v
retrieving revision 1.108
diff -u -p -r1.108 value_moves.c
--- engine/value_moves.c        12 Aug 2003 03:05:33 -0000      1.108
+++ engine/value_moves.c        21 Aug 2003 12:44:55 -0000
@@ -3296,57 +3296,42 @@ review_move_reasons(int *the_move, float
 void
 choose_strategy(int color, float score, float game_status)
 {
-
-  minimum_value_weight  = 1.0;
-  maximum_value_weight  = 1.0;
-  territorial_weight    = 1.0;
-  strategical_weight    = 1.0;
-  attack_dragon_weight  = 1.0;
-  invasion_malus_weight = 1.0;
-  followup_weight       = 1.0;
-
   TRACE("  Game status = %f (0.0 = start, 1.0 = game over)\n", game_status);
-

-  if (cosmic_gnugo) {
-
-    if ((game_status > 0.65) &&
-       ((color == BLACK && score < -15.0)
-       || (color == WHITE && score > 15.0))) {
-
-      /* We seem to be winning, so we use conservative settings */
-      minimum_value_weight  = 0.66;
-      maximum_value_weight  = 2.0;
-      territorial_weight    = 0.95;
-      strategical_weight    = 1.0;
-      attack_dragon_weight  = 1.1;
-      invasion_malus_weight = 1.3;
-      followup_weight       = 1.1;
-      TRACE("  %s is leading, using conservative settings.\n",
-               color == WHITE ? "White" : "Black");
-    }
-    else if (game_status > 0.16) {
-
-      /* We're not winning enough yet, try aggressive settings */
-      minimum_value_weight  = 0.66;
-      maximum_value_weight  = 2.0;
-      territorial_weight    = 1.4;
-      strategical_weight    = 0.5;
-      attack_dragon_weight  = 0.62;
-      invasion_malus_weight = 2.0;
-      followup_weight       = 0.62;
-
-      /* If we're getting desesperate, try invasions as a last resort */
-      if ((game_status > 0.75)  &&
-          ((color == BLACK && score > 25.0)
-           || (color == WHITE && score < -25.0)))
-        invasion_malus_weight = 0.2;
-
-      TRACE("  %s is not winning enough, using aggressive settings.\n",
-             color == WHITE ? "White" : "Black");
-    }
+  if ((game_status > 0.65) &&
+     ((color == BLACK && score < -15.0)
+     || (color == WHITE && score > 15.0))) {
+
+    /* We seem to be winning, so we use conservative settings */
+    minimum_value_weight  = 0.66;
+    maximum_value_weight  = 2.0;
+    territorial_weight    = 0.95;
+    strategical_weight    = 1.0;
+    attack_dragon_weight  = 1.1;
+    invasion_malus_weight = 1.3;
+    followup_weight       = 1.1;
+    TRACE("  %s is leading, using conservative settings.\n",
+            color == WHITE ? "White" : "Black");
+  }
+  else if (game_status > 0.16) {
+    /* We're not winning enough yet, try aggressive settings */
+    minimum_value_weight  = 0.66;
+    maximum_value_weight  = 2.0;
+    territorial_weight    = 1.4;
+    strategical_weight    = 0.5;
+    attack_dragon_weight  = 0.62;
+    invasion_malus_weight = 2.0;
+    followup_weight       = 0.62;
+
+    /* If we're getting desesperate, try invasions as a last resort */
+    if ((game_status > 0.75)  &&
+       ((color == BLACK && score > 25.0)
+        || (color == WHITE && score < -25.0)))
+      invasion_malus_weight = 0.2;
+
+    TRACE("  %s is not winning enough, using aggressive settings.\n",
+          color == WHITE ? "White" : "Black");
   }
-
 }








reply via email to

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