[gnugo-devel] (no subject)

[Stéphane Nicolet]

The patch below sets the valuation malus for suspected ineffective
owl attacks of worms in estimate_territorial_value() to 2.0+epsilon
instead of epsilon.

The regression delta is 4 PASS and 1 FAIL :


6 unexpected PASS!
     http://www.public32.com/regress/?arion.tst:6

1107 unexpected PASS!
    http://www.public32.com/regress/?nicklas1.tst:1107

163 unexpected FAIL: Correct 'B9', got 'C7'
    http://www.public32.com/regress/?trevor.tst:163

1300 unexpected PASS!
    http://www.public32.com/regress/?nngs.tst:1300

1750 unexpected PASS!
    http://www.public32.com/regress/?nngs.tst:1750


Stéphane



--- value_moves.c       Mon Jan 26 14:48:29 2004
+++ value_moves.c       Mon Jan 26 14:36:30 2004
@@ -1818,7 +1818,7 @@
          && dragon[aa].size == worm[aa].size
          && worm[aa].attack_codes[0] == WIN
          && attack_move_reason_known(pos, aa) != WIN) {
-       this_value = 0.05 * (2 * worm[aa].effective_size);
+       this_value = (2.0 + 0.05 * (2 * worm[aa].effective_size));
        TRACE("  %1m: -%f - suspected ineffective owl attack of worm 
%1m\n",
              pos, this_value, aa);
        tot_value -= this_value;


























... /// ... \\\ ...