IDS: 深化迭代步长由1改为2 (WAR)
原因: 对于以下棋局 r2 s000 t00 (2,4) 00:00.000 (2,8) 00:00.487 (2,6) 00:46.657 (2,5) 00:00.976 (2,2) 00:49.176 (2,3) 00:02.488 (3,6) 00:55.784 (1,6) 00:03.387 (1,4) 01:09.749 (3,5) 00:03.758 最后一步 (3,5) 而非封堵 (3,4) 最终输棋 调试时发现在迭代深度为5时 之前 (3,4) 排第1的顺序被打乱 故迭代为偶数以跳过5 和原来的程序对战结果 2:5-3.5, 所以这种改法为 WAR 方案
This commit is contained in:
parent
527fec2856
commit
a3301d19d0
|
@ -582,7 +582,7 @@ int NineChessAi_ab::alphaBetaPruning(int depth)
|
|||
|
||||
#ifdef IDS_SUPPORT
|
||||
// 深化迭代
|
||||
for (int i = 2; i < d; i++) {
|
||||
for (int i = 2; i < d; i += 2) {
|
||||
alphaBetaPruning(i, -INF_VALUE, INF_VALUE, rootNode);
|
||||
}
|
||||
|
||||
|
|
Loading…
Reference in New Issue