Harald Johnsen wrote:It's not wrong, but in his first post he is talking about reducing by two and there is three in the code
Hehe, didn't catch that.
I suppose the mistake comes from the null move reduction, depth - R - 1. There the reduction is in addition to the -1, because we are searching a subtree (after the null move), whereas here we are searching the same tree.
Harald Johnsen wrote:It's not wrong, but in his first post he is talking about reducing by two and there is three in the code
Hehe, didn't catch that.
I suppose the mistake comes from the null move reduction, depth - R - 1. There the reduction is in addition to the -1, because we are searching a subtree (after the null move), whereas here we are searching the same tree.
exactly. the problem was due to copy/paste. I should not have doing negamax at all, since I haven't done a move (or null move) that changes turns...
