Author: Daniel Clausen
Date: 11:11:19 12/01/03
Go up one level in this thread
On December 01, 2003 at 08:02:25, Roberto Nerici wrote: [snip] >>Well, one thing you could try is this: if your opponent just captured a piece, >>it's likely that you have to recapture (or capture something else) or you're >>just down material-wise. So you could reduce the allocated time by factor X and >>calculate normally. (which reduces your normal search depth a few plies) If >>after the allocated time, the recapture is really the best move and you didn't >>get a drastic fall in the score (e.g you're still in the [-0.2, +0.2] range of >>compared to the previous search or something) it's probably ok to do the >>recapture. > >Thanks, this is the sort of idea I was looking for. Are you suggesting that in >the limited move case, you look for the type of move, and then do a relatively >quick search just to check that it seems ok? Of course there will be positions >where there is something counter-intuitive just over the horizon, but that it >always going to be the risk. Note that I don't have that much experience with it since my engine (once again) can't even generate moves. Maybe someone with experience here could shed some more light on this. :) >All I have thought of so far is to modify the allowed drop off in score between >iterations, based on the number of moves available. i.e. if the engine normally >decides to search longer if score for depth I is X or more below the score for >depth (I-1), then for cases where there are only two moves available, make X a >bigger value. Does that make sense? I understand what you mean, but I can't quite follow the logic. I mean the goal is to find the best move. I don't see how the number of available (in the sense of legal, if I understand you correct here) should have an effect here. Do you think that will less legal moves you probably need less time to refute all the bad moves? (which in case of 2 legal moves can only be 1) Sargon
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.