Author: Bruce Moreland
Date: 22:31:51 12/18/02
Go up one level in this thread
On December 18, 2002 at 23:48:28, Russell Reagan wrote: >On December 18, 2002 at 22:55:56, Bruce Moreland wrote: > >>Let's say that your current depth is 5. You are going to make each of your >>moves, and then give black a 4-ply response. > >Here I wonder if this is a realistic problem. How often are you going to be in a >situation where you only have time for a 5 ply search? It seems like most >programs are into double digit plies in the first handful of seconds. At higher >search depths is this still a non-trivial problem? A 10-ply search has about a zillion 5-ply searches inside of it. >>Some programs cannot detect non-capturing mates at the horizon, so if this >>program is like that, black could have a mate in one, and it would not be >>detected. > >Is this a mate in one from the root position, or in a position at some point in >the null-move search? I'm a little confused here, because it doesn't seem >logical that you could miss a mate in one in a 5-ply search. In the reduced >search, black gets a move, and white gets a response. If black has a mate in >one, it will mate on its move, one ply before the search was to end, and it will >be detected. I'm talking about sub-trees all through here. From the root you have more time, but a big search is a bunch of small searches. At a critical point, if you make a mistake, you could be hosed. >>So the knife stops inches from white's heart, white thinks he is fine, and this >>whole line is discarded as "awesome for white". > >Hah! That's a nice way to put it. :) > >>Do you understand? If you don't, I will try again, because this is important. > >Yes it makes sense mostly. I see how an engine could miss things because of >null-move, but as seen above, I'm not sure I understand exactly how bad the >problem is. A 5-ply search missing a mate in one sounds like a bad problem to >me, but I'm not sure if that's what you meant or not (see above). If this >problem holds with increased depth, it could really cause problems. IE white >moves into a mate in nine even though it did a 13-ply search. Not good. > >>I think it's the single worst problem that you'll >>have if you use traditional null move search with R=2. > >So is it a good thing or a bad thing that this is "the single worst problem >you'll have if you use traditional null move search with R=2"? Do you mean, >"This is as bad as it gets! Wonderful!"? Or do you mean, "This is a horrible >problem that's hard to fix. Argh!"? The endgame zugzwang problem is not as bad as this problem. This problem affects middlegame play. It gets you mated if you are not careful. bruce
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.