Author: Robert Hyatt
Date: 12:17:58 12/23/98
Go up one level in this thread
On December 22, 1998 at 07:14:27, Dezhi Zhao wrote: >On December 22, 1998 at 01:14:12, Sylvain Lacombe wrote: > >> >>>I agree; there are usually a lot of forced mates in a given position, and the >>>first one found is usually NOT the fastest. >>> >>>James >> >>Well, if iterative deepening is implemented, it will find the fastest mate. If a >>mate his found at deep 4, their his no reason to continue the search cause there >>is nothing better than mate. But you need a condition for it to stop. >> >>The only thing i can think of for not implementing the PV play is the extra code >>it takes. >> >>Sylvain. > >The things are a little complicated than that , because of extensions >and null moves. You may find a mate in 6 at depth = 4. But if you continue >the search, you got a mate in 5 at depth = 5. > >Dezhi Zhao The only issue here to worry about is that once you have found a mate in N, then the next search (not next iteration, but next search after you play the best move in this search) had better be at least a mate in N-1. You can find the result of not doing this in many early computer chess events. The first one held had a program "Coko IV" blunder like this always playing a mate in 2 move, and eventually losing. The hard part is that finding mates where every move is a check is easy. And we often find mates in 12 when there is a mate in 5 on the board. I don't worry since the next search will mate in 11 or better... this happens when the mate in 5 has one (or more) quiet moves in it, which defeats shallow searches...
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.