Author: Robert Hyatt
Date: 17:04:55 05/31/98
Go up one level in this thread
On May 31, 1998 at 18:01:28, Len Spencer wrote: >The other night I happened to look at the "mate in 2" problem on the ICL >home page, and decided to try it out on some of the programs I have. >Needless to say I got some interesting results. > >The position: > >5QbK/r6p/2pRp1p1/2P1k3/2B1npPN/r1p1N/6P1/q1b1R1B1 w > >First I tried Crafty, thinking it should have no problem. But it >INSISTED on Rxe6+ mate in 3. So I tried several of the other programs I >have to see what they would do. Here's what I found: > >1. (I won't give away the answer!) Programs that got it right: > CM4000 > GNU Chess 4.14 Win32 > Cyrus > >2. Rxe6+ (forced mate in 3) > Crafty 15.8 > EXChess > >3. Qxg8 (mate in 3) > Sargon 4 > Grandmaster Chess (Zarkov 2.5 engine) > EdChess 3.0. > >Any ideas? Is the correct move getting lopped off early in the search >simply because it is a "waiting move"? here's what happens in my case. If a mate (shortest mate) has a quiet move, then quite frequently, crafty will find a longer mate quicker because it extends checks quite aggressively. And shallow searches can find deep mates if all moves are checks. And that same shallow search will miss the shorter mate because of the quiet move that can terminate the search before the mate was found... And once a mate is found, and the iteration is completed, crafty exits *if* the mate it found is one move (or better) shorter than the mate it found for its last move... In real chess, finding a mate in 2 over a mate in 3 has no importance whatsoever, so that I choose to exit and play the winning move...
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.