Author: Uri Blass
Date: 11:30:02 07/19/04
Go up one level in this thread
On July 19, 2004 at 14:13:39, Peter Berger wrote: >On July 19, 2004 at 14:03:43, Robert Hyatt wrote: > >>On July 19, 2004 at 11:55:43, Daniel Clausen wrote: >> >>>Hi >>> >>>I'm testing the position below with various engines available for Linux and >>>noticed that Crafty 19.15 prints funny things when failing high. >>> >>>[D]3qk3/8/8/8/8/8/PPPPPPPP/RNBQKBNR w - - 0 1 >>> >>>[snip] >>> >>> 9-> 3.58 33.23 1. e3 Ke7 2. Qg4 Qa5 3. Nf3 Qa8 4. >>> Qg7+ Kd6 5. Qd4+ Ke7 6. Ne5 >>> 10 4.43 +1 1. e3!! >>> 10 5.04 +3 1. e3!! >>> 10 8.10 +M 1. e3!! >>> 10 35.02 37.08 1. e3 Ke7 2. Qh5 Kd6 3. Qb5 Qh8 4. >>> Qb4+ Kd5 5. Qc4+ Kd6 6. Qd4+ Qxd4 7. >>> exd4 Kd5 >>> >>>[snip] >>> >>>I noticed the "+1" things in previous games but never saw a "+3" or "+M" so far. >>>I guess the "+3" (simliarly to "+1") means that Crafty got a fail high and knows >>>that that the move is at least +3 better since that was the window of the >>>re-search. But what is M? If it's a certain number, why don't print that? >>> >>>Sargon >> >> >>+1 means a fail high outside the normal aspiration search window of about .4 >>pawns. The +1 means we are upping the upper bound to current + 1.00. The +3 >>means we added another +2 which means we are winning at least a pawn, and maybe >>more. +M means we are upping the bound to +Mate, meaning that we are winning >>more than a piece (more than +3.0). > >Should have put this question into the other post. Have these values been tested >thoroughly? From watching testgames I regularly saw the +1 in PVs, but so far >I have never seen the +3 without being followed by a +M some time later - so >does this trick really save time over all? Did you see +3 only in cases that Crafty was already winning or also in different cases after a big fail? Uri
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.