Author: Uri Blass
Date: 04:55:45 12/14/02
Go up one level in this thread
On December 14, 2002 at 05:16:04, Uri Blass wrote: >On December 13, 2002 at 22:44:30, Howard Exner wrote: > >>[D] r1bq1b1r/pp2p1pp/3p4/1n1kn3/1P3B2/P6Q/5PPP/3RK1NR b K - bm Nd4; >> >>Given a bit more time what does Fritz 8 (or other engines) play here? >> >>Nice game. Can't help but be reminded of Bronstein's test games vs Deep Thought. > >Thanks for posting >This seems to expose a bug in latest movei > >It seems here not to evaluate repetition of the root position as a draw and >shows the following line > >depth=8 +4.67 d5c4 d1c1 c4d5 c1d1 d5c6 d1c1 c6b6 h3e3 b6a6 a3a4 c8g4 f4e5 d6e5 >a4b5 a6b5 >Nodes: 3532216 NPS: 155672 >Time: 00:00:22.69 > >Kc4 Rc1+ Kd5 Rd1+ is supposed to be evaoluated as a draw. >When I give it the game there is no problem. > >I will check later for the bug(I suspect not reading correctly castling rights >or ignoring repetition detection for the root position when it is the first >move). > >Uri I found the bug It turns out that movei simply did not update the hash table array correctly for the root position. It is not a problem in games because when white starts and the move is the first move there is no problem but it can be a problem at analysis. Movei remembered the hash key of the root position at zobkey[0] when it looked at zobkey[1] that was not initialized in the root position. I fixed that bug by changing the place of one line. 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.