Author: Peter Kappler
Date: 22:26:40 11/28/99
Go up one level in this thread
On November 29, 1999 at 00:32:17, Peter McKenzie wrote: >The following position is from a blitz game between Hossa and Beadle (LambChop) >on ICC: > >r3q1k1/ppp1rpp1/2n1b2p/8/2P2B2/3B4/PPPQ1RPP/5RK1 w - - > >Hossa played Bxh6 and wiped Beadle out. This position is an interesting test of >king safety evaluation, as there is no short range mate or win of material after >the sac although I'm pretty confident the sac is correct. > >I tried this position with LambChop, and it too liked Bxh6 giving it a score of >about +0.5 pawns from depths 4 thru 8. On depths 9-10 it switched to Qc3 with a >similar score of around +0.5, then at depth 11 it switched back to Bxh6 and the >score jumps to +1.4. Depth 12 is still Bxh6 at +1.6. > >We discussed this position a bit on ICC, Bob ran it on Crafty which didn't want >to play Bxh6. Peter Kappler ran it on Hiarcs which liked Bxh6 straight away >with a +1 score! Last night when I ran this, I had Hiarcs in "2-line" mode, where it displays the two best variations. (I'm not exactly sure how it does this - probably just two separate search threads...) In this mode, it finds Bxh6 in about a minute (K6-3/450) with a score of +1. 30 seconds later the score jumps to +1.5. But, when I ran it again tonight, I forced Hiarcs to only display one variation, and in this mode it didn't find Bxh6 in the first 2-3 minutes of searching. Instead, it prefers Qc3, with a score +0.8. Fritz seems to find this move much faster. It prefers Bxh6 immediately, then switches to Rf3 with a score of zero, and stays there for ~15 seconds when it finally returns back to Bxh6 with a score of +0.78. I think this is an excellent test position. Hope to see more analysis from other engines. --Peter
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.