Author: Albert Silver
Date: 13:17:46 06/16/99
Go up one level in this thread
On June 14, 1999 at 23:58:53, John Stanback wrote: >On June 14, 1999 at 14:14:21, Jeff Anderson wrote: > >>Even most beginners learn to recognize when to play the Bxh7 sac, but for some >>reason not one of my engines, commercial or freeware considers it in the >>position below. Even if the win is beyond its horizon, shouldn't most programs >>use tecniques to analyse lines like these deeper (extensions I think they are >>called.) This theme so common, wouldn't it perhaps even be in the programs >>interest to recognize positions where it could work so they can play it, and not >>fall victim to it? >>Here is such a position where my expensive programs fail to even consider the >>move. >> >>r1b2rk1/3pqppp/p1nNp3/3nP3/5P2/1N1B4/1PPBQ1PP/5RK1 w - - 0 18 > >I was sort of surprised that Zarkov finds Bxh7 in >2 minutes, 40 seconds on a K6-333. The score isn't all >that high though so it's probably a king safety heuristic >that's helping. Here's the output: > >00:01.0 35456 5> -17 c4 Nc7 Qe4 Qh4 Nc5 >00:02.0 84157 6> -17 c4 Nc7 Qe4 Qh4 Nc5 >00:02.3 103168 6 -40 c4 Nc7 Qe4 Qh4 g3 Qh6 >00:02.6 115584 6 -22 Nxc8 Raxc8 Bxa6 Ra8 Bc4 Ndb4 Be3 >00:07.0 364630 7> -22 Nxc8 Raxc8 Bxa6 Ra8 Bc4 Ndb4 Be3 >00:07.7 410496 7 -23 Nxc8 Rfxc8 Bxa6 Rcb8 Bc4 Nc7 Qd3 >00:23.2 1324777 8> -23 Nxc8 Rfxc8 Bxa6 Rcb8 Bc4 Nc7 Qd3 >00:24.7 1429120 8 -30 Nxc8 Rfxc8 Bxa6 Rcb8 Bc4 Ra2 Bxd5 exd5 Bc3 >00:29.4 1703296 8 -26 c4 Nc7 Qe4 g6 Nd4 Rb8 Bc3 Nxd4 Bxd4 >00:58.2 3350016 8 -18 Be4 f6 c4 Nc7 Na5 fxe5 Qd3 h6 Nxc6 dxc6 fxe5 >01:12.7 4216995 9> -18 Be4 f6 c4 Nc7 Na5 fxe5 Qd3 h6 Nxc6 dxc6 fxe5 >01:58.2 6579328 9 -4 Be4 f6 c4 Nc7 Na5 fxe5 Qd3 h6 Nxc6 dxc6 Bxc6 >02:40.6 8674844 9 34 Bxh7+ Kxh7 Qh5+ Kg8 Rf3 g6 Qh6 Qxd6 exd6 Rb8 Na5 >03:05.4 9838080 9 52 Bxh7+ Kxh7 Qh5+ Kg8 Rf3 Nf6 exf6 Qxf6 Rh3 Qh6 >03:28.6 11108858 10> 52 Bxh7+ Kxh7 Qh5+ Kg8 Rf3 Nf6 exf6 Qxf6 Rh3 Qh6 > > >John I put this to my engines in CB7 (with 16 megs hash on a K6-2/350) and here is what I got: Fritz 5.32 - Only places it first after 12 plies (after 11 it gives Bxh7+ a -0.09 score): 1.Bxh7+ Kxh7 2.Qh5+ Kg8 3.Rf3 g6 4.Qh6 f6 5.Rg3 Nxe5 0.81/12 This takes it about 18 minutes. Hiarcs 7.32 - Finds it after 8 plies in around 7 minutes: 1.Bxh7+ Kxh7 2.Qh5+ Kg8 3.Rf3 Nf6 4.exf6 Qxf6 5.Bc3 e5 6.Rh3 Qh6 7.Qxh6 gxh6 8.fxe5 Kh7 1.46/8 Junior 5.32 - After 3 hours it isn't even in it's top 4 choices. Albert Silver
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.