Author: Robert Hyatt
Date: 07:15:22 01/19/98
Go up one level in this thread
On January 19, 1998 at 09:56:16, Ernst A. Heinz wrote: >On January 19, 1998 at 09:49:00, Robert Hyatt wrote: > >>On January 19, 1998 at 07:54:25, Ernst A. Heinz wrote: >> >>>On January 19, 1998 at 00:05:42, Robert Hyatt wrote: >>> >>>>to avoid losing this game, you'd have to play this in 10 seconds or >>>>so, as it was a 5 0 game I believe... I certainly can't do it... >>> >>>"DarkThought" discards Bxe7 after 1 sec in iteration #8. >>> >>>=Ernst= >> >> >>Positionally or tactically? > >In iteration #8 tactically (score drops to +0.1). > >After a long think (> 15 min) it switches back to Bxe7 in iteration #12 >with a score of +0.56 -- actually not much better than the other good >alternatives. > >=Ernst= > the problem is, Bxe7 is much worse. I eventually end up at -2 or so letting it run a while. After Nc4 and the other merriment, it goes for the Bxg5 move, expecting hxg5 and Qxg5, but it finds that this can't be played and ends up a piece down for a pawn. your first fail low to .1 doesn't seem to be tactical, as my early scores are only +.5... and this is the point... the bad thing here is that singular extensions light this up instantly... I made Cray Blitz search this thing and at depth=6, 0 seconds it was most unhappy. But when you look at this, both black and white have few good moves. Particularly white who plays Bxe7, but after Nc4 must move the queen *and* hold the bishop, so Qg5 is the only choice. And after h6 Qh4 is the only choice, and after g5, the queen is forced to move but is also cut off from the bishop. So singular extensions find this all very quickly, with enough depth left over to figure out that after playing Bxg5, it isn't going to be able to play Qxg5 and also recover everything it has sacrificed... I'm going to play around with SE again. I don't know whether 100K nps is enough horsepower to do it, but I want to try, based on results I have seen by DB at reduced speeds. BTW my CB analysis was on a single C90 processor running at only 50K nodes per second, so SE apparently can work at slower speeds without killing performance totally. Needs more investigation. >>The point of Bxe7 is that it turns out to >>be aggressive as the devil, shredding black's king-side at the end of >>the >>PV, because it recovers the material on d6, but sacs the bishop on g5. >>If >>you don't like to attack, you might not like Bxe7 no matter what. This >>is >>something I've been working on for several months now. The score for >>Bxe7 >>is only +.5 or so, which is not much better than other alternatives when >>testing crafty... > >"DarkThought" surely likes to attack but in this case Bxe7 seems to be >too unclear at shallow depths ... > >=Ernst=
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.