Author: Sune Fischer
Date: 06:47:29 03/30/02
Go up one level in this thread
On March 30, 2002 at 09:22:36, Gian-Carlo Pascutto wrote: >On March 30, 2002 at 09:13:26, Sune Fischer wrote: > >>Hmm, must improve...: >>depth=8 -1.03 1.Bd3 Bxd3 2.Nxd3 Rd8 3.Rd1 Rd5 4.Be3 Qc6 >>depth=9 -1.91 1.Bd3 Bxd3 2.Nxd3 Rd8 3.Qc3 Qb5 4.Nb4 Bf6 5.Qe3 Bxb2 >> >>My new version of ChessCraft is blind as a bat, how deep is this line anyway, >>I'm not using any extensions other than regular q-search. > >7 ply (from looking at Sjeng's output) I don't get it, 7 -64 57 219161 Bxg6 hxg6 Qe4 Bf8 Qxg6+ Ng7 Qf7+ Kh7 Rfd1 Be2 Qg6+ Kg is 11 plies and still not a mate, has the pv been extended internally, I don't see this as a mate in fact white is down a bishop for 1 pawn (that's what my eval would see right here). I suspect the line is deeper than 9 plies. >>But all hope is not lost, I think it has a bug, in this position it wants to >>play: Ka1????? >> >>[D]4k3/7p/1r2pQ1P/8/p1p5/P1P1R3/K3B3/8 w - - bm Rxe6! >> >>depth=9 +14.84 1.Ka1 Kd7 2.Bxc4 Rd6 3.Bxe6 Kc6 4.Re1 Kb5 5.Qf4 >> >>That has to be a bug right, it seems 2 ply deep, funny thing is, it was actually >>able to reach this clearly won position, so I have no idea what's going on. >>Any tips as to how I should debug the search? > >Note that 3.Bxe6 should be 3.Bxe6+ ... this may or may not be related >to your problem. that is unrelated, it's because it can't see the move is a check until the move has been made (and the attack tables updated). >Compare perft numbers from this position to other programs. There may or >may not be a problem in detecting legal moves. > >Since it's only 2 ply deep, you can dump the search tree and look at it. I think you are right, it is also the only way I can think of. -S.
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.