Author: Anthony Cozzie
Date: 13:01:12 09/11/03
Go up one level in this thread
On September 11, 2003 at 14:00:02, Ed Schröder wrote: >I am currently trying singular extensions again, while running test positions I >got this amazing result: > >[d]8/pR4pk/1b6/2p5/N1p5/8/PP1r2PP/6K1 b - - bm Rxb2; > >Found after just 9 seconds and 11 plies. Without the algoritm it takes 14 or 15 >plies and several minutes of course. I am now looking for the crazy game Gandalf >vs Ferret of a few years ago during the WCM in Maastricht (I believe) where >Ferret miraculous won from a lost position due to singular extensions. > >Thanks in advance, > >Ed Maybe your pruning is hurting you? Zappa gets it in 21 seconds (P4-2.4). 1... c4-c3 2. b2xc3 Rd2xa2 3. Na4xb6 a7xb6 4. Rb7xb6 Ra2-a3 5. c3-c4 Ra3-c3 6. Rb6-b7 Rc3xc4 7. Kg1-f2 Rc4-c2 8. Kf2-f1 = (-0.23) Depth: 13/22 00:00:08.01 5143kN 1... Rd2xb2 2. Na4xb2 c4-c3 3. Rb7xb6 c5-c4 4. Rb6-b4 a7-a5 5. Rb4-b5 c3-c2 6. Kg1-f2 c2-c1=Q 7. Kf2-g3 Qc1-a1 8. a2-a4 c4-c3 + (0.07) Depth: 14/24 00:00:20.56 14077kN = (1.72) Depth: 14/25 00:00:56.07 38094kN = (1.72) Depth: 15/29 00:01:32.84 62357kN I've putzed around with singular extensions. There has to be something good about 'extend forcing moves' but I've never really found a good implementation. I'm also considering switching to MTD(F) which would preclude using them, I think. I just looked over Ferret-Gandalf, and that is one heck of a combination. Even 5-10 moves into it, Zappa still claimed black was winning by several pawns; the combination must be 30 plies deep at least. Ferret didn't handle the closed Ruy very well; but that was one heck of a swindle. anthony
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.