Author: Richard Pijl
Date: 06:50:53 12/27/02
Go up one level in this thread
On December 27, 2002 at 08:45:28, Uri Blass wrote: >On December 27, 2002 at 07:48:51, Richard Pijl wrote: > >>On December 27, 2002 at 06:14:13, Jonathan Lee wrote: >> >>>[D]6bk/ppp2p1p/5P1P/8/8/8/8/1K6 w - - 0 1 >> >>I'm not sure what you want to see here. >>Baron 0.99.4 beta 3 sees a very clear win for black (no mate score though): > >I guess some bug with null move pruning. > Yes, I think it is. I had to collect my car from the dealer so I walked a bit, clearing my head and thinking about what could be the cause. I already knew it had to be something with forward pruning. I'm not doing futility pruning in the endgame so it had to be null-move. I'm using double nullmove, and the zugzwang resilience is based on the principle that two null-moves in a row should kill all zugzwang traps. However, it only works correctly when the condition for performing a nullmove is symmetrical: You need to be able to do two nullmoves in a row in order to let it work. It wasn't in my implementation. I allowed a nullmove for the side that still had at least one minor or major piece, even if the other side had none. I changed that now. The result is shown below Now The Baron finds the draw in 5 seconds on my machine: ply time nodes score pv 13(16). 0 27733 -687 b1a2 a7a5 a2a3 b7b5 a3b3 a5a4 b3b4 c7c6 b4c3 14(18)& 0 32771 -680 b1a2 a7a5 a2a3 b7b5 a3b3 a5a4 b3b4 c7c6 b4c3 c6c5 c3d3 a4a3 14(18). 0 41909 -680 b1a2 a7a5 a2a3 b7b5 a3b3 a5a4 b3b4 c7c6 b4c3 c6c5 c3d3 a4a3 15(20)- 0 103432 -711 b1a2 b7b5 15(20)& 0 125123 -716 b1a2 b7b5 a2a3 c7c5 a3b3 15(20)& 0 126503 -711 b1c2 b7b5 15(20). 0 126506 -711 b1c2 b7b5 16(20)- 0 301783 -743 b1c2 c7c5 16(20)& 0 309332 -743 b1c2 c7c5 c2b3 a7a5 b3a4 16(20). 0 309336 -743 b1c2 c7c5 c2b3 a7a5 b3a4 17(24)& 0 489798 -718 b1c2 17(24)+ 0 495425 -707 b1b2 17(24)+ 0 497454 -613 b1b2 17(24)& 0 548522 -598 b1b2 a7a6 b2c2 b7b5 17(24). 0 557259 -598 b1b2 a7a6 b2c2 b7b5 18(25)& 0 893811 -598 b1b2 a7a6 b2c2 a6a5 c2b3 c7c5 18(25). 0 912123 -598 b1b2 a7a6 b2c2 a6a5 c2b3 c7c5 19(21)& 1 1015764 -598 b1b2 a7a6 b2c2 a6a5 c2b3 c7c5 b3b2 19(25). 1 1130125 -598 b1b2 a7a6 b2c2 a6a5 c2b3 20(29)& 1 1973016 -609 b1b2 c7c5 b2b3 b7b5 b3c3 b5b4 c3b3 a7a5 b3b2 a5a4 b2c2 20(29). 2 2234160 -609 b1b2 c7c5 b2b3 b7b5 b3c3 b5b4 c3b3 a7a5 b3b2 a5a4 b2c2 21(28)& 2 3088841 -612 b1b2 c7c5 b2b3 b7b5 b3c3 b5b4 c3b3 a7a5 b3b2 a5a4 b2c2 c5c4 c2b1 a4a3 b1a2 21(28). 3 3643986 -612 b1b2 c7c5 b2b3 b7b5 b3c3 b5b4 c3b3 a7a5 b3b2 a5a4 b2c2 c5c4 c2b1 a4a3 b1a2 22(27)& 4 4479317 -612 b1b2 c7c5 b2b3 b7b5 b3c3 c5c4 c3b4 a7a6 b4c3 a6a5 c3c2 22(29). 4 5097144 -612 b1b2 c7c5 b2b3 b7b5 b3c3 c5c4 c3b4 a7a6 b4c3 a6a5 c3c2 23(27)+ 4 5120156 -545 b1b2 23(27)+ 4 5122243 -452 b1b2 23(29)& 5 5355797 -241 b1b2 a7a6 b2a2 b7b5 a2a3 a6a5 a3b2 23(33)& 5 5893701 0 b1c2 a7a6 c2b2 b7b5 b2c3 a6a5 c3b2 23(33). 6 6591955 0 b1c2 a7a6 c2b2 b7b5 b2c3 a6a5 c3b2 24(35)& 7 7355720 0 b1c2 a7a6 c2b2 b7b5 b2c3 a6a5 c3b2 24(35). 8 8848398 0 b1c2 a7a6 c2b2 b7b5 b2c3 a6a5 c3b2 25(36)& 9 10236296 0 b1c2 a7a6 c2b2 b7b5 b2c3 a6a5 c3b2 25(36). 12 12701090 0 b1c2 a7a6 c2b2 b7b5 b2c3 a6a5 c3b2 26(38)& 14 15124678 0 b1c2 a7a6 c2b2 b7b5 b2c3 a6a5 c3b2 26(38). 18 19227652 0 b1c2 a7a6 c2b2 b7b5 b2c3 a6a5 c3b2
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.