Author: Howard Exner
Date: 12:51:58 08/16/98
Go up one level in this thread
On August 16, 1998 at 13:17:25, Frank Schneider wrote: >Hi Howard, > >here are Gromits results on your avoid move positions: > >8/7p/R7/7k/8/p5K1/8/r7 b - - am a3a2; >I think Gromit likes a2 too and would need very long >to avoid it. > >Q6n/8/8/8/8/6K1/pk6/8 w - - am Qxh8; >8/1Q5n/8/8/8/2p5/7K/2k5 w - - am Qxh7; >No problem (endgametablebases) > >1kb5/8/p1p3p1/Pp1p1p1p/1P1P3P/5PP1/2PN4/5K2 w - - am=f3f4; >I'm not sure if Gromit knows that f4 would be wrong but he >likes Nd2-b3 better. see below for corrected epd on this one(Bishop should be on b7 instead of c8) > >8/5p2/p3p1p1/1pbkP2p/5P1P/1P4P1/P3KB2/8 w - - am Bxc5; >Bxc5 after 3 minutes and very likely later as well. This am Bxc5 and am Rd5 in #7 are quite deep. I'm not sure how deep a program would need to go to solve that both these moves lose. > >1r6/5k2/p4p1K/5R2/7P/8/6P1/8 w - - am Kh5; >Gromit want's to play Kh5 (-95) at first but finds Kh7 after 249 sec. (-13). > >8/6p1/1p5p/1R2k3/4p3/1P2P3/1K4PP/3r4 b - - am Rd5; >This is a nice position and Gromit has no idea what's wrong about Rd5. > >8/8/6p1/3Pkp2/4P3/2K5/6P1/n7 w - - am e4f5; bm d6; >fail-low to <-400 after about 50 sec. (iteration 14) >fail-high of d6 after 197 sec. (iteration 16) > >8/8/6p1/3PkP2/8/2K5/6P1/n7 b - - am=g6f5; bm Kxf5; >Gromit likes g6f5 with a score of 0 (iteration 19) but then >fails high on Kxf5 after 491 sec (#19) > >And the LCTII Fin4-Position >5k2/7R/4P2p/5K2/p1r2P1p/8/8/8 b - - am h5; >After 470 sec. there is a fail-high on h3, but possibly >Gromit would change it's idea again later. > >I added the times to the Rebel-results below. > >Frank > >On August 15, 1998 at 18:44:02, Howard Exner wrote: > >>Thanks again to Steven Edwards for correcting the syntax of the epd. >>Here I hope is the corrected version. >> >>8/7p/R7/7k/8/p5K1/8/r7 b - - am a2; >>Q6n/8/8/8/8/6K1/pk6/8 w - - am Qxh8; >>8/1Q5n/8/8/8/2p5/7K/2k5 w - - am Qxh7; >>1k6/1b6/p1p3p1/Pp1p1p1p/1P1P3P/5PP1/2PN4/5K2 w - - am f4 I should have included the best move for position #4 I think white has to play Kf2-Ke3-Kf4 idea as it is the only doorway for the king to enter. After Nb3 black should be able to hold on with Bc8, followed by f4 and Bf5 plan. This is a test of the blocked position idea. After the sequence 1.Nb3 Bc8 2.Nc5 Kb7 3.Kf2 f4! 4. gxf4 Bf5! the position is locked up. Rebel8 thinks it has an advantage but unless I'm missing something I see no way for white to capitalize. >>8/5p2/p3p1p1/1pbkP2p/5P1P/1P4P1/P3KB2/8 w - - am Bxc5; >>1r6/5k2/p4p1K/5R2/7P/8/6P1/8 w - - am Kh5; >>8/6p1/1p5p/1R2k3/4p3/1P2P3/1K4PP/3r4 b - - am Rd5; >>8/8/6p1/3Pkp2/4P3/2K5/6P1/n7 w - - am exf5; bm d6; >>8/8/6p1/3PkP2/8/2K5/6P1/n7 b - - am gxf5; bm Kxf5; >> >> >>Here are the results of Rebel8. >> >>Pos Rebel8 Default Rebel8.4G Gromit >> >>1 >15:00 >15:00 - >>2 3:50 0:28 <0:05 >>3 4:47 4:24 <0:05 >>4 0:05 0:50 <0:05 (luck?) >>5 >15:00 >15:00 - >>6 2:10 1:37 4:09 >>7 >15:00 13:24 - >>8 3:00 5:30 3:17 >>9 >15:00 1:12 8:11 >> >>Rebel8.4G is sel=4 with aggressive style >>Machine is a K6-233 60mb hash >> >>Position 8 and 9 are from the same game. It appears >>Rebel default is playing d6 in #8 for the wrong reason >>as it cannot find the correct followup from #9. >> >>I'd like to see a program have some code for position #1. >>Some sort of penalty for advancing the rook pawn to the seventh rank >>when no immediate material gain is in site. This is a very common occurance >>in endgames. >> >>As posted earlier positions 2 and 3 were found immediately by CM4000. >>Do other programs include knowledge for these two basic endings?
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.