Author: Uri Blass
Date: 00:51:22 12/27/01
I downloaded WAC.epd from Dann corbit's site in order to test my program and I see that there are some cases when there is more than one solution and at least in the first 2 cases the solutions are not correct [D]r1bqk2r/ppp1nppp/4p3/n5N1/2BPp3/P1P5/2P2PPP/R1BQK2R w KQkq - bm Ba2 Nxf7; id "WAC.022"; I am not sure if Nxf7 wins the game I see that Ba2 is one of the solutions and I do not understand it Nxf7 seems clearly better for positional reasons but I am not sure if it is winning for white Is Ba2 clearly better than other posiiible moves like Nxe6? Here is Deep Fritz's analysis after Ba2 and after Nxe6 and After Ba2(I give only depth 13 because smaller depthes are not interesting) New position r1bqk2r/ppp1nppp/4p3/n5N1/3Pp3/P1P5/B1P2PPP/R1BQK2R b KQkq - 0 1 Analysis by Deep Fritz: 1...b6 2.Qh5 Ng6 3.0-0 Bb7 4.Bxe6 fxe6 5.Nxh7 Kf7 6.Ng5+ = (0.00) Depth: 13/41 00:01:35 52878kN (Blass, Tel-aviv 27.12.2001) After Nxe6 New position r1bqk2r/ppp1nppp/4N3/n7/2BPp3/P1P5/2P2PPP/R1BQK2R b KQkq - 0 1 Analysis by Deep Fritz: 1...Bxe6 2.Bxe6 0-0 3.Ba2 c5 4.Bg5 h6 5.Bh4 cxd4 6.cxd4 Qc7 7.Bxe7 = (-0.03) Depth: 13/37 00:01:20 46390kN (Blass, Tel-aviv 27.12.2001) second example is WAC 31 [D]rb3qk1/pQ3ppp/4p3/3P4/8/1P3N2/1P3PPP/3R2K1 w - - bm Qxa8 d6 dxe6; id "WAC.031 Qxa8 is a bad move and I do not understand how it is one of the solutions d6 dxe6 are good moves and g3 is also a winning move that does not appear in the solutions I already tested my program in the first 31 problems of WAC and it could solve almost all of them in less than a second. The only problem that it could not solve in a reasonable time is WAC 2 I guess that it needs depth 13 in order to solve it and the estimated time that it needs to get this depth is at least some hours. The hardest problem for it (from the problems that it solved) was WAC 22 but I consider it as a positional problem because there are many moves that win a pawn for white to get equality in material. Qh5 is one of them and my program found Nxf7 at depth 4 but changed it's mind at depth 6 to Qh5 only to change it's mind later again to Nxf7 at depth 7. The search techniques are similiar to TSCP except limiting the qsearch to 7 plies I believe that it is illogical to let the qsearch to explode and it is also illogical to let it to explode only when the program search deeper. Qsearch is selective search and if you search many plies forward then you cannot trust your score and if you cannot trust your score there is no logical reason to waste more nodes in order to find a score and it seems better to me to save time and to return static evaluation. Uri
This page took 0.03 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.