Author: Rémi Coulom
Date: 03:31:29 07/23/99
Hi, This is a set of problems I collected from positional blunders of my program. They are very easy tests that might be of interest to programmers writing an evaluation function. I am a weak chess player and would be interested in opinions about the comment I made for each test because I am not sure that I am right. Especially, I do not understand well the positions with a question mark that are at the end of this list. Most of these positions were obtained by commenting games with Fritz5.32. Remi ############################################################################ # # If a problem contains only bm moves, it is solved when one of them # is chosen as the best move. If a problem contains only am moves, it # is solved when any move not in the am list is chosen. If a problem has # both am and bm fields, it is solved when all bm moves are given a better # evaluations than all am moves. This means that a bm/am problem can be # solved with a choice of best move that is not in the bm list. For # instance, on the starting position "am f4; bm e4" would be possible, which # only means that the program should give a better evaluation to e4 than to # f4, not that e4 is better than any other move. Of course, to solve such # problems an engine must support some form of n-best move or, better, the # capability to evaluate a list of moves from a given starting position. # # The goal of this positional database is to test the quality of the # evaluation function of a program, not its search. As a consequence, # problems are chosen so that they can be solved very fast if the right # knowledge is implemented correctly. # # "grep pos0 posdb.epd" should strip this file from comments and tests that # need confirmation # ############################################################################ # # We are in the ending now, it is time to centralize the King r2k3r/ppp1bppp/3n4/8/3N4/2N5/PPP1bPPP/3RK2R w K - id "pos0000"; bm Kxe2; # # Create an outside passed pawn 8/3q1k2/5ppp/8/p1R5/P1K1N2P/1P3PP1/8 w - - id "pos0001"; bm b3 b4; # # Block passed pawns + control g file + King safety + Do not expose f pawns 3r2k1/pq6/2r4p/2p1Q3/2Pp1p1p/1P3P2/P4P2/1R1R2K1 w - - id "pos0002"; bm Kh2; am Qxf4; # # King safety + nnRP exchange r5kr/pp3pp1/4p1p1/1Qn5/3BnP2/2P3Pq/PP2BR1P/R5K1 w - - id "pos0003"; bm Rg2 Bf1; am Bxc5; # # Create a passed pawn and destroy Black pawn structure at the price of # a strong Black bishop pair and the exchange of a bishop for 3 pawns. r4rk1/p3pp1p/b1p5/2P1N1p1/4PB2/2b2K2/P4PPP/3R3R w - - id "pos0004"; bm Nxc6; # # f6?? weakens e6 and the pawn shelter of the short castle r3k2r/pp1b1ppp/2n1p3/q7/3P4/P1PB1N2/3Q1PPP/R4RK1 b kq - id "pos0005"; bm Rc8 O-O Rd8 Rb8; am f6; # # b6 is bad because of Rb5 Qa4 Rh5. r3k2r/pp1b2pp/2n1pp2/q7/3P4/P1PB1N2/3Q1PPP/RR4K1 b kq - id "pos0006"; bm Nd8; am b6; # # Rc2 is the only way to provide the King with a shelter (Kc1) 1rq1r1k1/5ppp/b7/b1pNp3/4P3/1B3Q2/PP3PPP/2RK3R w - - id "pos0007"; bm Rc2; # # Bxc3 wins a pawn, but it is a positional catastrophe r1b2rk1/pppp1ppp/5n2/4q3/1b2P3/2NQ4/PPPB1PPP/2KR1B1R b - - id "pos0008"; am Bxc3; # # Weakening the a7 pawn is not worth the loss of control of the center r1bq1rk1/pp2bppp/2nppn2/8/2BNP3/2N1B3/PPP2PPP/R2Q1RK1 w - - id "pos0009"; am Nxc6; r1bqkbnr/pp3ppp/2np4/1Bp1p3/4P3/5N2/PPPP1PPP/RNBQ1RK1 w kq - id "pos0010"; am Bxc6; # # Bxc3+ weakens the pawn structure but not that much. It does not compensate # for the open b file. rn1qk2r/ppp2ppp/5p2/3p1b2/1b1P4/2NBP3/PPP2PPP/R2QK1NR b KQkq - id "pos0011"; am Bxc3+; bm Bxd3; # # The rook should stay on the back rank to be able to control open files 3r2k1/Qp4q1/p1p3pp/3b4/4p1P1/P3P3/1P2B1PP/2R3K1 w - - id "pos0012"; am Rc2; bm Qb6; # # Do not block the h1 Rook with the white King r5k1/1pq1nppp/p2b4/3P1b2/2B5/PPN2NP1/1B2KPP1/R6R w - - id "pos0013"; am Kf1; bm Rhd1; # # Bxg7 was played by TCB 0043 because a King not on the back rank was # given a penalty. It looks like a very bad move. r2q1rk1/1pnbb1np/p2p1ppB/2pP4/4P3/1QN5/PP1NBPPP/R4RK1 w - - id "pos0014"; am Bxg7; # # To move the Knight to e3 1q5k/pr2bp1n/2ppb3/2p1p3/2P1P1rp/1PNP1N2/P3QPPB/R3R2K w - - id "pos0015"; bm Nd1; # # Weakens Black player's pawn structure but costs a good bishop and opens # the f file for black rooks in a closed position. r4rk1/3bqppp/pp2p1n1/n2pP3/2pP4/P1P1QN2/1PBN1PPP/R4RK1 w - - id "pos0016"; am Bxg6; ############################################################################### # # Protect the pawn shield ? 4r1k1/1ppn3p/p1bprppN/8/qPPQPB2/2P3RP/P3R1P1/6K1 b - - id "pos"; bm Kg7; am Kf8 Kh8 # # To be analyzed further rnbq1rk1/pp3ppp/3b1n2/1N1p4/3P4/1PP2N2/5PPP/R1BQKB1R w KQ - id "pos"; am Nxa7 # # Analysis of this endgame ? I guess it is a draw. ppp/pp2p1n1/n2pP3/2pP4/P1P1QN2/1PBN1PPP/R4RK1 w - - id "pos"; # moves by TCB : which ones were errors in the pawn endgame ? [Event "WinBoard 4.0.2, Turnier 4"] [Site "Hamburg/Trier"] [Date "1999.04.16"] [Round "10.1"] [White "TC Bishop 0043"] [Black "ZChess 1.20"] [Result "0-1"] 1. d4 d5 2. c4 e6 3. Nc3 Nc6 4. Nf3 Bb4 5. Qb3 Qe7 6. a3 Bxc3+ 7. Qxc3 Nf6 8. e3 O-O 9. Bd3 Re8 10. b3 dxc4 11. Bxc4 e5 12. dxe5 Nxe5 13. Nxe5 Qxe5 14. Qxe5 Rxe5 15. Bb2 Rg5 16. Bxf6 gxf6 17. Rc1 c6 18. g3 Bg4 19. Be2 Be6 20. Bc4 Bg4 21. Rf1 Re5 22. Rc2 Kf8 23. Kd2 Bf5 24. Rcc1 b5 25. Be2 Rd8+ 26. Ke1 Rd6 27. b4 Be4 28. f3 Bd3 29. Kf2 Bxe2 30. Kxe2 Red5 31. Rcd1 Rxd1 32. Rxd1 Rxd1 33. Kxd1 Ke7 34. Kc2 Kd6 35. e4 c5 36. bxc5+ Kxc5 37. Kc3 a5 38. Kd3 b4 39. a4 Kd6 40. Ke2 Ke5 41. Kd3 f5 42. exf5 Kxf5 43. Kd4 h5 44. Kc4 Ke5 45. Kd3 h4 46. Kc4 Kf6 47. Kd4 hxg3 48. hxg3 Kf5 49. Ke3 Ke5 50. Kd3 Kd5 51. g4 f6 52. f4 Kc5 53. Ke3 Kc4 54. f5 Kd5 55. Kd3 Ke5 56. Ke3 b3 57. Kd3 Kf4 58. Kc3 Kxg4 59. Kxb3 Kxf5 60. Kb2 Ke6 61. Ka2 Kd5 62. Kb1 Kc5 63. Kb2 Kb4 64. Ka1 f5 65. Kb1 Kxa4 66. Kc1 Kb5 67. Kb2 Kc4 68. Kb1 Kd3 69. Ka2 f4 70. Kb3 f3 71. Ka4 Kd4 72. Kb5 f2 73. Kxa5 f1=Q 74. Kb6 Qf6+ 75. Ka7 Kc5 76. Kb7 Qf7+ 77. Ka8 Kc6 78. Kb8 Qb7# 0-1 # # This is a complicated position that would need more analysis : # f4 can be followed by Bh7+ with an interesting king attack (Ne1 Qh2 Kg2 Rh1) # e4 looks like another good defensive move # g6 seems too dangerous, allowing for instance Qg5 & open f file r2q1rk1/pp1n1pp1/2pnp3/6P1/Pb1P4/3BP3/1PQ2PN1/R1B2RK1 w - - id "pos"; bm f4 e4; am g6; # # useless move ? 1r1qr1k1/p4ppp/2pb1n2/3p1b2/3B4/1BN5/PPPQ1PPP/1R3RK1 w - - id "pos"; am Kh1; # # Attack white King ? 1k4rr/p1p5/1p6/1R1P2p1/2q2p1p/7P/P4PP1/1Q1R2K1 b - - id "pos"; bm g4; am Rd8; # # ? 3r1rk1/pb4p1/2npp2p/2p2p2/1qP1B2N/1P2P3/P4PPP/R2QR1K1 w - - id "pos"; bm Bxc6; am Bf3; # # ? 3r2k1/pb4p1/2n1pr1p/2p2p2/1qPp3N/1P2PB2/P3QPPP/R2R2K1 w - - id "pos"; bm Bxc6 exd4 a3; am Kf1;
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.