WCSAC re-revisited

[sje]

WCSAC (1,001 Wining Chess Sacrifices and Combinations by Fred Reinfeld) is a tactical test suite that's been around for a while and was one of the first released in EPD format.

I've been running Symbolic against WCSAC with a maximum time limit per problem starting at one second and doubling each time through the set. I'm using a 3 GHz Pentium 4 box and a fairly small transposition table.

At 4096 seconds, 983 solved.
At 8192 seconds, 984 solved.
At 16384 seconds, 984 solved.
At 32768 seconds, 984 solved.
Now running 65536 second limit.

The seventeen tough problems are:

Code: Select all

8/1bp4p/p2p4/1p3rbk/4p3/1BP5/PP3PP1/3RR1K1 w - - acd 17; acn 7735890261; bm g4+; fmvn 1; hmvc 0; id WCSAC.0025; pes +1.131; pv a4 Re5 g3 Kg6 Bc2 Kg7 axb5 Rxb5 b4 Bf6 Re3 Re5 Rde1 a5 Bxe4 Rxe4 Rxe4 Bxe4 Rxe4 Bxc3; scv 09:06:07.94; solve 0; sts TimeLimitNode; suv 09:06:07.94; tbpc 0;
2b5/1p3r2/p1p2rk1/P1Pp1pp1/3P2P1/4P2P/2B1K3/1R3R2 w - - acd 19; acn 9535805492; bm gxf5+; fmvn 1; hmvc 0; id WCSAC.0027; pes +0.838; pv Kd2 Be6 Rf2 Bc8 Rbf1 Rf8 Rg2 Rh8 Rb1 Rh7 Bd3 Rxh3 Bxf5+ Bxf5 gxf5+; scv 09:06:07.89; solve 0; sts TimeLimitNode; suv 09:06:07.89; tbpc 0;
4rrk1/ppp1qppp/5n2/2b1n1N1/2B3b1/1QN5/PB4PP/4RR1K w - - acd 17; acn 6287162579; bm Bxf7+; fmvn 1; hmvc 0; id WCSAC.0037; pes +2.607; pv Nd5 Nxd5 Rxe5 Qxe5 Bxe5 Rxe5 Nxf7 Re3 Qxb7 Nf6 Ng5+ Kh8 Qxc7 Bb4 Qf4 Re1 Nf7+ Rxf7 Bxf7 Rxf1+ Qxf1 Bd6; scv 09:06:07.91; solve 0; sts TimeLimitNode; suv 09:06:07.91; tbpc 0;
r1bq1rk1/pp2bp1p/2p2np1/3p2B1/3P4/2NQ2N1/PPP2PPP/4RRK1 w - - acd 16; acn 6287781389; bm Rxe7; fmvn 1; hmvc 0; id WCSAC.0062; pes +0.664; pv f4 Bd6 f5 Kg7 Nce2 h6 Bxf6+ Qxf6 fxg6 Qxg6 Qb3 a5 Nf4 Bxf4 Rxf4 a4 Qc3 Bg4; scv 09:06:07.94; solve 0; sts TimeLimitNode; suv 09:06:07.94; tbpc 0;
2r1r1k1/1p1qbp1p/3p1p2/p2P4/3Q4/5N2/PP2RPPP/4R1K1 w - - acd 17; acn 6948581674; bm Rxe7; fmvn 1; hmvc 0; id WCSAC.0247; pes +1.500; pv h3 Bf8 Qxf6 Rxe2 Qg5+ Bg7 Rxe2 Re8 Rxe8+ Qxe8 b3 Qc8 Nh4 h6 Qd2 b6 Qe3 b5 Nf3 Qa8; scv 09:06:08.59; solve 0; sts TimeLimitNode; suv 09:06:07.91; tbpc 0;
r2r2k1/ppb1qpp1/2n2n1p/2P1p3/1P1p4/P3PN2/1BQ1NPPP/3RK2R b K - acd 16; acn 6469927049; bm d3; fmvn 1; hmvc 0; id WCSAC.0253; pes +0.846; pv dxe3 Rxd8+ Rxd8 fxe3 Nd5 Qb3 b6 e4 Nf6 Qa4 Qd7 Nc3 bxc5 bxc5 Qd3 Kf2 Rb8 Ba1 Na5; scv 09:06:08.33; solve 0; sts TimeLimitNode; suv 09:06:07.86; tbpc 0;
rn1q1rk1/pbppbppp/1p2p3/8/2P5/2n2NP1/PPQPPPBP/R1B1R1K1 w - - acd 17; acn 5791510767; bm Ng5; fmvn 1; hmvc 0; id WCSAC.0315; pes +0.033; pv dxc3 f5 Bf4 Na6 b4 Bf6 Qd2 Qe8 c5 Rc8 cxb6 cxb6 Rac1 e5 Bg5 e4 Nd4 Bxg5 Qxg5; scv 09:06:08.35; solve 0; sts TimeLimitNode; suv 09:06:07.87; tbpc 0;
rqb2rk1/1p2pp1p/p5p1/3Nb3/P1B2n2/1Q3N2/1P3PPP/2RR2K1 w - - acd 15; acn 6539342980; bm Nb6; fmvn 1; hmvc 0; id WCSAC.0327; pes +1.635; pv Nxe7+ Kh8 Nxc8 Rxc8 g3 Bf6 gxf4 Qxf4 Rc2 Rab8 h3 b5 axb5 axb5 Bd5 Rxc2 Qxc2 Rg8; scv 09:06:08.32; solve 0; sts TimeLimitNode; suv 09:06:07.86; tbpc 0;
3r4/1p3k1p/p1b1p1p1/4Q1Pn/2B3KP/4pP2/PP2R1N1/6q1 b - - acd 16; acn 7364112938; bm Rd4+; fmvn 1; hmvc 0; id WCSAC.0472; pes +1.243; pv Bd7 Kh3 Qh1+ Kg4 b5 Bb3 Ke7 Nxe3 Qg1+ Kh3 Rf8 Nd5+ Ke8 Nf6+ Rxf6 Qb8+ Kf7 gxf6 Qf1+ Rg2 Qh1+ Qh2 Nf4+ Kg3 Qxg2+ Qxg2 Nxg2 Kxg2 Kxf6; scv 09:06:08.35; solve 0; sts TimeLimitNode; suv 09:06:07.89; tbpc 0;
r4rk1/2p2pp1/pppb1q1p/4pN2/3PPn2/5Q1P/PPP2PP1/R1B2RK1 b - - acd 17; acn 6555834146; bm exd4; fmvn 1; hmvc 0; id WCSAC.0475; pes -1.381; pv g5 Bxf4 gxf4 Qh5 Qg6 Qxh6 exd4 Qxg6+ fxg6 Nxd4 Rae8 Rfe1 Be5 c3 Bxd4 cxd4 Re6 e5 Rfe8 g3 f3 Rac1; scv 09:06:08.53; solve 0; sts TimeLimitNode; suv 09:06:07.92; tbpc 0;
2rr1b2/8/ppkq1pP1/3pp3/2p5/2BPPpP1/BQ3P2/1RRN1n1K b - - acd 15; acn 6613870754; bm Qb8; fmvn 1; hmvc 0; id WCSAC.0580; pes -2.265; pv Rb8 dxc4 d4 Qc2 dxc3 Nxc3 Qd3 Nd5 Qxc2 Rxc2 Nxg3+ fxg3 Bg7 Kg1 Rh8 Rcb2 b5 cxb5+ axb5; scv 09:06:08.39; solve 0; sts TimeLimitNode; suv 09:06:07.89; tbpc 0;
5k2/p1q2p1p/2Pr2p1/4Q3/P7/6P1/5P1P/2R3K1 w - - acd 15; acn 9138289340; bm Qxd6+; fmvn 1; hmvc 0; id WCSAC.0641; pes +6.560; pv Re1 Kg8 Qe8+ Kg7 Re7 Qxe7 Qxe7 Rxc6 Qxa7 Rc4 a5 h5 h4; scv 09:06:07.89; solve 0; sts TimeLimitNode; suv 09:06:07.89; tbpc 0;
rnb1kb1r/pp3p2/2p4p/7q/3PN2P/1B4p1/PPP2KP1/R1BQN2R w kq - acd 18; acn 6837252835; bm Ke3; fmvn 1; hmvc 0; id WCSAC.0824; pes +1.274; pv Kf1 Qxd1 c3 Qxb3 axb3 Be6 Bf4 Be7 Nxg3 Na6 b4 O-O-O; scv 09:06:07.88; solve 0; sts TimeLimitNode; suv 09:06:07.88; tbpc 0;
rnb3kr/1p1nqppp/p3p3/2ppP3/3P1N2/2NB1Q2/PPP2PP1/R3K2R w KQ - acd 15; acn 6849556469; bm Bxh7+; fmvn 1; hmvc 0; id WCSAC.0875; pes -0.254; pv Nfxd5 exd5 Nxd5 Qd8 Rxh7 Rxh7 Bxh7+ Kxh7 Qxf7 Nc6 e6; scv 09:06:07.86; solve 0; sts TimeLimitNode; suv 09:06:07.86; tbpc 0;
3q4/p2nbk1p/2R1p1pB/1p1rPp2/8/5N1P/PPQ2PP1/7K w - - acd 16; acn 7176340340; bm Rxe6; fmvn 1; hmvc 0; id WCSAC.0876; pes +1.306; pv Rc8 Qb6 Rh8 Qc5 Qxc5 Rxc5 Rxh7+ Ke8 Rh8+ Kf7 Ra8 Rc7 Be3 Nb6 Rb8 Nc4 Rxb5 Nxe3 fxe3; scv 09:06:08.84; solve 0; sts TimeLimitNode; suv 09:06:07.85; tbpc 0;
2rnr1k1/ppq2p1p/5Qp1/2N5/4b3/2P1B1b1/PP2B1P1/3R1R1K w - - acd 16; acn 6653050555; bm Bh6; fmvn 1; hmvc 0; id WCSAC.0915; pes +2.298; pv Bg4 Bxg2+ Kxg2 Be5 Qh4 Ra8 Rd7 Qb6 Qf2 Qb5 a4 Qc4 Be2 Qa2 Nxb7 Nxb7; scv 09:06:07.89; solve 0; sts TimeLimitNode; suv 09:06:07.89; tbpc 0;
8/3R2pp/5k2/5P2/5BP1/7K/4q2P/8 b - - acd 17; acn 9737730824; bm Qf3+; fmvn 1; hmvc 0; id WCSAC.0991; pes +0.841; pv h5 Rd6+ Kf7 gxh5 Qf3+ Bg3 Qxh5+ Kg2 Qxf5 Kg1 Kg8 Rc6 Qb1+ Kf2 Qa2+ Ke3 Qb3+ Kf2 Qb2+ Ke3 Qa3+ Kd2 Kf7 Rc7+ Kg6 Ke2; scv 09:06:58.93; solve 0; sts TimeLimitNode; suv 09:06:07.93; tbpc 448462;

[Dirt]

Code: Select all

2r1r1k1/1p1qbp1p/3p1p2/p2P4/3Q4/5N2/PP2RPPP/4R1K1 w - - acd 17; acn 6948581674; bm Rxe7; fmvn 1; hmvc 0; id WCSAC.0247; pes +1.500; pv h3 Bf8 Qxf6 Rxe2 Qg5+ Bg7 Rxe2 Re8 Rxe8+ Qxe8 b3 Qc8 Nh4 h6 Qd2 b6 Qe3 b5 Nf3 Qa8; scv 09:06:08.59; solve 0; sts TimeLimitNode; suv 09:06:07.91; tbpc 0;

[d]2r1r1k1/1p1qbp1p/3p1p2/p2P4/3Q4/5N2/PP2RPPP/4R1K1 w - -
Just looking at this one, Rxe7 would be an excellent move if black takes back with his rook in order to keep his queen out of danger. However, taking with the queen actually looks quite safe.

Are you trying to verify these solutions or do you already know them to be correct?

[sje]

Are you trying to verify these solutions or do you already know them to be correct?

I'm looking for cases where the book move is less effective than the move produced by a very, very long search. The idea is to add a sufficiently better analysis move to the test suite as appropriate.

The 65,536 seconds run results will be done in about a week.

[sje]

Misc EPD opcode decodings:

pes - pawn equivalent score/symbol
scv - wall clock time used [[DDD:]HH:]MM:SS[.hh]
sts - search termination status
suv - CPU time used [[DDD:]HH:]MM:SS[.hh]
tbpc - tablebase probe count

[Dirt]

Are you trying to verify these solutions or do you already know them to be correct?

I'm looking for cases where the book move is less effective than the move produced by a very, very long search. The idea is to add a sufficiently better analysis move to the test suite as appropriate.

Then I guess at some point you plan to check the computer lines with the aid of human insight, just not yet.

[sje]

Then I guess at some point you plan to check the computer lines with the aid of human insight, just not yet.

At what point is human insight not going to be of much help? The editor of the book was a New York State Champion, more than once I believe, and a successful chess author including ghostwriting. But his insight has been proven wrong more than once including a missed mate in one in WCSAC. On the seventeen positions in question, I think Reinfeld in each case either missed a better move or made a mistake in his analysis that made his solution look rather better than it really is.

I think that a mega second search (about twelve days) with a reasonable program on a reasonable machine will produce a correct answer, or at least one of several correct answers. If there is any correct answer to be had, that is.

The end result will be a revised WCSAC that includes solution moves produced by deep analysis. I'll add these without deleting the original solution moves as that has been the tradition with suite editing (in most cases).

[sje]

Here's WCSAC.0025:
[d]8/1bp4p/p2p4/1p3rbk/4p3/1BP5/PP3PP1/3RR1K1 w - - 0 1
After 65,536 seconds:

Code: Select all

8/1bp4p/p2p4/1p3rbk/4p3/1BP5/PP3PP1/3RR1K1 w - - acd 18; acn 15540932816; bm g4+; fmvn 1; hmvc 0; id WCSAC.0025; pes +1.131; pv a4 Re5 g3 Kg6 Bc2 Kg7 axb5 Rxb5 b4 Bf6 Re3 Re5 Rde1 a5 Bxe4 Rxe4 Rxe4 Bxe4 Rxe4 Bxc3; scv 18:12:18.42; solve 0; sts TimeLimitNode; suv 18:12:15.89; tbpc 0;

[sje]

WCSAC.0027:
[d]2b5/1p3r2/p1p2rk1/P1Pp1pp1/3P2P1/4P2P/2B1K3/1R3R2 w - - 0 1

Code: Select all

2b5/1p3r2/p1p2rk1/P1Pp1pp1/3P2P1/4P2P/2B1K3/1R3R2 w - - acd 19; acn 19183013498; bm gxf5+; fmvn 1; hmvc 0; id WCSAC.0027; pes +1.058; pv Kd2 Be6 Rf2 Bc8 Rbf1 Rf8 Bd3 R6f7 Rg2 Kh6 gxf5 Bxf5 h4 g4 Rxf5 Rxf5 Bxf5 Rxf5 Rxg4 Rf2+ Kd3 Kh5; scv 18:12:15.87; solve 0; sts TimeLimitNode; suv 18:12:15.87; tbpc 0;

[sje]

WCSAC.0037:
[d]4rrk1/ppp1qppp/5n2/2b1n1N1/2B3b1/1QN5/PB4PP/4RR1K w - - 0 1

Code: Select all

4rrk1/ppp1qppp/5n2/2b1n1N1/2B3b1/1QN5/PB4PP/4RR1K w - - acd 17; acn 12908645742; bm Bxf7+; fmvn 1; hmvc 0; id WCSAC.0037; pes +2.584; pv Nd5 Nxd5 Rxe5 Qxe5 Bxe5 Rxe5 Nxf7 Re3 Qxb7 Nf6 Qxc7 Rc3 Nd6+ Rxc4 Nxc4 Rc8 Qg3 Be6 Ne5; scv 18:12:15.85; solve 0; sts TimeLimitNode; suv 18:12:15.85; tbpc 0;

[pijl]

Here's WCSAC.0025:
[d]8/1bp4p/p2p4/1p3rbk/4p3/1BP5/PP3PP1/3RR1K1 w - - 0 1
After 65,536 seconds:

Code: Select all

8/1bp4p/p2p4/1p3rbk/4p3/1BP5/PP3PP1/3RR1K1 w - - acd 18; acn 15540932816; bm g4+; fmvn 1; hmvc 0; id WCSAC.0025; pes +1.131; pv a4 Re5 g3 Kg6 Bc2 Kg7 axb5 Rxb5 b4 Bf6 Re3 Re5 Rde1 a5 Bxe4 Rxe4 Rxe4 Bxe4 Rxe4 Bxc3; scv 18:12:18.42; solve 0; sts TimeLimitNode; suv 18:12:15.89; tbpc 0;

CTD selects g4 from the start. Perhaps a material (+ possible passer) score problem?

Code: Select all

ply   score   time           nodes pv
 7.   1.023   0:00.11        45451 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 h5h4 h2h3
 8&   0.995   0:00.17        70543 g2g4 h5g4 b3e6 h7h5 g1g2 h5h4 e6f5 g4f5 g2h3 a6a5
 8.   0.995   0:00.28       113695 g2g4 h5g4 b3e6 h7h5 g1g2 h5h4 e6f5 g4f5 g2h3 a6a5
 9&   1.063   0:00.50       197399 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 h5h4 h2h3 a6a5
 9.   1.063   0:00.65       269799 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 h5h4 h2h3 a6a5
10&   0.937   0:00.98       405721 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 g5h4 e1e2 f5g4 b2b3 h4f6
10.   0.937   0:01.51       635753 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 g5h4 e1e2 f5g4 b2b3 h4f6
11&   1.021   0:02.18       920594 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 g5h4 e1e2 f5g4 d1e1 g4f3 h2g1
11.   1.021   0:03.03      1289584 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 g5h4 e1e2 f5g4 d1e1 g4f3 h2g1
12&   0.964   0:05.00      2075476 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 f5f4 e1g1 h5h4 g1g2 g5f6 d1g1 b7d5
12.   0.964   0:07.90      3311117 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 f5f4 e1g1 h5h4 g1g2 g5f6 d1g1 b7d5
13&   0.917   0:11.79      5000362 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 h5h4 a2a3 f5f4 e1g1 b7c8 d1e1 h4h3 g1g5 f4g5 e1e4
13.   0.917   0:23.81      9810679 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 h5h4 a2a3 f5f4 e1g1 b7c8 d1e1 h4h3 g1g5 f4g5 e1e4
14&   0.925   0:35.98     14926511 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 g5h4 e1f1 f5f4 h2h3 h4f6 f1e1 b7c8 h3g2 c8e6 a2a3
14.   0.925   0:55.67     23043824 g2g4 h5g4 b3e6 h7h5 e6f5 g4f5 g1h2 g5h4 e1f1 f5f4 h2h3 h4f6 f1e1 b7c8 h3g2 c8e6 a2a3
15&   0.927   1:29.39     36841941 g2g4 h5g4 b3e6 h7h5 a2a4 g4f4 e6f5 f4f5 a4b5 a6b5 g1h2 g5h4 e1e2 f5g4 d1a1 b7d5
15.   0.927   2:21.10     57902063 g2g4 h5g4 b3e6 h7h5 a2a4 g4f4 e6f5 f4f5 a4b5 a6b5 g1h2 g5h4 e1e2 f5g4 d1a1 b7d5
16&   0.942   3:46.26     93580429 g2g4 h5g4 b3e6 h7h5 a2a4 h5h4 e6f5 g4f5 a4b5 a6b5 g1h2 f5f4 e1g1 b7c6 d1a1 c6d5 a1a7 c7c6

Richard.