Page 1 of 1
Mate in 15 (Difficult for most programs)
Posted: Sat May 19, 2012 7:26 pm
by biscuit1953
2N5/8/8/p3PP2/2P1P3/p5B1/P2KP1pP/5krb b - - 1 1
Chessmaster 11 can only find mate in 20 after 30 minutes on my Acer Aspire laptop.
Re: Mate in 15 (Difficult for most programs)
Posted: Sat May 19, 2012 8:13 pm
by zullil
Since Black can only play a4, let's just consider this position:
2N5/8/8/4PP2/p1P1P3/p5B1/P2KP1pP/5krb w - - 0 1
Critter-1.4 (8 threads, 4 GB hash) solves quickly:
Code: Select all
29/60 00:26 158858134 6066760 +M15 Bf4 Kf2 Be3+ Kf1 Kd3 Ke1 Bxg1 Kf1 Be3 g1=Q Bxg1 Bg2 e6 Bh3 e7 Kxg1 e8=Q Kg2 Qh5 Kxh2 Ke3 Kg3 Qf3+ Kh4 Kf4 Bxf5 Qg3+ Kh5 Qg5#
Re: Mate in 15 (Difficult for most programs)
Posted: Sat May 19, 2012 8:24 pm
by biscuit1953
I must admit that I have had little experience with UCI programs and never used Critter before but, it certainly seems to be among the strongest.
Re: Mate in 15 (Difficult for most programs)
Posted: Sat May 19, 2012 8:33 pm
by zullil
biscuit1953 wrote:I must admit that I have had little experience with UCI programs and never used Critter before but, it certainly seems to be among the strongest.
I don't use Houdini, but it might be faster on this one. Perhaps others will post their results soon.
Re: Mate in 15 (Difficult for most programs)
Posted: Sat May 19, 2012 10:58 pm
by biscuit1953
I'm starting to think that Critter is the only program that can find mate in 15. Several engines I tested have failed to find it in including Rybka 2.2.
Re: Mate in 15 (Difficult for most programs)
Posted: Sun May 20, 2012 3:02 am
by zullil
biscuit1953 wrote:I'm starting to think that Critter is the only program that can find mate in 15. Several engines I tested have failed to find it in including Rybka 2.2.
Stockfish-2.2.2 (8 threads, 2GB hash) finds mate-in-15, but not quickly:
Code: Select all
info depth 34 seldepth 50 score mate 15 nodes 139722721779 nps 11740061 time 11901362 multipv 1 pv g3f4 f1f2 f4e3 f2f1 d2d3 f1e1 e3g1 e1f1 g1c5 g2g1r c5g1 f1g1 e5e6 h1e4 d3e4 g1f2 e6e7 f2e2 e7e8q e2d2 e4d4 d2c2 e8e1 c2b2 e1d1
I'm confident that other engines will solve this too.
Re: Mate in 15 (Difficult for most programs)
Posted: Sun May 20, 2012 6:38 pm
by ernest
zullil wrote:Since Black can only play a4, let's just consider this position:
Houdini 2.0c finds it quite fast (on old dualcore):
2N5/8/8/4PP2/p1P1P3/p5B1/P2KP1pP/5krb w - - 0 1
Analysis by Houdini 2.0c w32 Nlm:
........
1.Bf4 Kf2 2.Be3+ Kf1 3.Kd3 Ke1 4.Bxg1 Kf1 5.Be3 g1Q 6.Bxg1 Bg2 7.Be3 Ke1 8.e6 Bf1 9.e7 Bxe2+ 10.Kd4 Bh5 11.f6 Kd1 12.Nd6 Kc2 13.f7 Kb2 14.f8Q Ka1 15.Qf1+ Kxa2 16.Kd3 Bg4 17.Qf2+ Be2+ 18.Qxe2+ Ka1 19.Bd4+ Kb1 20.Qc2#
+- (#20) Depth: 29/58 00:00:07 29451kN
1.Bf4 Kf2 2.Be3+ Kf1 3.Kd3 Ke1 4.Bxg1 Kf1 5.Be3 g1Q 6.Bxg1 Bg2 7.Be3 Bh3 8.e6 Bg4 9.e7 Bh5 10.f6 Ke1 11.Nd6 Bxe2+ 12.Kd4 Bh5 13.f7 Bxf7 14.Nxf7 Kf1 15.e8Q Ke2 16.Qg8 Kf3 17.Qg3+ Ke2 18.Qf2+ Kd1 19.Qd2#
+- (#19) Depth: 30/58 00:00:11 47677kN
1.Bf4 Kf2 2.Be3+ Kf1 3.Kd3 Ke1 4.Bxg1 Kf1 5.Be3 g1Q 6.Bxg1 Bg2 7.e6 Kxg1 8.e7 Kf2 9.e8Q Ke1 10.Qe5 Kf2 11.Qg3+ Kf1 12.Ke3 Kg1 13.f6 Kh1 14.Kf2 Bxe4 15.Qg1#
+- (#15) Depth: 31/58 00:00:18 82438kN