ECMGCP 1495: Additional solution?

[Tord Romstad]

[d]5rk1/p1p1br2/1p1p2qp/4p3/2PPn1pP/3Q2P1/PPN1P1K1/2RR2B1 b - -
The only solution move given in the EPD file is Bxh4. Glaurung wants to play Rf3, with a clearly winning score. Is this an additional solution, or do I have a new bug?

Glaurung's analysis (iMac Core Duo 2 GHz, 128 MB hash):

Code: Select all

 2     +1.39   00:00      696 Qf5 Kh2
 2     +1.57   00:00      797 h5 Kh2
 2     +1.88   00:00      922 c5 dxe5 dxe5
 3     +1.47   00:00     1263 c5 dxe5 dxe5 Kh2
 3     +1.78   00:00     3001 Bxh4 gxh4 g3 dxe5 dxe5
 4     +1.16   00:00     3455 Bxh4 gxh4 g3 Kh1 g2+ Kh2
 4     +1.35   00:00     4015 c5 dxe5 dxe5 Kh2 h5
 5     +1.41   00:00     8583 c5 dxe5 dxe5 Qd5 Nxg3 Kxg3
 5     +1.65   00:00    14188 exd4 Bxd4 Bxh4 gxh4 g3
 6     +1.25   00:00    22019 exd4 Bxd4 Bf6 Kh2 Bxd4 Nxd4 Rf2+ Kh1
 6     +2.00   00:00    33709 Rf3 exf3 gxf3+ Qxf3 Rxf3 Kxf3 Nxg3 dxe5 Bxh4 exd6
                              cxd6
 7     +2.31   00:00    63703 Rf3 exf3 gxf3+ Qxf3 Rxf3 Kxf3 Nxg3 dxe5 dxe5 Bh2
                              e4+ Ke3
 8     +2.31   00:00    97273 Rf3 exf3 gxf3+ Qxf3 Rxf3 Kxf3 Nxg3 dxe5 dxe5 Bh2
                              e4+ Ke3
 9     +2.51   00:00   184383 Rf3 exf3 gxf3+ Qxf3 Rxf3 Kxf3 Qf5+ Kg2 Qg4 Bf2
                              Bxh4 Ne3
10     +2.65   00:00   342596 Rf3 exf3 gxf3+ Qxf3 Rxf3 Kxf3 Nxg3 dxe5 Bxh4 Rd4
                              Qf5+ Kg2 Ne2
11     +3.29   00:00   728863 Rf3 exf3 gxf3+ Qxf3 Rxf3 Kxf3 Nxg3 dxe5 dxe5 Bh2
                              e4+ Ke3 Nf5+ Kf4 Nxh4 Ke3 Bg5+ Bf4
12     +3.86   00:01    1696k Rf3 exf3 gxf3+ Qxf3 Rxf3 Kxf3 Nxg3 dxe5 dxe5 Bf2
                              e4+ Kg2 Ne2+ Kf1 Nxc1 Rxc1 Kf7
13     +4.10   00:09   13187k Rf3 exf3 gxf3+ Qxf3 Bxh4 dxe5 Bxg3 Ne3 Rxf3 Kxf3
                              Bxe5 Rc2 Ng3 Rh2 Ne4
14     +4.55   00:14   21581k Rf3 exf3 gxf3+ Qxf3 Bxh4 Be3 Rxf3 Kxf3 Nxg3 Rg1
                              Qf5+ Kg2 Ne2 Kh2+ Nxg1 Rxg1+ Kf7 Nb4 Bg5
15     +5.06   00:28   41994k Rf3 exf3 gxf3+ Qxf3 Bxh4 g4 Rxf3 Kxf3 Ng3 Kg2 Ne2
                              Kh3 Nxc1 Rxc1 Bg5 Be3 exd4
16     +4.76   00:50   74986k Rf3 exf3 gxf3+ Qxf3 Bxh4 dxe5 Rxf3 Kxf3 Qf5+ Ke2
                              Nxg3+ Kd2 Qxe5 Nd4 Bg5+ Kc2 Qe4+ Kb3 Bxc1 Rxc1
                              Qd3+ Ka4 Ne4
17     +6.31   02:11  199736k Rf3 exf3 gxf3+ Qxf3 Bxh4 dxe5 Rxf3 Kxf3 Nxg3 Rd4
                              Qf5+ Rf4 Qd3+ Kg4 Ne2 Ne3 Bg5 Rd1 Qg6 Rff1 Bxe3+
                              Kf3
Nodes: 204325840
Nodes/second: 1512852
Best move: Rf3
Ponder move: exf3

Tord

[Uri Blass]

It is clearly an additional solution.
The simplest way to check if this is an additional solution is to try other programs after Rf3 and see that they also can see winning score for black.

[Eelco de Groot]

Hello Tord,

I suppose both moves are winning quickly. Pro Deo 1.5 Test16 chooses 1.Bxh4:

5rk1/p1p1br2/1p1p2qp/4p3/2PPn1pP/3Q2P1/PPN1P1K1/2RR2B1 b - -

00:00:00 1.00 3.53 1..exd4 2.Nxd4
00:00:00 1.07 3.74 1..d5 2.dxe5
00:00:00 1.10 3.79 1..Qf5
00:00:00 1.11 4.23 1..a6
00:00:00 1.12 4.26 1..c5
00:00:00 1.19 4.29 1..c6
00:00:00 1.30 4.31 1..a5
00:00:00 2.00 1.91 1..a5 2.Kh1
00:00:00 2.02 2.53 1..c5 2.Kh1 cxd4
00:00:01 2.19 2.64 1..d5 2.Kh1 dxc4
00:00:03 3.00 2.64 1..d5 2.Kh1 dxc4
00:00:05 3.17 2.71 1..exd4 2.Kh1 Bg5
00:00:07 4.00 2.65 1..exd4 2.Kh1 Bg5 3.Ra1
00:00:08 4.14 2.65 1..Bxh4
00:00:08 4.14 2.94 1..Bxh4 2.gxh4 Rf3 3.Be3 Rh3 4.dxe5
00:00:08 5.00 2.70 1..Bxh4 2.gxh4 Rf3 3.Be3 exd4 4.Qxd4 Rh3
00:00:08 5.01 2.92 1..exd4 2.Kh2 Bg5 3.Ra1
00:00:08 5.25 2.92 1..Bg5
00:00:08 5.25 4.02 1..Bg5 2.Ra1 Be7 3.dxe5 dxe5
00:00:08 6.00 2.30 1..Bg5 2.Rb1 Qh7 3.Kh1
00:00:08 6.01 2.79 1..exd4 2.Nxd4 c5 3.Nc6 d5 4.Nxe7 Rxe7 5.cxd5
00:00:09 7.00 2.52 1..exd4 2.Nxd4 a6 3.Kh1 Rf2 4.b4 Bg5
00:00:09 7.02 2.52 1..Rf3
00:00:09 7.02 3.82 1..Rf3 2.exf3 gxf3 3.Kf1 Qxg3 4.Rd2 Nxd2 5.Qxd2 Qh3 6.Ke1 Bxh4 7.Kd1 Qf1 8.Ne1 Qxg1
00:00:09 8.00 3.89 1..Rf3 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 c5 5.dxe5 Nxg3
00:00:10 9.00 3.77 1..Rf3 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.dxe5 Bxh4 6.Rd4
00:00:12 10.00 3.77 1..Rf3 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.dxe5 Bxh4 6.Rd4
00:00:16 11.00 3.91 1..Rf3 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Bf6 5.dxe5 Bxe5 6.h5
00:00:19 11.02 3.91 1..Bxh4
00:00:35 12.00 4.50 1..Rf3 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.dxe5 Bxh4 6.Rd4 Qf5 7.Kg2 Ne4 8.exd6 cxd6
00:00:44 12.01 4.50 1..Bxh4
00:00:52 12.01 4.54 1..Bxh4 2.gxh4 Rf3 3.dxe5 Rxd3 4.exd3 Rf3 5.Bd4 Qh5 6.dxe4 Qxh4
00:01:22 13.00 4.68 1..Bxh4 2.gxh4 Rf3 3.Ne3 Rg3 4.Kh1 Rf2 5.Bh2 Rh3 6.Nf1 Rxd3 7.Rxd3 Rxe2
00:04:50 14.00 5.15 1..Bxh4 2.gxh4 Rf3 3.Rf1 Rxd3 4.Rxf8 Kxf8 5.exd3 Nd2 6.Ne1 Qf6

After 1. Rf3

[Event "?"]
[Site "?"]
[Date "2007.06.28"]
[White "?"]
[Black "?"]
[Result "*"]
[FEN "5rk1/p1p1b3/1p1p2qp/4p3/2PPn1pP/3Q1rP1/PPN1P1K1/2RR2B1 w - -"]

1..Rf3

5rk1/p1p1b3/1p1p2qp/4p3/2PPn1pP/3Q1rP1/PPN1P1K1/2RR2B1 w - -

00:00:00 1.00 0.07 2.exf3 gxf3 3.Kh1 Nxg3 4.Kh2
00:00:00 2.00 -0.33 2.exf3 gxf3 3.Kh1 Qxg3
00:00:00 3.00 -0.33 2.exf3 gxf3 3.Kh1 Qxg3
00:00:00 4.00 -5.20 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.dxe5 dxe5
00:00:00 4.03 -4.51 2.Be3 Rxg3 3.Kh2 Rf2 4.Kh1 Rh3 5.Kg1 g3
00:00:00 5.00 -4.71 2.Be3 Rxg3 3.Kh2 Bxh4 4.dxe5 Rh3 5.Kg1 Bf2 6.Bxf2 Rxd3 7.exd6
00:00:00 5.01 -4.71 2.exf3
00:00:00 5.01 -4.71 2.exf3
00:00:00 5.01 -3.61 2.exf3 gxf3 3.Kf1 Qxg3 4.Rd2 Nxd2 5.Qxd2 Qh3 6.Ke1 Bxh4 7.Kd1 Qf1 8.Ne1 Qxg1
00:00:00 6.00 -3.62 2.exf3 gxf3 3.Kf1 Qxg3 4.Rd2 Qh3 5.Ke1 Bxh4 6.Rf2 Qh1 7.Qf1 Nxf2
00:00:00 7.00 -3.64 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.dxe5 Bxh4 6.Rd4
00:00:01 8.00 -3.64 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.dxe5 Bxh4 6.Rd4
00:00:02 9.00 -3.95 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.Bh2 e4 6.Ke3 Ne2
00:00:06 10.00 -4.30 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.Bf2 Bxh4 6.Bxg3 Qxg3 7.Ke4 Qg6 8.Ke3
00:00:21 11.00 -4.37 2.exf3 gxf3 3.Qxf3 Rxf3 4.Kxf3 Nxg3 5.dxe5 Bxh4 6.Rd4 Qh5 7.Rg4 Kh8 8.exd6 Bg5 9.Kxg3 Bxc1
00:02:32 12.00 -4.41 2.exf3 gxf3 3.Qxf3 Bxh4 4.g4 Rxf3 5.Kxf3 Ng5 6.Ke2 Qe4 7.Kd2 Nf3 8.Kc3 Bg5 9.Kb3 Bxc1
00:06:52 13.00 -4.65 2.exf3 gxf3 3.Qxf3 Bxh4 4.dxe5 Rxf3 5.Kxf3 Qf5 6.Ke2 Nxg3 7.Kd2 Ne4 8.Ke2

Pro Deo 1.5 Test16.eng 256 Mb TT Athlon 2009 MHz

As you can see the last PV is coming more in line with what Glaurung is computing so I think the additional solution from Glaurung is correct!

Regards, Eelco

[Tord Romstad]

Uri and Eelco,

Thanks for confirming that Rf3 also wins! I agree that testing with another program is generally the easiest way to verify things like this, but today I didn't have any other programs available, and not much time (I'm leaving for the OPCCC later today).

Tord

[Eelco de Groot]

Good Luck in Lodz with Glaurung Tord! Have a safe trip! Will the iMac notebook have to do as your tournament machine? Or can you maybe borrow some machine from the Tournament organization. I mean against Rybka if it had been playing it would not have been enough...

Regards, Eelco