Really nice puzzle from Peter's suite

[Jouni]

[d]3rqQ2/3ppppp/7k/R7/8/8/1PPP2K1/8 w - - 0 1

Rems solves instantly

1.Ra3 Qxf8 2.Rh3+ Kg5 3.Rg3+ Kf5 4.Rf3+ Ke6 5.Re3+ Kd6 6.Rd3+ Kc7 7.Rc3+ Kb8 8.Rb3+ Ka8 9.Ra3+ Kb7 10.Rb3+
= (0.00) Depth: 14/32 00:00:00 1146kN, tb=137
1.Ra3 Qxf8 2.Rh3+ Kg5 3.Rg3+ Kf5 4.Rf3+ Ke6 5.Re3+ Kd6 6.Rd3+ Kc7 7.Rc3+ Kb8 8.Rb3+ Ka8 9.Ra3+ Kb7 10.Rb3+
= (0.00) Depth: 15/32 00:00:01 2246kN, tb=280
1.Ra3 Qxf8 2.Rh3+ Kg5 3.Rg3+ Kf5 4.Rf3+ Ke6 5.Re3+ Kd6 6.Rd3+ Kc7 7.Rc3+ Kb8 8.Rb3+ Ka8 9.Ra3+ Kb7 10.Rb3+
= (0.00) Depth: 16/32 00:00:01 3266kN, tb=457

Is it sure, that Ra3 is only drawing move?

[peter]

Is it sure, that Ra3 is only drawing move?

Pity, I don't know the originary source by my own neither (otherwise it would be given with .epd) but you could try out especially with Rems, that solves instantly because of its internal MultiPV- mode, why don't you let the engine show more then one output- line?

Crystal solves SMP (26 threads of the 16x3.5 GHz CPU, having other threads busy at the moment too) with MultiPV=3 at once too, here's it's output started with empty hash:

Engine: Crystal251008 (8192 MB)
von the Stockfish developers (see AUTHORS f
Using 26 threads
-----
15 0:01 -0.15 1.Ta3 Dxf8 2.Th3+ Kg5 3.Tg3+ Kf5 4.Tf3+ Ke5 5.Te3+ Kf4 6.Tf3+ (42.536.717) 23320
15 0:01 -3.55 1.Dxe8 Txe8 2.c4 Tc8 3.d3 d6 4.b4 Kg6 5.Kf3 h5 6.Td5 Kh6 7.b5 Tb8 8.Tf5 f6 9.d4 g6 10.Tc5 h4 11.Kg2 (42.536.717) 23320
15 0:01 -4.40 1.Ta6+ d6 2.Dxe8 Txe8 3.b4 Tb8 4.Ta4 Kg5 5.c4 Kf6 6.b5 e5 7.d3 h5 8.Ta7 Tc8 9.Kg3 g5 10.Ta1 Ke6 11.Th1 Th8 12.Te1 f5 13.d4 e4 14.c5 (42.536.717) 23320
-----
-----
23 2:19 -0.20 1.Ta3 Dxf8 2.Th3+ Kg5 3.Tg3+ Kf4 4.Tf3+ Ke4 5.Te3+ Kd4 6.Td3+ Kc4 7.Tc3+ Kb5 8.Tb3+ Kc5 9.Tc3+ (2.186.908.819) 15693
22 2:19 -4.95 1.Dxe8 Txe8 2.b4 d6 3.c4 Tc8 4.d3 Kg6 5.b5 h5 6.Kf3 f5 7.Ta4 Kf6 8.d4 g5 9.b6 Ke6 10.Ke2 Tb8 11.c5 Kd5 12.Ta5 (2.186.908.819) 15693
22 2:19 -5.55 1.Ta6+ d6 2.Dxe8 Txe8 3.b4 Tb8 4.c3 Kg5 5.Ta7 Kf6 6.Kg3 h5 7.Kh4 Th8 8.Kg3 g5 9.c4 Tb8 10.b5 e5 11.Ta3 e4 (2.186.908.819) 15693

And after a little Forward- Backward single primary and back to MultiPV=2:

3rqQ2/3ppppp/7k/R7/8/8/1PPP2K1/8 w - -

Engine: Crystal251008 (8192 MB)
von the Stockfish developers (see AUTHORS f

25 1:23 -0.10 1.Ta3 Dxf8 2.Th3+ Kg5 3.Tg3+ Kh5
4.Th3+ (1.383.183.259) 16506
TB:3.048.321

24 1:23 -5.30 1.Dxe8 Txe8 2.b3 Kg6 3.Kf3 Tc8 4.c3 d6
5.Ta7 Kf6 6.Kg4 h6 7.Ta1 Kg6 8.Ta5 f6
9.c4 Kf7 10.Ta2 g6 11.d4 h5+ 12.Kh4 Tb8
13.Tb2 g5+ 14.Kxh5 (1.383.183.259) 16506
TB:3.048.321

So, I guess yes, Ra3 is the only drawing move.

Edit: Could well be, I saw the position at one of the first times here as the nr.100 collected by lech (Marek) way back then:

viewtopic.php?p=941510#p941510

[peter]

Edit- time over, but I yet wanted to mention, that Sting solves that MultiPV=1 at once too, even single threaded (3.5GHz, 1024Mb hash):

Code: Select all

3rqQ2/3ppppp/7k/R7/8/8/1PPP2K1/8 w - - 0 1

Analysis by Sting Black Hole 10x (based on Stockfish 2.1.1) 64bit:

1.Dxe8 Txe8 2.b4 Kg6 3.b5 Tc8 4.c3 Tb8 5.c4 Kf5 6.Kf3 d5 
  Schwarz steht deutlich besser.  
  -+ (-1.73)  Tiefe: 10   00:00:00
...
1.Dxe8 Txe8 2.b4 Kg6 3.Kf3 Tb8 4.c3 h5 5.Kf4 h4 6.Kg4 Th8 7.Tg5+ Kf6 8.Tf5+ Ke6 9.Tf3 h3 10.Te3+ Kd5 11.Td3+ Kc4 12.Td4+ Kb5 
  Schwarz steht deutlich besser.  
  -+ (-2.34)  Tiefe: 20   00:00:05  12641kN
1.Ta3 Dxf8 2.Th3+ Kg5 3.Tg3+ Kf6 4.Tf3+ Kg6 5.Tg3+ Kf6 
  Schwarz steht deutlich besser.  
  -+ (-2.26 ++)  Tiefe: 21   00:00:07  18775kN
...
1.Ta3 Dxf8 2.Th3+ Kg5 3.Tg3+ Kf6 4.Tf3+ Kg6 5.Tg3+ Kf6 
  Schwarz steht etwas besser.  
  =/+ (-0.58 ++)  Tiefe: 21   00:00:08  21877kN
1.Ta3 Dxf8 2.Th3+ Kg5 3.Tg3+ Kf6 4.Tf3+ Kg6 5.Tg3+ Kf5 6.Tf3+ Ke6 7.Te3+ Kd6 8.Td3+ Kc7 9.Tc3+ Kb7 10.Tb3+ Kc6 11.Tc3+ Kb5 
  Die Stellung ist ausgeglichen.  
  = (-0.04)  Tiefe: 21   00:00:08  22436kN
...
1.Ta3 Dxf8 2.Th3+ Kg5 3.Tg3+ Kf6 4.Tf3+ Kg6 5.Tg3+ Kf5 6.Tf3+ Ke6 7.Te3+ Kd6 8.Td3+ Kc7 9.Tc3+ Kb7 10.Tb3+ Kc6 11.Tc3+ Kb5 12.Tb3+ Kc5 13.Tc3+ Kd5 14.Td3+ Ke5 15.Te3+ Kf4 16.Tf3+ Ke4 17.Te3+ Kd4 18.Td3+ Kc4 19.Tc3+ Kb4 
  Die Stellung ist ausgeglichen.  
  = (-0.04)  Tiefe: 30   00:00:25  67922kN

[Norm Pollock]

Possible missing castling permissions in Peter's test suite. Castling can affect the analysis for determining the best move, even if castling is not itself the best move. A missing castling move denies that side the option to castle.

Code: Select all

Line 113
rn2kr2/1pp1b1p1/4p3/1Pp1P2p/p1P1NP1P/P7/1B4P1/3RR1K1 w - - bm g3; hmvc 0; id "Opening, Hiarcs_! ?!._? 113_?_512, 113";
rn2kr2/1pp1b1p1/4p3/1Pp1P2p/p1P1NP1P/P7/1B4P1/3RR1K1 w q - bm g3; hmvc 0; id "Opening, Hiarcs_! ?!._? 113_?_512, 113";

Line 138
4k1br/1K1p1n1r/2p2pN1/P2p1N2/2P3pP/5B2/P2P4/8 w - - bm Kc8; hmvc 0; id "ERET 103_?_512, 138";
4k1br/1K1p1n1r/2p2pN1/P2p1N2/2P3pP/5B2/P2P4/8 w k - bm Kc8; hmvc 0; id "ERET 103_?_512, 138";

Line 183
r1bq1rk1/1p1nbppp/p3p3/2ppP3/5B1P/2PBP3/PP1N1PP1/R2QK2R w - - bm Bxh7+; hmvc 0; id "Hard, Positions_Corbit, Dann_2022, 183";
r1bq1rk1/1p1nbppp/p3p3/2ppP3/5B1P/2PBP3/PP1N1PP1/R2QK2R w KQ - bm Bxh7+; hmvc 0; id "Hard, Positions_Corbit, Dann_2022, 183";

[peter]

Possible missing castling permissions in Peter's test suite. Castling can affect the analysis for determining the best move, even if castling is not itself the best move. A missing castling move denies that side the option to castle.

Code: Select all

Line 113
rn2kr2/1pp1b1p1/4p3/1Pp1P2p/p1P1NP1P/P7/1B4P1/3RR1K1 w - - bm g3; hmvc 0; id "Opening, Hiarcs_! ?!._? 113_?_512, 113";
rn2kr2/1pp1b1p1/4p3/1Pp1P2p/p1P1NP1P/P7/1B4P1/3RR1K1 w q - bm g3; hmvc 0; id "Opening, Hiarcs_! ?!._? 113_?_512, 113";

Line 138
4k1br/1K1p1n1r/2p2pN1/P2p1N2/2P3pP/5B2/P2P4/8 w - - bm Kc8; hmvc 0; id "ERET 103_?_512, 138";
4k1br/1K1p1n1r/2p2pN1/P2p1N2/2P3pP/5B2/P2P4/8 w k - bm Kc8; hmvc 0; id "ERET 103_?_512, 138";

Line 183
r1bq1rk1/1p1nbppp/p3p3/2ppP3/5B1P/2PBP3/PP1N1PP1/R2QK2R w - - bm Bxh7+; hmvc 0; id "Hard, Positions_Corbit, Dann_2022, 183";
r1bq1rk1/1p1nbppp/p3p3/2ppP3/5B1P/2PBP3/PP1N1PP1/R2QK2R w KQ - bm Bxh7+; hmvc 0; id "Hard, Positions_Corbit, Dann_2022, 183";

Thanks for pointing that out, Norm, even if in the given examples I don't see any change of bm by giving the missing syntax, regards