Sorry, I missed this reply. This isn't a separate version, I'm still looking for a way to include these changes in Stockfish with hopefully an ELO increase as well. No success so far. The null move change is about -10 elo in self play as presented. The evaluation change seems to be a 0 ELO change.pichy wrote:Can you make this version available as a download?
What engine can see the best move Bf8+!
Moderators: hgm, Rebel, chrisw
-
- Posts: 568
- Joined: Tue Dec 12, 2006 10:10 am
- Full name: Gary Linscott
Re: Can you make this version available for more testing ?
-
- Posts: 1136
- Joined: Sun Feb 14, 2010 10:02 pm
Re: What engine can see the best move Bf8+!
The second position (set by me) is much simpler for engines.lech wrote:The good result of Houdini and Critter can be accidental. It seems that both the engines are not experts in zugzwang positions.
Stockfish can solve it on a very high depth (time) thanks to problems with VALUE_DRAW = 0;
My Sting solves it in 19 mintes on a very slow computer.
[d] 6R1/8/2pB3k/2P4p/5p1q/5P2/4P1K1/8 w - - 0 1
Sting SF 2.0 (beta):I modified this position to try to prove that the result of Houdini and Critter can be accidental.Code: Select all
33/46 09:11 493.008.329 894.091 0,00 Rg8e8 Qh4g3+ Kg2f1 Qg3h3+ Kf1g1 Kh6g6 Bd6xf4 Qh3h4 Bf4d6 Kg6f7 Re8e4 Qh4e1+ Kg1g2 h5h4 Re4e7+ Kf7f6 Re7e8 Qe1c1 Re8e4 h4h3+ Kg2xh3 Qc1f1+ Kh3g4 Qf1g1+ Bd6g3 Qg1xc5 Bg3h4+ Kf6f7 Re4e7+ Kf7g6 Re7e6+ Kg6f7 Re6f6+ Kf7g7 34+ 19:00 887.273.483 777.766 +0,08 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 Qe1xe2+ Kg2h1 Qe2c4 Bf6a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 34+ 19:01 887.722.040 777.648 +0,16 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 Qe1xe2+ Kg2h1 Qe2c4 Bf6a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 34+ 19:02 888.489.821 777.511 +0,28 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 Qe1c1 Kg2h2 Qc1c4 Kh2h1 Qc4c2 Bf6a1 Qc2c4 e2e4 f4xe3/ep Rg7g4+ 34+ 19:05 890.489.703 777.094 +0,73 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 Qe1c1 Kg2h2 Qc1c4 Kh2h1 Qc4c2 Bf6a1 Qc2c4 e2e4 f4xe3/ep Rg7g4+ 34+ 19:06 891.013.252 777.000 +1,14 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 h5h4 Kg2h2 34+ 19:11 894.206.774 776.589 +2,67 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 h5h4 Kg2h2 34+ 19:19 900.172.186 776.114 +4,04 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 Qe1xe2+ Kg2h1 Qe2e3 Rg7e7+ Kh8g8 Re7xe3 f4xe3 Kh1g2 Kg8f7 Bf6d4 e3e2 Kg2f2 h5h4 Kf2xe2 Kf7g6 Ke2f2 Kg6h5 Kf2g2 Kh5g5 Bd4e3+ Kg5f6 Kg2h3 Kf6e5 Kh3xh4 Ke5e6 Kh4g4 Ke6d5 f3f4 Kd5e6 f4f5+ Ke6e5 Kg4g5 34+ 19:32 909.108.248 775.679 +6,09 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 Qe1xe2+ Kg2h1 Qe2xf3+ Kh1h2 Qf3e3 Bf6a1 h5h4 Ba1b2 Qe3xc5 Rg7g5+ 34/55 19:38 913.254.956 775.258 +6,98 Bd6f8+ Kh6h7 Rg8g7+ Kh7h8 Bf8e7 Qh4e1 Be7f6 Qe1xe2+ Kg2h1 Qe2e6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6h6 Rg7g6+ Kh8h7 Rg6xh6+ Kh7xh6 Kh1xh2 Kh6g5 Kh2h3 Kg5f5 Kh3h4 Kf5e6 Kh4g4 Ke6d7 Kg4xf4 Kd7c8 Bd4e3 Kc8b7 Kf4g4 Kb7a6 f3f4 Ka6b7 f4f5
[d] 7k/6R1/2p5/2P4p/5p1q/5P2/1B2P3/6K1 b - - 0 1
After 1...Qe1+ 2.Kh2 Qh4+ 3.Kg2 Qe1 4.Bf6 it leads to known position from previous solution.
But now, the main zugzwang position
[d] 7k/6R1/2p2B2/2P4p/5p2/5P2/4P1K1/4q3 b - - 0 1
appears after the logic move Bf6.
Sting solves it quickly.
[d] 7k/6R1/2p5/2P4p/5p1q/5P2/1B2P3/6K1 b - - 0 1
Sting SF 2.0 (beta):What about others ?Code: Select all
37/26 01:48 73.698.635 677.588 0,00 Qh4e1+ Kg1h2 Qe1h4+ Kh2g1 38- 01:56 79.491.692 682.695 -0,08 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1xe2+ Kg2h1 Qe2e6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6b2 Qe6a2 Bb2d4 Qa2c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 38- 01:57 80.048.240 682.987 -0,16 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1xe2+ Kg2h1 Qe2e6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6b2 Qe6a2 Bb2d4 Qa2c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 38- 02:01 83.281.432 684.474 -0,74 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1xe2+ Kg2h1 Qe2e6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6b2 Qe6a2 Bb2d4 Qa2c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 38- 02:05 85.831.229 685.536 -1,15 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1c1 Kg2h2 Qc1c4 Bf6b2 Qc4a2 Bb2d4 Qa2c4 Bd4a1 Qc4c2 Kh2h1 Qc2c4 e2e4 h5h4 Ba1f6 h4h3 Bf6a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 Qc4xc5 Rg7g5+ 38- 02:12 91.339.350 689.514 -1,76 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1xe2+ Kg2h1 Qe2e6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 38- 02:16 94.249.334 690.471 -2,67 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1c1 Kg2h2 Qc1c2 Bf6a1 Qc2c4 Kh2h1 Qc4c2 e2e4 Qc2c4 Ba1f6 Qc4e6 e4e5 Qe6c4 e5e6 Qc4xe6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 38- 02:19 96.722.871 691.569 -4,04 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1c1 Kg2h2 Qc1c2 Bf6a1 Qc2c4 Kh2h1 Qc4c2 e2e4 Qc2c4 Ba1f6 Qc4e6 e4e5 Qe6c4 e5e6 Qc4xe6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 38- 02:39 110.893.568 697.373 -6,09 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1xe2+ Kg2h1 Qe2e6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6h6 Rg7g6+ Kh8h7 Rg6xh6+ Kh7xh6 Kh1xh2 Kh6h7 Kh2h3 Kh7g6 Kh3g4 Kg6f7 Kg4xf4 Kf7g6 Kf4g3 Kg6f5 Bd4e3 Kf5e5 f3f4+ Ke5f5 Kg3f3 Kf5g6 Kf3g4 Kg6h6 f4f5+ Kh6g7 Kg4g5 Kg7g8 Be3d4 Kg8f7 f5f6 Kf7f8 Kg5g6 Kf8g8 f6f7+ Kg8f8 Bd4g7+ 38- 03:06 130.786.713 700.679 -9,17 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1xe2+ Kg2h1 Qe2e6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 Qc4b5 Rg7b7+ Kh8g8 Rb7xb5 c6xb5 c5c6 38/58 03:11 134.104.611 701.541 -7,75 Qh4e1+ Kg1h2 Qe1h4+ Kh2g2 Qh4e1 Bb2f6 Qe1xe2+ Kg2h1 Qe2e6 Bf6d4 Qe6c4 Bd4a1 h5h4 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h4h3 Ba1f6 Qc4e6 Bf6d4 Qe6c4 Bd4a1 h3h2 Ba1f6 Qc4e6 Bf6d4 Qe6h6 Rg7g6+ Kh8h7 Rg6xh6+ Kh7xh6 Kh1xh2 Kh6h7 Bd4e5 Kh7h6 Be5xf4+ Kh6g7 Bf4g5 Kg7f7 Bg5h4 Kf7e8 Kh2h3 Ke8d7 Bh4g5 Kd7c8 f3f4 Kc8c7 f4f5 Kc7d7 Kh3g4 Kd7c7 f5f6 Kc7d7 f6f7
The zugzwang layout is just ready.
Additionally, The move Bf6 is one move earlier.
Sting needs 10x less time and it is a logical reaction.
Thus, Houdini and Critter should solve the second (my) position immediately.
Unfortunately ...... , but the move Be7->f6 is only a tactical solution for both!
Maybe, I can't be friendly, but let me be useful.
-
- Posts: 1136
- Joined: Sun Feb 14, 2010 10:02 pm
Re: What engine can see the best move Bf8+!
Gary, your changes don't work to help in solwing of zugzwang positions.gladius wrote:Well, with the Stockfish modifications posted above (which I'm still testing to try and find a +ELO change), here is what it looks like:lech wrote:What about others ?
Code: Select all
Analysis by Stockfish 120410 64bit 326 64bit: 1...Qe1+ 2.Kh2 Qh4+ 3.Kg1 = (0.00) Depth: 32/20 00:00:02 6075kN 1...Qe1+ 2.Kh2 Qh4+ 3.Kg2 Qe1 4.Bf6 Qc1 5.Kh2 Qc2 6.Ba1 h4 7.Kh1 Qc4 8.e4 h3 9.Bf6 Qe6 10.e5 Qc4 11.e6 Qxe6 12.Bd4 Qc4 13.Ba1 h2 14.Bf6 Qe6 15.Bd4 Qc4 16.Ba1 = (0.08) Depth: 33/43 00:00:02 7868kN 1...Qe1+ 2.Kh2 Qh4+ 3.Kg2 Qe1 4.Bf6 Qc1 5.Kh2 Qc2 6.Ba1 h4 7.Kh1 h3 8.Kh2 Qc4 9.Bf6 Qc2 10.Bd4 Qd1 11.Bb2 Qc2 12.Bf6 Qc4 13.e4 Qe6 14.e5 = (0.16) Depth: 33/45 00:00:03 11101kN 1...Qe1+ 2.Kh2 Qh4+ 3.Kg2 Qe1 4.Bf6 Qc1 5.Kh2 Qc2 6.Ba1 Qc4 7.Kh1 Qc2 8.e4 Qc4 9.Bf6 Qe6 10.e5 Qc4 11.Kh2 Qxc5 12.e6 Qf5 13.Ba1 ² (0.28) Depth: 33/54 00:00:03 11702kN 1...Qe1+ 2.Kh2 Qh4+ 3.Kg2 Qe1 4.Bf6 Qc1 5.Kh2 Qc2 6.Ba1 Qc4 7.Kh1 Qc2 8.e4 Qc4 9.Bf6 Qe6 10.e5 Qc4 11.Kh2 Qc3 12.Kh3 Qxc5 13.Kh4 Qe3 14.Kg5 c5 15.Kh6 ... +- (85.58) Depth: 37/77 00:01:27 263mN 1...Qe1+ 2.Kh2 Qh4+ 3.Kg2 Qe1 4.Bf6 Qxe2+ 5.Kh1 Qe6 6.Bd4 Qc4 7.Ba1 h4 8.Bf6 Qe6 9.Bd4 Qc4 10.Ba1 h3 11.Bf6 Qe6 12.Bd4 Qc4 13.Ba1 h2 14.Bf6 Qe6 15.Bd4 Qe3 16.Rd7+ Kg8 17.Bxe3
It is the same accidental and side tactical effect as in Houdini and Critter.
Maybe, I can't be friendly, but let me be useful.
-
- Posts: 2564
- Joined: Thu Mar 09, 2006 3:04 am
Re: Can you make this version available for more testing ?
I was not referring to an Elo increase, it is not just about Elo If Stockfish solve this position and other similar it is an improvement by itself. Lets not forget that not all the people who buy or get an engine all that they are looking for is an Elo increasegladius wrote:Sorry, I missed this reply. This isn't a separate version, I'm still looking for a way to include these changes in Stockfish with hopefully an ELO increase as well. No success so far. The null move change is about -10 elo in self play as presented. The evaluation change seems to be a 0 ELO change.pichy wrote:Can you make this version available as a download?
PS: What do you do with an engine that play stronger, but can not find simple position like this one
-
- Posts: 2564
- Joined: Thu Mar 09, 2006 3:04 am
Re: What engine can see the best move Bf8+!
Sting is a very interesting open source software!
Sting SF is open source code software (thread) based on Stockfish (2.1.1) without exe-files.
Sting SF is to help detect, impossible to find or time consuming for engines, sacrifices.
Unfortunately, but just Stockfish is very weak to find difficult sacrifices and thus is not a good tool to make any analysis.
Sting SF project should very help you to get a reliable result in analysis.
Sting SF uses the option : “Sacrifice detection” (0 – 8, 8 = default, 0 = disabled).
BTW. I tested a strength (Sacrifice detection = 8). Sting seems to be stronger than Stockfish.
Sting SF is open source code software (thread) based on Stockfish (2.1.1) without exe-files.
Sting SF is to help detect, impossible to find or time consuming for engines, sacrifices.
Unfortunately, but just Stockfish is very weak to find difficult sacrifices and thus is not a good tool to make any analysis.
Sting SF project should very help you to get a reliable result in analysis.
Sting SF uses the option : “Sacrifice detection” (0 – 8, 8 = default, 0 = disabled).
BTW. I tested a strength (Sacrifice detection = 8). Sting seems to be stronger than Stockfish.
-
- Posts: 568
- Joined: Tue Dec 12, 2006 10:10 am
- Full name: Gary Linscott
Re: What engine can see the best move Bf8+!
They are not intended to help with zugzwang position, just adding understanding about understanding revealed checks.lech wrote:Gary, your changes don't work to help in solwing of zugzwang positions.
It is the same accidental and side tactical effect as in Houdini and Critter.
I haven't looked at your patches for Sting though. Are they available anywhere?
-
- Posts: 568
- Joined: Tue Dec 12, 2006 10:10 am
- Full name: Gary Linscott
Re: Can you make this version available for more testing ?
It's easy to make an engine solve a given position more quickly. But if it reduces the ability to find the correct move in 99 other positions, it's not so great. Elo is the best tool we have for measuring that right now.pichy wrote:I was not referring to an Elo increase, it is not just about Elo If Stockfish solve this position and other similar it is an improvement by itself. Lets not forget that not all the people who buy or get an engine all that they are looking for is an Elo increase
PS: What do you do with an engine that play stronger, but can not find simple position like this one
I'm trying to find the best of both worlds, where the engine understands more motifs (in this case, revealed checks), and plays better.
-
- Posts: 3232
- Joined: Mon May 31, 2010 1:29 pm
- Full name: lucasart
Re: What engine can see the best move Bf8+!
I find that hard to believe... Can you give more details on the test you conducted ?pichy wrote: I tested a strength (Sacrifice detection = 8). Sting seems to be stronger than Stockfish.
-
- Posts: 2564
- Joined: Thu Mar 09, 2006 3:04 am
Re: What engine can see the best move Bf8+!
I did'nt conducted any tests, this is what it say on this download pagelucasart wrote:I find that hard to believe... Can you give more details on the test you conducted ?pichy wrote: I tested a strength (Sacrifice detection = 8). Sting seems to be stronger than Stockfish.
http://www.chess2u.com/t4988-sting-sf-10-is-available
-
- Posts: 2564
- Joined: Thu Mar 09, 2006 3:04 am
Re: What engine can see the best move Bf8+!
I find it hard to believe that most of the people who post here are only interested in getting the engines that that have the higher rating as if all that they want to do is to pit engine vs engine. I want more from an engine than just a high Elo, such as being able to solve simple positions that even a Human Master can solve.pichy wrote:I did'nt conducted any tests, this is what it say on this download pagelucasart wrote:I find that hard to believe... Can you give more details on the test you conducted ?pichy wrote: I tested a strength (Sacrifice detection = 8). Sting seems to be stronger than Stockfish.
http://www.chess2u.com/t4988-sting-sf-10-is-available