Sting SF 16

[lech]

As I earlier wrote, I am working at a very interesting alternative in solving. it is the auto-playing.
A problem is how to solve such positions:

[d]4b3/5p2/2p4p/P1p1PN2/2pn2P1/P2k4/Pr1npKN1/B7 w - - 0 1
The Neghina's study after 8 easy moves (g2e1 d3c3 f5g3 e8d7 a5a6 d7c8 a6a7 c8b7) leads to:

[d]8/Pb3p2/2p4p/2p1P3/2pn2P1/P1k3N1/Pr1npK2/B3N3/ w - - 0 1
It is a "draw" position where Black is much stronger, but White is able to win this position thanks to a very long "lazy" king play and many zugzwang
positions. A problem is how to help to avoid it that the draw result "satisfies" the weaker side in "draw" positions. I found an interesting solution, but it needs still some time to end it.

A weak dual core (threads = 1; hash = 1024M) solves it with 30 second per move.

position fen 4b3/5p2/2p4p/P1p1PN2/2pn2P1/P2k4/Pr1npKN1/B7 w - -
go movetime 30000 play
g2e1 d3c3 f5g3 e8d7 a5a6 d7c8 a6a7 c8b7 f2g2 b7a8 g2h3 a8b7 g4g5 h6g5 e5e6 f7e6 h3g4 b7a8 g4h5 a8b7 h5g6 b7a8 g6f7 a8b7 f7e8 b7a8 e8d7 a8b7 d7d8 b7a8 d8e8 g5g4 e8d7 a8b7 d7d8 b7a8 d8c8 e6e5 c8d7 a8b7 d7e8 b7a8 e8f7 a8b7 f7f6 b7a8 f6g6 a8b7 g6h5 b7a8 h5g4 a8b7 g4h5 b7a8 h5g6 a8b7 g6f7 b7a8 f7e8 a8b7 e8d7 b7a8 d7c8 e5e4 c8d7 a8b7 d7e8 b7a8 e8f7 a8b7 f7g6 b7a8 g6f6 a8b7 f6f7 b7a8 f7e8 a8b7 e8d7 b7a8 d7c8 e4e3 c8d7 a8b7 d7e8 b7a8 e8f7 a8b7 f7g6 b7a8 g6f6 a8b7 f6f7 b7a8 f7e8 a8b7 e8d7 b7a8 d7c8 a8b7 c8b8 b7a6 a1b2 stop
go movetime 3000 play
c3b2 a7a8q a6b5 a3a4 d2f1 g3e2 d4e2 a4b5 c6b5 a8g2 f1g3 g2f3 c4c3 f3e3 c5c4 e3g5 b5b4 g5b5 b4b3 b5c4 b3a2 e1d3 b2c2 d3b4 c2c1 c4a2 c1d1 a2d5 d1e1 b4c2 e1f2 d5f7 f2g1 c2e3 e2d4 f7f4 d4f5 e3f5 g3f5 f4f5 g1g2 b8c7 g2g3 c7d6 c3c2 f5c2 g3f3 d6e5 f3e3 c2d1 e3f2 e5f4 f2g2 d1c2 g2f1 f4f3 f1g1 c2g2 Black is mated

[lech]

A next example proving that auto-playing can be a wonderful idea, not only to display full solutions (based on chess rules: draw or mate), but to solve
some very difficult positions too.

Popov Georgy & Hanyan Aleksei (part of study - full with black pawn h5)
[d]qn6/b1Kp3p/p1pB3p/p7/P1NkP3/2pP4/2B5/8 w - - 0 1

position fen qn6/b1Kp3p/p1pB3p/p7/P1NkP3/2pP4/2B5/8 w - - 0 1
go movetime 10000 play
c7c8 h6h5 d6f4 d4c5 f4e3 c5b4 e3c1 b4c5 c1a3 c5d4 a3d6 a7b6 c4b6 a8a7 b6c4 a7a8 d6f4 d4c5 f4e3 c5b4 e3c1 b4c5 c1a3 c5d4 a3d6 h5h4 d6f4 d4c5 f4e3 c5b4 e3c1 b4c5 c1a3 c5d4 a3d6 h4h3 d6f4 d4c5 f4e3 c5b4 e3c1 b4c5 c1a3 c5d4 a3d6 h7h5 d6f4 d4c5 f4e3 c5b4 e3c1 b4c5 c1a3 c5d4 a3d6 h5h4 d6f4 d4c5 f4e3 c5b4 e3c1 b4c5 c1a3 c5d4 a3d6 h3h2 d6h2 d4c5 h2d6 c5d4 d6f4 d4c5 f4e3 c5b4 e3c1 b4c5 c1a3 c5d4 a3d6 h4h3 d6f4 d4c5 f4e3 c5b4 e3c1 b4c5 c1a3 c5d4 a3d6 a8a7 d6h2 d4c5 h2g1 c5b4 g1a7 d7d5 a7b8 d5c4 d3c4 b4a3 b8f4 a3b2 c2d3 c3c2 d3c2 b2c2 e4e5 c2b3 e5e6 b3c4 e6e7 c4d3 e7e8q d3c3 e8e3 c3c2 f4e5 c6c5 e3c3 c2d1 c3d3 d1c1 e5c3 h3h2 d3d2 c1b1 d2b2 Black is mated

[Scally]

Hi,

Compiling the new code on a Raspberry Pi produces this error on trying to start the engine:
Failed to allocate 1024 MB for transposition table.

I changed the 1024 to 32 in ucioption.cpp as per Sting SF15 and all works well.

Thanks,

Al.

[Vinvin]

Hi !
This position should be evaluated as a draw (because white king goes to hide on h1 and g1).
[d]8/8/8/3k2p1/4N3/2b1P1pP/4K1P1/8 b - - 0 1

[lech]

Hi !
This position should be evaluated as a draw (because white king goes to hide on h1 and g1).
[d]8/8/8/3k2p1/4N3/2b1P1pP/4K1P1/8 b - - 0 1

Indeed Sting can do it by auto-playing (extended searching).
position fen 8/8/8/3k2p1/4N3/2b1P1pP/4K1P1/8 b - -
go movetime 1000 play
d5e4 e2f1 e4e3 f1g1 e3e2 g1h1 c3d4 h3h4 g5g4 h4h5 d4e3 h5h6 e2d1 h6h7 e3d4 h7h8n d4h8 h1g1 h8b2 g1f1 d1d2 f1g1 d2e2 g1h1 b2c1 h1g1 c1d2 g1h1 d2c3 h1g1 c3e1 g1h1 e1b4 h1g1 b4a3 g1h1 a3b2 h1g1 b2d4 g1h1 e2e3 h1g1 d4f6 g1f1 f6c3 f1g1 c3e5 g1f1 e5b8 f1g1 e3d2 g1f1 b8c7 f1g1 d2e2 g1h1 c7f4 h1g1 e2d2 g1f1 f4d6 f1g1 d6e5 g1f1 e5h8 f1g1 d2e2 g1h1 h8a1 h1g1 a1e5 g1h1 e5d6 h1g1 d6f4 g1h1 e2d3 h1g1 f4e5 g1f1 d3d2 f1g1 e5c3 g1f1 c3b4 f1g1 b4d6 g1h1 d6e7 h1g1 e7c5 g1f1 c5a3 f1g1 d2d1 g1h1 a3f8 h1g1 f8b4 g1f1 d1c2 f1g1 b4a3 g1h1 a3e7 h1g1 e7d6 g1h1 d6e5 h1g1 e5a1 g1h1 a1b2 h1g1 b2e5 g1f1 e5d4 draw

[Vinvin]

Hi !
Here's another one where Sting-SF cannot see the draw :
[D]4K1bn/5p2/5PpB/6P1/3k4/8/8/7q w - - 0 4

Variation from viewtopic.php?p=666297#p666297

[lech]

Hi !
Here's another one where Sting-SF cannot see the draw :
[D]4K1bn/5p2/5PpB/6P1/3k4/8/8/7q w - - 0 4

Variation from viewtopic.php?p=666297#p666297

To understand what happens here, there is needed to know and understand chess, chess rules and alpha-beta search.

Try to get it and you will see that Sting is ok.

[lech]

Hi !
Here's another one where Sting-SF cannot see the draw :
[D]4K1bn/5p2/5PpB/6P1/3k4/8/8/7q w - - 0 4

Variation from viewtopic.php?p=666297#p666297

To understand what happens here, there is needed to know and understand chess, chess rules and alpha-beta search.

Try to get it and you will see that Sting is ok.

When you stop searching you know that a position can be equal (missing gain) or not, ONLY.
Then, if you want (need), to get the full solution and result (mate or draw), start auto-playing.

[Spliffjiffer]

hello Marek, i tried your "autoplay-mode" myself and it works excellently, very intereting to juggle with it !
one question: is it possible for me to use my complete 4 threads that i have to let Sting autoplay positions... if so, how ?
ah...beside this: is it possible to abbort the autoplay-process by using a certain command maybe ?

[lech]

hello Marek, i tried your "autoplay-mode" myself and it works excellently, very intereting to juggle with it !
one question: is it possible for me to use my complete 4 threads that i have to let Sting autoplay positions... if so, how ?
ah...beside this: is it possible to abbort the autoplay-process by using a certain command maybe ?

to use more threads (4):
setoption name threads value 4
to stop solution:
click button ctrl (don't hold) and type stop