Anti-Engine Puzzle

[peter]

Good job of Sting's without doubt too.
Mating score isn't really necessary, if engine obviously "sees" it should work within 50 moves, otherwise 0.00- near eval would arise, and of course the line should show progress, which Sting one's clearly does too. Probably Sting would make the mate <50 in play out too, (question remains, if without black blunders) didn't try it neither, and still, if it's to be seen as a study, the 50 moves don't count at all anyhow.

Yet I think it's the engines' problem as for eval still mainly, "thinking" 50 moves draw to arise before mate.

That is nonsense. If an engine sees +3, it just means it is a minor ahead, but the position is still to complex to recognize it as a draw (which a minor ahead typically is). So it is probably not KBK. There is zero evidence that it could force a checkmate from that position.

What ist nonsense?
That Chest gives a #46 line?

http://talkchess.com/forum3/viewtopic.p ... 60#p886360

Reproduced here:

http://talkchess.com/forum3/viewtopic.p ... 91#p886391

And that you can keep such lines in hash (if it's big enough) as DTM<50 with SF at Backward in output too? Already with only 4G hash SF 11 made it back to 8th move with a 150+ eval here:

http://talkchess.com/forum3/viewtopic.p ... 79#p886379

And you saw the illegal position here, did you?

http://talkchess.com/forum3/viewtopic.p ... 02#p886402

And Vincent's Sting had not only +3, it was +110 here, did you see?

http://talkchess.com/forum3/viewtopic.p ... 03#p886403

And what's nonsense in my supposing, that the wrong 0.00- near eval can come from drawing lines within search- horizon without mate "in sight" only?
If it wasn't 50 moves rule, where do you think, 0.00 should come from with this unbalanced material otherwise?

[Paloma]

BTW,
it's a Study by Seoane,J 1922 La Strategie#0915

[Dann Corbit]

BTW,
it's a Study by Seoane,J 1922 La Strategie#0915

Could you be so kind as to post the substance of the study here?
I did not find it at YACPDB or PDB server.

[Michel]

It seems this can be easily solved by constructing a TB. Only about 10 million positions to consider (assuming that after the white king moves to f1 the configuration of 7 pieces in the lower right corner is never touched).

[hgm]

Indeed, that is what I did. Just remove those occupied squares from the board. Not entirely correct, because it ignores that e1 and e2 are not accessible by the black King. But that is a disadvantage for white, and if it doesn't spoil the win, than this apparently did not matter. Another extra handicap for white is that the EGT geenrator considered one of the Knights royal, and thus would not dare to step it next to the black King, even when protected.

But because the end-game is only half won with these handicaps, it would still be interesting to see how it would fare without those.

[Paloma]

BTW,
it's a Study by Seoane,J 1922 La Strategie#0915

Could you be so kind as to post the substance of the study here?
I did not find it at YACPDB or PDB server.

[pgn][Event "La Strategie#0915"]
[Site "?"]
[Date "1922.??.??"]
[Round "?"]
[White "Seoane=J"]
[Black "(+0345.03e2c6)"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/8/2k5/3N4/8/4N1p1/4Kprp/B5bn w - - 0 1"]
[PlyCount "91"]
[EventDate "1922.??.??"]

{La Strategie/1.} 1. Kf1 Kd6 (1... Kc5 2. Be5 Kb5 3. Nc3+ Kc5 4. Bg7 Kb4 5. Bd4
Ka5 6. Nc4+ Ka6 7. Nd6 Ka5 8. Bc5 Ka6 9. Nd5 Ka5 10. Nb4 Ka4 11. Nc6 Kb3 12.
Bb4 Kc2 13. Ne5 Kb3 14. Nd3 Kc2 15. Nc5) 2. Bd4 Kc6 3. Nc3 Kd6 4. Nc4+ Ke6 5.
Be5 Kd7 6. Na5 Ke6 7. Bb8 Kd7 8. Nd5 Ke6 9. Nb6 Kf5 10. Nac4 Ke4 11. Be5 Kf5
12. Bc3 Ke6 13. Bd4 Ke7 14. Be5 Ke6 15. Bc7 Kf5 16. Nd2 Ke6 17. Ne4 Kf5 18. Nc5
Kg4 19. Nc4 Kf3 20. Be5 Kg4 21. Nd6 Kf3 22. Bd4 Kg4 23. Ne6 Kh5 24. Bf6 Kg6 25.
Be7 Kh5 26. Nf4+ Kg4 27. Ng6 Kh5 28. Ne5 Kh6 29. Nf5+ Kh5 30. Ng7+ Kh6 31. Bf6
Kh7 32. Nf5 Kg8 33. Bg7 Kh7 34. Bh6 Kg8 35. Ne7+ Kh8 36. Ng4 Kh7 37. Bf8 Kh8
38. Ne3 Kh7 39. N3f5 Kh8 40. Bg7+ Kh7 41. Bh6 Kh8 42. Bf8 Kh7 43. Ne3 Kh8 44.
Ng4 (44. N3d5 {<or>}) 44... Kh7 45. Nf6+ Kh8 46. Ng6# 1-0

[/pgn]

[Dann Corbit]

Thanks you, Paloma

[Pio]

Indeed, that is what I did. Just remove those occupied squares from the board. Not entirely correct, because it ignores that e1 and e2 are not accessible by the black King. But that is a disadvantage for white, and if it doesn't spoil the win, than this apparently did not matter. Another extra handicap for white is that the EGT geenrator considered one of the Knights royal, and thus would not dare to step it next to the black King, even when protected.

But because the end-game is only half won with these handicaps, it would still be interesting to see how it would fare without those.

Isn’t the easiest way to change the PST for king and put it as 2000
For white king on e2

[hgm]

There is no PST in an EGT generator. And no white King either, the way I used it.

[Pio]

There is no PST in an EGT generator. And no white King either, the way I used it.

I am sorry. What I meant was that changing the PST for the white king should help solving the puzzle but it won’t fix it. Not generating moves for white king on e2 (and removing the moves from and to where the black pieces are (except the black king) should make any fast and bug free engine solve this problem in a couple of seconds since there are approximately 10 million positions.