Hard Puzzle for July

[peter]

Do I have to look for new DTM now in the 2.g8Q- line?

Yes. We were quite not sure if White still wins here.

Ok, will really need some hardware- time for other things now first, so this time probably indeed till some later on regards

[peter]

Do I have to look for new DTM now in the 2.g8Q- line?

Yes. We were quite not sure if Black still wins here.

Don't know, why you doubt the position with White's King displaced so little only, not to be won after the much shorter way to next pawn- or capture- move in 2.g8Q- line. The longer one with 2.Kxe7 transposes anyhow.

Did just a quick Forward- Backward, maybe DTM is really a little longer then with original King- position, but because Nh1 has to be captured anyhow after Black coverted at g8 and Queens have to be exchanged, there for sure isn't any danger of coming to 50 moves rule- draw.

5K2/2p1r1P1/P1Pp4/b1pP4/1qBp2p1/3P1pPp/3N1P1B/k5BN b - - 0 1

1... Qxd2 2. g8=Q Re5 3. Qg6 Bb6 4. Kf7 Qg5 5. Qxg5 Rxg5 6. Ke7 Kb2 7. Kd8 Rg8+
8. Kd7 Kc2 9. Ke6 Kd1 10. Kf6 Ke2 11. Ke6 Kf1 12. Kf6 Kg2 13. Bb3 Kxh1 14. Ke6
Ba5 15. Kd7 Ra8 16. Bc4 Kg2 17. Bb5 Rh8 18. Bc4 Be1 19. Kxc7 Bxf2 20. Bxf2 Kxf2
21. a7 Kg2 22. Bg1 Kxg1 23. Kb7 f2 24. a8=Q Rxa8 25. Kxa8 f1=Q 26. c7 Kf2 27.
c8=Q Kxg3 28. Qe6 h2 29. Kb7 h1=Q 30. Qxd6+ Qf4 31. Qc7 Qxc7+ 32. Kxc7 Qh7+ 33.
Kc6 Kf2 34. d6 g3 35. Bd5 Qxd3 36. d7 Qa6+ 37. Kc7 g2 38. d8=Q Qa5+ 39. Kc8
Qxd8+ 40. Kxd8 g1=Q 41. Kc7 d3 42. Be4 Qg7+ 43. Kc6 Qf6+ 44. Kxc5 d2 45. Bc2
Qc3+ 46. Kd5 Qxc2 47. Ke6 d1=Q 48. Kf6 Qd6+ 49. Kg5 Qcg6+ 50. Kh4 Qdg3#

Probably with some small detours, so no entitlement for shortest DTM, but with SlowChess' fine mate search mating output can be kept rather quickly in hash backward to 15th move of a slightly different line like this:

1... Qxd2 2. g8=Q Re5 3. Qg6 Bb6 4. Kf7 Qg5 5. Qxg5 Rxg5 6. Ke7 Kb2 7. Kd8 Rg8+
8. Kd7 Kc2 9. Ke6 Kd1 10. Kd7 Ke2 11. Ke6 Kf1 12. Kd7 Kg2 13. Ke7 Kxh1 14. Bb5
Kg2 15. Kd7 :

6r1/2pK4/PbPp4/1BpP4/3p2p1/3P1pPp/5PkB/6B1 b - - 0 1

Analysis by SlowChess Blitz 2.6 sse:

15...La5 16.Lc4 Th8 17.La2 Le1 18.Kxc7 Lxf2 19.Lxf2 Kxf2 20.a7 Kg2 21.Kb7 f2 22.a8D Txa8 23.Kxa8 f1D 24.c7 Df8+ 25.Kb7 Kxh2 26.c8D Dxc8+ 27.Kxc8 Kxg3 28.Kd7 h2 29.Kxd6 h1D 30.Kc7 Dh2 31.d6 Dxa2 32.d7 Dd5 33.d8D Dxd8+ 34.Kxd8 Kf4 35.Kc7 g3 36.Kc6 Ke5 37.Kxc5 g2 38.Kc6 g1D 39.Kc5 Dc1+ 40.Kb4 Dc3+ 41.Kb5 Kd6 42.Kb6 Db4+ 43.Ka7 Kc6 44.Ka6 Db6#
-+ (-#30) Tiefe: 36/128 00:53:41 26380MN, tb=211466421

[Hobacle]

Do I have to look for new DTM now in the 2.g8Q- line?

Yes. We were quite not sure if Black still wins here.

Don't know, why you doubt the position with White's King displaced so little only, not to be won after the much shorter way to next pawn- or capture- move in 2.g8Q- line. The longer one with 2.Kxe7 transposes anyhow.

Did just a quick Forward- Backward, maybe DTM is really a little longer then with original King- position, but because Nh1 has to be captured anyhow after Black coverted at g8 and Queens have to be exchanged, there for sure isn't any danger of coming to 50 moves rule- draw.

5K2/2p1r1P1/P1Pp4/b1pP4/1qBp2p1/3P1pPp/3N1P1B/k5BN b - - 0 1

1... Qxd2 2. g8=Q Re5 3. Qg6 Bb6 4. Kf7 Qg5 5. Qxg5 Rxg5 6. Ke7 Kb2 7. Kd8 Rg8+
8. Kd7 Kc2 9. Ke6 Kd1 10. Kf6 Ke2 11. Ke6 Kf1 12. Kf6 Kg2 13. Bb3 Kxh1 14. Ke6
Ba5 15. Kd7 Ra8 16. Bc4 Kg2 17. Bb5 Rh8 18. Bc4 Be1 19. Kxc7 Bxf2 20. Bxf2 Kxf2
21. a7 Kg2 22. Bg1 Kxg1 23. Kb7 f2 24. a8=Q Rxa8 25. Kxa8 f1=Q 26. c7 Kf2 27.
c8=Q Kxg3 28. Qe6 h2 29. Kb7 h1=Q 30. Qxd6+ Qf4 31. Qc7 Qxc7+ 32. Kxc7 Qh7+ 33.
Kc6 Kf2 34. d6 g3 35. Bd5 Qxd3 36. d7 Qa6+ 37. Kc7 g2 38. d8=Q Qa5+ 39. Kc8
Qxd8+ 40. Kxd8 g1=Q 41. Kc7 d3 42. Be4 Qg7+ 43. Kc6 Qf6+ 44. Kxc5 d2 45. Bc2
Qc3+ 46. Kd5 Qxc2 47. Ke6 d1=Q 48. Kf6 Qd6+ 49. Kg5 Qcg6+ 50. Kh4 Qdg3#

Probably with some small detours, so no entitlement for shortest DTM, but with SlowChess' fine mate search mating output can be kept rather quickly in hash backward to 15th move of a slightly different line like this:

1... Qxd2 2. g8=Q Re5 3. Qg6 Bb6 4. Kf7 Qg5 5. Qxg5 Rxg5 6. Ke7 Kb2 7. Kd8 Rg8+
8. Kd7 Kc2 9. Ke6 Kd1 10. Kd7 Ke2 11. Ke6 Kf1 12. Kd7 Kg2 13. Ke7 Kxh1 14. Bb5
Kg2 15. Kd7 :

6r1/2pK4/PbPp4/1BpP4/3p2p1/3P1pPp/5PkB/6B1 b - - 0 1

Analysis by SlowChess Blitz 2.6 sse:

15...La5 16.Lc4 Th8 17.La2 Le1 18.Kxc7 Lxf2 19.Lxf2 Kxf2 20.a7 Kg2 21.Kb7 f2 22.a8D Txa8 23.Kxa8 f1D 24.c7 Df8+ 25.Kb7 Kxh2 26.c8D Dxc8+ 27.Kxc8 Kxg3 28.Kd7 h2 29.Kxd6 h1D 30.Kc7 Dh2 31.d6 Dxa2 32.d7 Dd5 33.d8D Dxd8+ 34.Kxd8 Kf4 35.Kc7 g3 36.Kc6 Ke5 37.Kxc5 g2 38.Kc6 g1D 39.Kc5 Dc1+ 40.Kb4 Dc3+ 41.Kb5 Kd6 42.Kb6 Db4+ 43.Ka7 Kc6 44.Ka6 Db6#
-+ (-#30) Tiefe: 36/128 00:53:41 26380MN, tb=211466421

OK. This is my problem. Thanks.
Because this version is much harder for engines on my computer.

[peter]

OK. This is my problem. Thanks.
Because this version is much harder for engines on my computer.

Is it, indeed?
Didn't notice much difference as for my trials with engines, yet to say I this time didn't have to let them try as much as before ponderin on their own alone, but just started with already known lines by myself. And I already knew the best engines for my special ways of trying at Forward- Backward from that very positions in question, a work of its own always and nowadays even more, to find out among the many fishes in the sea, which one's swimming fastest along the lines of interest.

As for 0.00- near eval to get them jumping I prefer Crystal here, having gone into several moves of the lines, it sooner changes to -+ here, than engines without fortress detection, more nullmove- pruning and more LMRs like SF dev.do, if 0.00 arises nearer to the ends of the lines again, it might be correct as for 50 moves- draw threatening.

Specialised mate search is fine of SF MateFinder (pity not any more under development since Joseph Ellis changed to Crystal instead), HekaStockfish by Majkel Novak, Honey by Michael Byrne and, as I found out not until lately, SlowChess. SF dev. of course is good enough and often best yet still, asl long as there isn't much Zugzwang involved and LMRs don't matter so much.

As we saw at some mating- lines too short because of ignoring 50 moves by SlowChess, the output- lines as for mate- scores are quite reliable of SlowChess' too, yet as for 50 moves better check them with e.g.Crystal yet too, SlowChess sometimes evaluates forced mate higher then along the output- lines to be found 50 moves- draw, especially without tbs used, which anyhow yet is faster for forced mate lines than togehter with 6men loaded.

Sting is a capacity of its own at all of those tricky studies, especially when it comes to Zugzwang and Lazy King eval, it's DTM- outputs aren't as quickly to be gotten now and then compared to the other specialised ones named above.

Chest and Gustav (by Olaf Jenkner) of course not to forget, but most of the really long solution- lines are not to be solved in interim time without exactly fitting parameters of their own set manually.

Good luck and have fun with your further work composing chess (choss ) problems, at least as much fun as I had with them (even if sometimes there was little bit of anger and shame in between too, not being able to solve them as quickly and easily always as I expected, but that's in such difficult ones' nature) regards

[fishpov]

Hi
Where we can get a compile version of
MateFinder 050320 ?
Thanks

[peter]

Where we can get a compile version of
MateFinder 050320 ?

Pity, the link I had it from, doesn't work anymore, this is the latest github- site I found:

https://github.com/jhellis3/Stockfish/r ... /tag/MF_10

[fishpov]

Thanks for the link but this is MateFinder 021218 64 not the lastest version I think.

[peter]

Thanks for the link but this is MateFinder 021218 64 not the lastest version I think.

For sure not, but I simply don't find a newer working version, not to mention a compile anymore. My biggest hope was Ipman, but at his site, the Dez.-18 version is the lates one too:

http://ipmanchess.yolasite.com/compiles.php

I had my latest version downloaded at I.Ivec's

http://chess.ultimaiq.net/stockfish.html

at a time, when Matefinder- compiles (togehter with the sources of course) were updated there regularly, pity I didn't keep the sources of this version, so no more chance for a BYO- compile neither.
Nowadays there's still the fine CorChess together with SF dev., Matefinder since a rather long time isn't anymore, Crystal was instead of MF first, then this one stopped at the site given too.
Sorry regards