Cooked composition?

[MOBMAT]

Vladislav Tarasiuk & A. Khandurin
3rd Prize Phenix 2019
[d]2R5/B7/6K1/8/4k3/5r2/7p/7N b - - 1 4
I found the above position here:
http://www.arves.org/arves/index.php/nl ... henix-2019

I threw every engine I could at it and not one of them could solve the problem as published, which is supposed to be a mate in 11.

The knight is sacrificed twice on h1 first for a winning Rook ending and then for a surprising Zugzwang. The underpromotion of a black knight after ...Kg4 on the very same thematic square h1 adds a touch of humour to the main theme. 1. Nf5 [1. Ne6? Ke4 2. Bf2 gf2 3. Rc1 Kf3 4. Kf6 h2! 5. Kf5 f1Q 6. Rf1 Kg2=] 1... h2 2. Ng3! [2. Bf2? h1Q 3. Bg3 Ke4=] 2... Rf3 3. Nh1! [3. Bb8? Ke3! 4. Ke6 Rg3 5. Rh8 Rg8! 6. Rh2 Ke4!=] 3... Ke4 [3... Rf1 4. Rc4! Rh1 5. Bb8 Ke3 6. Bh2! Rh2 7. e5+»] 4. Kg6 Ra3! [4... Rf1 5. Ng3 Kf3 6. Nf1! h1Q 7. Rf8 Kg2 /Rg4 8. Ne3 Kg3 9. Bb8 Kh3 /Kh4 10. Rh8#] 5. Bb8 Ra1 6. Ng3! [6. Nf2? Kf3 7. Rf8 Kg2=] 6... Kf3 7. Rc3 Kg2 [7... Kg4!? 8. Ne4 h1N! (8... h1Q? 9. Nf2 Kh4 10. Bg3#) 9. Nf6 Kh4 10. Rc2 Ra3 11. Rd2 Rb3 12. Bd6 Re3 13. Rg2 Rf3 14. Ne4 Rb3 (or 14... Rf1 15. Rd2 Kg4 16. Ng5+») 15. Kf5 Rf3 16. Bf4] 8. Rc2 Kh3 9. Nh1!! [9. Rc3 Kg2 10. Rc2 Kh3 11. Nh1 would only be a waste of time] [Try 9. Ne4? Ra6! (9... h1Q? 10. Ng5 Kg4 11. Rc4 Qe4 12. Re4#) 10. Kh5 Ra5! (10... h1Q?? 11. Ng5#) 11. Ng5 Rg5 12. Kg5 h1Q=] 9... Rh1 10. Kh5! zugzwang 10... Rg1 11. Rh2#

So, thinking there was something wrong, i did retro analysis on it and wow, the composers must not have used an engine to validate the problem since I found more than one mistake. This is surprising since it is a problem from last year and if memory serves, there were strong engines back then. LOL

Comments are based on the scores that SF DEV 20071122 returns:
4. ... Rf1 is better.
5. ... Ra6+ delays a bit
7. ... Kg4 might be the same outcome
9. ... Rg1+ is the best move, not Rxh1 (mate in 2). Isn't the point of problems/puzzles to find the BEST responses, not the worse?

So, the author's got 3rd place, but it appears neither they nor the judges verified it.
I could not get any engine to return a mate score from the starting position (though some engines report a winning Syzygy score), can someone with more threads and memory give it a shot?

[Paloma]

This is sthe study from the beginning:
[pgn]
[Event "3rd Prize Phenix 2019"]
[Site ""]
[Date ""]
[Round ""]
[White "Vladislav Tarasiuk & A. Khandurin"]
[Black ""]
[Result "1-0"]
[Annotator "Y. Afek"]
[SetUp "1"]
[FEN "2R5/B4K2/8/8/3NPk2/6pp/5r2/8 w - - 0 1"]
[PlyCount "21"]

{The knight is sacrificed twice on h1first for a winning Rook ending and then for a surprising Zugzwang. The underpromotion of a black knight after ...Kg4 on the very same thematic square h1 adds a touch of humour to the main theme.} 1. Nf5 (1. Ne6+ $2 Kxe4+ 2. Bxf2 gxf2 3. Rc1 Kf3 4. Kf6 h2 $1 5. Kf5 f1=Q 6. Rxf1+ Kg2 $11) 1... h2 2. Nxg3 $1 (2. Bxf2 $2 h1=Q 3. Bxg3+ Kxe4 $11) 2... Rf3 3. Nh1 $1 (3. Bb8+ $2 Ke3+ $1 4. Ke6 Rxg3 5. Rh8 Rg8 $1 6. Rxh2 Kxe4 $1 $11) 3... Kxe4+ (3... Rf1 4. Rc4 $1 Rxh1 5. Bb8+ Ke3 6. Bxh2 $1 Rxh2 7. e5 $18) 4. Kg6 Ra3 $1 (4... Rf1 5. Ng3+ Kf3 6. Nxf1 $1 h1=Q 7. Rf8+ Kg2 {/Rg4} 8. Ne3+ Kg3 9. Bb8+ Kh3 {/Kh4} 10. Rh8#) 5. Bb8 Ra1 6. Ng3+ $1 (6. Nf2+ $2 Kf3 7. Rf8+ Kg2 $11) 6... Kf3 7. Rc3+ Kg2 (7... Kg4 $5 8. Ne4 h1=N $1 (8... h1=Q $2 9. Nf2+ Kh4 10. Bg3#) 9. Nf6+ Kh4 10. Rc2 Ra3 11. Rd2 Rb3 12. Bd6 Re3 13. Rg2 Rf3 14. Ne4 Rb3 ({or} 14... Rf1 15. Rd2 Kg4 16. Ng5 $18) 15. Kf5 Rf3+ 16. Bf4) 8. Rc2+ Kh3 9. Nh1 $3 (9. Rc3 Kg2 10. Rc2+ Kh3 11. Nh1 {would only be a waste of time}) ({ Try} 9. Ne4 $2 Ra6+ $1 (9... h1=Q $2 10. Ng5+ Kg4 11. Rc4+ Qe4+ 12. Rxe4#) 10. Kh5 Ra5+ $1 (10... h1=Q $4 11. Ng5#) 11. Ng5+ Rxg5+ 12. Kxg5 h1=Q $11) 9... Rxh1 10. Kh5 $1 {zugzwang} Rg1 11. Rxh2# 1-0
[/pgn]

But it's never a mate in 11

The Pos. you posted after white's fourth move lead to a mate in 34!
Best move for black 4...Rf1. Then 5.Ng3+ is mate in 34
After 4...Ra3 is 5.Bb8 mate in 33
All with 7men Lomonosov Tb

[Event "?"]
[Site "?"]
[Date "?"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "1-0"]
[SetUp "1"]
[FEN "2R5/B7/6K1/8/4k3/5r2/7p/7N b - - 0 1"]
[Move_number "0"]
[Is_demo_mode "false"]

1... Rf1 2. Ng3+ Kd5 3. Rh8 Rg1 4. Bb8 h1=Q 5. Rxh1 Rxh1 6. Nxh1 Kc5 7. Ng3 Kb5 8. Kf6 Ka5 9. Nf5 Kb5 10. Nd6+ Kb6 11. Ke6 Kc6 12. Ba7 Kc7 13. Kd5 Kd7 14. Nb5 Ke7 15. Bd4 Kd7 16. Be5 Ke7 17. Nd6 Kd8 18. Nf5 Kd7 19. Bf4 Kc8 20. Kc6 Kd8 21. Bg3 Kc8 22. Ne7+ Kd8 23. Kd6 Ke8 24. Ke6 Kd8 25. Bh2 Ke8 26. Bc7 Kf8 27. Nf5 Ke8 28. Ng7+ Kf8 29. Kf6 Kg8 30. Ne6 Kh8 31. Bf4 Kg8 32. Kg6 Kh8 33. Kf7 Kh7 34. Nf8+ Kh8 35. Be5# 1-0

[Ajedrecista]

Hello:

With the given solution, once the board reaches seven pieces (after 1. Nf5 h2 2. Nxg3 Rf3 3. Nh1 Kxe4), there is a DTZ (distance to zero) in 23 according to Syzygy online EGTB (KRBNvKRP):

https://syzygy-tables.info/?fen=2R5%2FB ... +w+-+-+0+4

[d]2R5/B4K2/8/8/4k3/5r2/7p/7N w - - 0 4

The only winning move is 4. Kg6, that is, the position of the diagramme of the original post.

Mainline from Syzygy EGTB after 3. ... Kxe4.

Analysis at Lichess.org

The other post confirms the issue with Lomonosov EGTB and DTM metric (distance to mate).

Though the composition starts with 9 pieces OTB, not 7.

Regards from Spain.

Ajedrecista.

[Alayan]

This is a tablebase position...

Black gets mated in 11 if he goes for the longest possible DTZ to avoid 6-men TB, with Ra3, as the capture resetting DTZ in the longest line is also checkmate. But that's after 4. Kg6 from the composition... By allowing an earlier simplification to 6-men TB, black can delay the checkmate significantly longer. The alternative to 7. ... Kg2 (that shortens DTZ), 7 ... Kg4 is marked as !? though the provided line goes to move 16 without checkmate because of course there is none. Rxh1?? in the composition main line is a gross blunder, and any decent engine can trivially prove it to be so.

What's the point of a composition with such big mistakes ?

[Deberger]

This is a very fine composition.

It's "White to play and win" not "Mate in 11."

The annotated exposition has "errors" such as 9...Rxh1, because variations which allow White to play Rxh2 are indeed trivial.

[MOBMAT]

Sorry for the messed up FEN

[d]2R5/B4K2/8/8/3NPk2/6pp/5r2/8 w - - 0 1

So is it M38 from the beginning?