Shocking misevaluation by SF

[Spliffjiffer]

for example i struggle mostly with this line atm:

1. Kg1 Rc2 2. Ra6 Ke4 3. Ra4+ Rc4 4. Ra8
[d]R7/8/6p1/4n2p/2r1k2P/6P1/5P2/6K1 b - - 7 4

[Eelco de Groot]

...2 clear lines that should help you (and SF also) to see that its winning

[d]8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - - 0 1

[pgn]
[Event "?"]
[Site "?"]
[Date "2017.08.08"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - -"]

1. Kg1 Rc2 2. Kf1 Ng4 3. Rf8 Ke6 4. Kg1 Ne5 5. Re8+ Kf6
6. Ra8 Nd3 7. Rd8 Rd2 8. Re8 Kf5 9. Rf8+ Ke6 10. Rf3
(10. Kf1 Ra2 11. Rf3 Ra1+ 12. Ke2 Ne5 13. Rf4 Ke7 14. Re4
Kf6 15. Re3 Kf5) 10... Rd1+ 11. Kh2 Ne5 12. Re3 Rd2 13. Kg1
Kd5 14. Ra3 Nd3 15. Ra8 Ke6 16. Re8+ Kf6 17. Rf8+ Kg7
18. Rf3 Rd1+ 19. Kh2 Ne5 20. Ra3 Rd2 21. Kg2 Nd3 22. Ra7+
Kf6 23. Ra6+ Kf5 24. Ra5+ Ke4 25. Ra4+ Kd5 26. Ra5+ Kd4
27. Ra4+ Kc3 28. Ra3+ Kc4 $17 *
[/pgn]

Unfortunately at least my version of Stockfish is so bad at this that it loses the solution when going upwards again, Especially if you leave it too long in a position with White to move. Part of this is Lazy SMP that will pick the highest scoring line. If this line is bad and deep at the same time it will start overwriting everything with new scores and can erase your whole solution. At least that is what I found. Up to here from the above pgn everything is still okay and it sees that Black is winning:

[D]5R2/6k1/6p1/7p/7P/3n2P1/3r1P2/6K1 w - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

28 1:03 -5.06 18.Rf3 Rd1+ 19.Kh2 Ne5 20.Ra3 Ng4+
21.Kg2 Rd2 22.Rf3 Ne5 23.Ra3 Nd3
24.Ra7+ Kf6 25.Ra6+ Kf5 26.Ra5+ Ke4
27.Ra4+ Kd5 28.Ra5+ Kd4 29.Ra4+ Kc3
30.g4 Rxf2+ 31.Kg3 (993.313.152) 15669

28 1:03 -7.25 18.Rc8 Nxf2 19.Kg2 Ng4+ 20.Kf3 Rd3+
21.Kf4 Rd4+ 22.Kf3 Rb4 23.Re8 Nh2+
24.Ke2 Rb2+ 25.Kd3 Rb3+ 26.Ke4 Rxg3
27.Rc8 Ng4 28.Rc7+ Kh6 29.Kf4 Rd3
30.Rc8 Rb3 31.Ke4 (993.313.152) 15669

27 1:03 -7.19 18.Ra8 Nxf2 19.Kf1 Rb2 20.Ra6 Kh6
21.Ra8 Ng4 22.Ra3 Rb7 23.Kg1 Re7
24.Ra8 Re3 25.Kh1 Rf3 26.Kg2 Rb3
27.Rc8 Rb2+ 28.Kh1 Rh2+ 29.Kg1 Rd2
30.Rh8+ Kg7 31.Rb8 (993.313.152) 15669

(This is backtracked from the positon after 28... Kc4)

This is still good

5R2/8/5kp1/7p/7P/3n2P1/3r1P2/6K1 b - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

34 8:44 -5.12 17...Kg7 18.Rf3 Rd1+ 19.Kh2 Ne5
20.Ra3 Ng4+ 21.Kg2 Rd2 22.Rf3 Ne5
23.Ra3 Nd3 24.Ra7+ Kf6 25.Ra6+ Kf5
26.Ra5+ Ke4 27.Ra4+ Kd5 28.Ra5+ Kd4
29.Ra4+ Kc3 30.g4 Rxf2+ (8.726.257.294) 16630

34 8:44 -2.24 17...Ke7 18.Rf3 Rd1+ 19.Kh2 Ke6
20.Re3+ Kd5 21.Rf3 Ne5 22.Rf8 Rb1
23.Rd8+ Ke6 24.Re8+ Kf6 25.Ra8 Ng4+
26.Kg2 Rb2 27.Kg1 Kf5 28.Rf8+ Ke6
29.Kg2 Kd5 30.Rf4 Rd2 (8.726.257.294) 16630

34 8:44 -2.24 17...Ke6 18.Kf1 Ra2 19.Kg1 Ke7
20.Rf3 Rd2 21.Kf1 Ne5 22.Rf4 Ke6
23.Rf8 Ra2 24.Re8+ Kf6 25.Kg1 Ng4
26.Rf8+ Kg7 27.Rf3 Rb2 28.Kg2 Rd2
29.Kg1 Ne5 30.Re3 Kf6 (8.726.257.294) 16630

One half move up

[D]4R3/8/5kp1/7p/7P/3n2P1/3r1P2/6K1 w - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

37 259:52 -4.93 17.Ra8 Ke7 18.Ra6 Nxf2 19.Rxg6 Ng4
20.Rc6 Nf6 21.Kf1 Rb2 22.Rc5 Ke6
23.Rc3 Nd5 24.Rc8 Ke5 25.Re8+ Kf5
26.Kg1 Rc2 27.Kf1 Ra2 28.Re1 Nf6
29.Re7 Rc2 30.Kg1 (259.309.940.484) 16630

37 259:52 -5.12 17.Rf8+ Kg7 18.Rf3 Rd1+ 19.Kh2 Ne5
20.Ra3 Rd2 21.Kg2 Nd3 22.Ra7+ Kf6
23.Ra6+ Kf5 24.Ra5+ Ke4 25.Ra4+ Kd5
26.Ra5+ Kd4 27.Ra4+ Kc3 28.g4 Rxf2+
29.Kg3 Rf1 30.gxh5 (259.309.940.484) 16630

37 259:52 -6.87 17.f3 Ne5 18.Kf1 Ra2 19.Rc8 Ra3
20.Kg2 Kg7 21.g4 hxg4 22.fxg4 Nxg4
23.Rc6 Rb3 24.Rc8 Ne3+ 25.Kf3 Nf5+
26.Kf2 Rb2+ 27.Kf3 Nxh4+ 28.Ke3 Nf5+
29.Kf3 Rb4 30.Rc7+ (259.309.940.484) 16630

But I waited too long and now it loses it one ply deeper in the same position...

4R3/8/5kp1/7p/7P/3n2P1/3r1P2/6K1 w - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

38 386:20 -2.26 {wrong} 17.Rf8+ Ke7 {wrong} 18.Rf3 Rd1+ 19.Kh2 Ke6
20.Re3+ Kd5 21.Rf3 Rd2 22.Kg1 Ke6
23.Rf8 Ke5 24.Rf3 Nc5 25.Ra3 Ne4
26.Re3 Kd4 27.Rf3 Ra2 28.Rf8 Ke5
29.Kg2 Nf6 30.Kg1 (387.226.768.361) 16704

38 386:20 -6.29 17.Ra8 Kf7 18.Ra7+ Ke6 19.Ra6+ Kf5
20.f3 Ne1 21.Ra5+ Kf6 22.Ra6+ Kg7
23.Ra7+ Kh6 24.f4 Nf3+ 25.Kf1 Rb2
26.f5 Nh2+ 27.Ke1 gxf5 28.Ra5 Kg6
29.Ra6+ Kg7 30.Ra7+ (387.226.768.361) 16704

37 386:20 -6.87 17.f3 Ne5 18.Kf1 Ra2 19.Rc8 Ra3
20.Kg2 Kg7 21.g4 hxg4 22.fxg4 Nxg4
23.Rc6 Rb3 24.Rc8 Ne3+ 25.Kf3 Nf5+
26.Kf2 Rb2+ 27.Kf3 Nxh4+ 28.Ke3 Nf5+
29.Kf3 Rb4 30.Rc7+ (387.226.768.361) 16704

I could still repair that going one ply up to a Black position, but a little bit further up it lost it again.


[D]R7/8/6p1/3k3p/7P/3n2P1/3r1P2/6K1 b - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

40 18:12 -2.19 {wrong} 15...Ke6 16.Rf8 Rc2 17.Rf3 Rc1+
18.Kh2 Rd1 19.Re3+ Kd5 20.Rf3 Rd2
21.Kg1 Ne5 22.Rf8 Ke6 23.Re8+ Kf6
24.Kf1 Ra2 25.Kg1 Rc2 26.Ra8 Ng4
27.Rb8 Rd2 28.Re8 Kf7 (17.261.854.047) 15793

40 18:12 -2.19 {possibly wrong but not the main line} 15...Ke4 16.Rf8 Ra2 17.Rf6 Ne5
18.Rf4+ Kd5 19.Kg2 Re2 20.Rf8 Ke6
21.Rb8 Kd6 22.Re8 Rb2 23.Ra8 Ng4
24.Rf8 Ke7 25.Rf3 Rd2 26.Kg1 Ke6
27.Rf8 Ne5 28.Rb8 Kd5 (17.261.854.047) 15793

40 18:12 -2.19 {possibly wrong but not the main line} 15...Ke5 16.Re8+ Kd6 17.Rg8 Ne5
18.Re8 Ra2 19.Kg2 Ng4 20.Rf8 Ke6
21.Rf3 Ne5 22.Rb3 Nc4 23.Rf3 Nd6
24.Re3+ Kf5 25.Rf3+ Ke5 26.Rf4 Rc2
27.Rf3 Ke6 28.Re3+ Kf6 (17.261.854.047) 15793

or:

[pgn]
[Event "?"]
[Site "?"]
[Date "2017.08.08"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - -"]

1. Kg1 Rc2 2. Rb6 Ke4 3. Rb4+ Rc4 4. Rb6 Ng4 5. Rb3 Rc1+
6. Kg2 Rc2 7. Rf3 Ke5 8. Rf8 Ke6 9. Rf3 Rd2 10. Kg1 Ne5
11. Rf4 Nd3 12. Rf3 Rd1+ 13. Kh2 Ne5 14. Re3 Rd2 15. Kg2
(15. Kg1 Kd5) 15... Kf5 $17 *
[/pgn]

[Eelco de Groot]

I could still repair that going one ply up to a Black position, but a little bit further up it lost it again.


[D]R7/8/6p1/3k3p/7P/3n2P1/3r1P2/6K1 b - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

40 18:12 -2.19 {wrong} 15...Ke6 16.Rf8 Rc2 17.Rf3 Rc1+
18.Kh2 Rd1 19.Re3+ Kd5 20.Rf3 Rd2
21.Kg1 Ne5 22.Rf8 Ke6 23.Re8+ Kf6
24.Kf1 Ra2 25.Kg1 Rc2 26.Ra8 Ng4
27.Rb8 Rd2 28.Re8 Kf7 (17.261.854.047) 15793

40 18:12 -2.19 {possibly wrong but not the main line} 15...Ke4 16.Rf8 Ra2 17.Rf6 Ne5
18.Rf4+ Kd5 19.Kg2 Re2 20.Rf8 Ke6
21.Rb8 Kd6 22.Re8 Rb2 23.Ra8 Ng4
24.Rf8 Ke7 25.Rf3 Rd2 26.Kg1 Ke6
27.Rf8 Ne5 28.Rb8 Kd5 (17.261.854.047) 15793

40 18:12 -2.19 {possibly wrong but not the main line} 15...Ke5 16.Re8+ Kd6 17.Rg8 Ne5
18.Re8 Ra2 19.Kg2 Ng4 20.Rf8 Ke6
21.Rf3 Ne5 22.Rb3 Nc4 23.Rf3 Nd6
24.Re3+ Kf5 25.Rf3+ Ke5 26.Rf4 Rc2
27.Rf3 Ke6 28.Re3+ Kf6 (17.261.854.047) 15793

This is a Black position and Ke6 was Failing High again, but it took almost 700 minutes to resolve the FH (or FL for White if you will) at depth 41. This is the downside of not returning hash hits in PV nodes, or, more generally, not having a learning engine. My ideal would be to permanently store all PV nodes like in Jeremy Bernstein's learning Stockfish but hopefully in shorter files. Just for cases like this as long as the evaluation still fails. But his version has all these external library functions that I don't think are really necessary and for a beginner are hard to fathom what they do.
Old Glaurung also had a very limited learn file (well more like a not permanent cache I believe, forgot all the details of it) but Tord removed the code because users did not understand it...

41 707:26 -4.57 15...Ke6 16.Re8+ Kf6 17.Rf8+ Kg7 18.Rf3 Rd1+ 19.Kh2 Ne5 20.Ra3 Rd2 21.Kg1 Nd3 22.Ra7+ Kg8 23.Ra6 Nxf2 24.Rxg6+ Kh7 25.Rf6 Ng4 26.Rb6 Kg7 27.Rb5 Nf6 28.Re5 Kg6 (671.521.967.070) 15820
40 707:26 -2.19 15...Ke4 16.Rf8 Ra2 17.Rf6 Ne5 18.Rf4+ Kd5 19.Kg2 Re2 20.Rf8 Ke6 21.Rb8 Kd6 22.Re8 Rb2 23.Ra8 Ng4 24.Rf8 Ke7 25.Rf3 Rd2 26.Kg1 Ke6 27.Rf8 Ne5 28.Rb8 Kd5 (671.521.967.070) 15820
40 707:26 -2.19 15...Ke5 16.Re8+ Kd6 17.Rg8 Ne5 18.Re8 Ra2 19.Kg2 Ng4 20.Rf8 Ke6 21.Rf3 Ne5 22.Rb3 Nc4 23.Rf3 Nd6 24.Re3+ Kf5 25.Rf3+ Ke5 26.Rf4 Rc2 27.Rf3 Ke6 28.Re3+ Kf6 (671.521.967.070) 15820
-------------------------------------------------------------------------------------


[D]R7/8/6p1/3k3p/7P/3n2P1/3r1P2/6K1 b - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

43 900:47 -4.57 15...Ke6 16.Re8+ Kd7 17.Rg8 Nxf2
18.Rg7+ Ke8 19.Rxg6 Ng4 20.Rg7 Nf6
21.Ra7 Rd7 22.Ra4 Kf7 23.Kg2 Ke6
24.Kf3 Kf5 25.Rf4+ Ke5 26.Ra4 Rd5
27.Ke3 Rb5 28.Kf3 Rb3+ (861.316.405.408) 15936

43 900:47 -2.25 15...Ke4 16.Rf8 Rc2 17.Kg2 Ke5
18.Kg1 Ke6 19.Rf3 Rd2 20.Kf1 Ke7
21.Kg1 Ne5 22.Re3 Kf6 23.Re4 Nd3
24.Re8 Kf5 25.Rf8+ Ke6 26.Rf3 Ke7
27.Kf1 Ne5 28.Rf4 Ra2 (861.316.405.408) 15936

43 900:47 -2.25 15...Ke5 16.Rf8 Ke6 17.Rf3 Ke7
18.Kf1 Nc5 19.Re3+ Kf6 20.Rf3+ Ke5
21.Re3+ Kd4 22.Re8 Ne4 23.Rd8+ Kc5
24.Rc8+ Kd6 25.Rf8 Ra2 26.Kg2 Ke5
27.Re8+ Kf5 28.Rf8+ Nf6 (861.316.405.408) 15936

or:

[pgn]
[Event "?"]
[Site "?"]
[Date "2017.08.08"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - -"]

1. Kg1 Rc2 2. Rb6 Ke4 3. Rb4+ Rc4 4. Rb6 Ng4 5. Rb3 Rc1+
6. Kg2 Rc2 7. Rf3 Ke5 8. Rf8 Ke6 9. Rf3 Rd2 10. Kg1 Ne5
11. Rf4 Nd3 12. Rf3 Rd1+ 13. Kh2 Ne5 14. Re3 Rd2 15. Kg2
(15. Kg1 Kd5) 15... Kf5 $17 *
[/pgn]

[Spliffjiffer]

after some further resurch i came to the conclusion that the r+n vs r endgame is winning for black...these are 2 mainlines and should be nearly perfect play for white

this time from the point where the last pawn has been taken (after 13.rxf6)



[d]8/8/5Rp1/2k1n2p/7P/1r4P1/4KP2/8 w - - 0 1

1. Kf1 Rb2 2. Kg1 Kd4 3. Rf4+ Kd3 4. Kf1 Ng4 5. Rf3+ Ke4 6. Rf4+ Ke5 7. Kg1 Rd2 8. Rf8 Ke6 9. Re8+ Kf7 10. Re4 Kg7 11. Re7+ Kf6 12. Re4 Ne5 13. Rf4+ Kg7 14. Re4 Nd3 15. Re7+ Kf6 16. Re8 Kf5 17. Rf8+ Ke6 18. Rf3 Rd1+ 19. Kh2 Ne5 20. Re3 Rd2 21. Kg1 Kd5 22. Ra3 Nd3 23. Ra8 Ke6 24. Re8+ Kf6 25. Rf8+ Kg7 26. Rf3 Rd1+ 27. Kh2 Ne5 28. Ra3 Rd2 29. Kg2 Nd3 30. Ra7+ Kf6 31. Ra6+ Kf5 32. Ra5+ Ke4 33. Ra4+ Kd5 34. Ra5+ Kd4 35. Ra4+ Kc3 36. Ra3+ Kc4 -+


1. Kf1 Rb2 2. Rf4 Ng4 3. Kg1 Kd6 4. Kf1 Ke6 5. Kg1 Ne5 6. Re4 Kf6 7. Re3
Rd2 8. Re4 Nd3 9. Re8 Kf5 10. Rf8+ Ke6 11. Rf3 Rd1+ 12. Kh2 Ne5 13. Re3 Rd2 14. Kg1 Kf6 15. Re4 Nd3 16. Re8 Kf5 17. Rf8+ Ke6 18. Rf3 Rd1+ 19. Kh2 Ne5 20. Re3 Rd2 21. Kg1 Kd5 22. Ra3 Nd3 23. Ra8 Ke6 24. Re8+ Kf6 25. Rf8+ Kg7 26. Rf3 Rd1+ 27. Kh2 Ne5 28. Ra3 Rd2 29. Kg2 Nd3 30. Ra7+ Kf6 31. Ra6+ Kf5 32. Ra5+ Ke4 33. Ra4+ Kd5 34. Ra5+ Kd4 35. Ra4+ Kc3 36. Ra3+ Kc4 -+

maybe so can make this diagram to replay, dont know how to do this

greetings, Peer

update: ups its not exactly the same position, sry for that
but it doesnt matter much, believe me...the ending is won

[Lyudmil Tsvetkov]

after some further resurch i came to the conclusion that the r+n vs r endgame is winning for black...these are 2 mainlines and should be nearly perfect play for white

this time from the point where the last pawn has been taken (after 13.rxf6)



[d]8/8/5Rp1/2k1n2p/7P/1r4P1/4KP2/8 w - - 0 1

1. Kf1 Rb2 2. Kg1 Kd4 3. Rf4+ Kd3 4. Kf1 Ng4 5. Rf3+ Ke4 6. Rf4+ Ke5 7. Kg1 Rd2 8. Rf8 Ke6 9. Re8+ Kf7 10. Re4 Kg7 11. Re7+ Kf6 12. Re4 Ne5 13. Rf4+ Kg7 14. Re4 Nd3 15. Re7+ Kf6 16. Re8 Kf5 17. Rf8+ Ke6 18. Rf3 Rd1+ 19. Kh2 Ne5 20. Re3 Rd2 21. Kg1 Kd5 22. Ra3 Nd3 23. Ra8 Ke6 24. Re8+ Kf6 25. Rf8+ Kg7 26. Rf3 Rd1+ 27. Kh2 Ne5 28. Ra3 Rd2 29. Kg2 Nd3 30. Ra7+ Kf6 31. Ra6+ Kf5 32. Ra5+ Ke4 33. Ra4+ Kd5 34. Ra5+ Kd4 35. Ra4+ Kc3 36. Ra3+ Kc4 -+


1. Kf1 Rb2 2. Rf4 Ng4 3. Kg1 Kd6 4. Kf1 Ke6 5. Kg1 Ne5 6. Re4 Kf6 7. Re3
Rd2 8. Re4 Nd3 9. Re8 Kf5 10. Rf8+ Ke6 11. Rf3 Rd1+ 12. Kh2 Ne5 13. Re3 Rd2 14. Kg1 Kf6 15. Re4 Nd3 16. Re8 Kf5 17. Rf8+ Ke6 18. Rf3 Rd1+ 19. Kh2 Ne5 20. Re3 Rd2 21. Kg1 Kd5 22. Ra3 Nd3 23. Ra8 Ke6 24. Re8+ Kf6 25. Rf8+ Kg7 26. Rf3 Rd1+ 27. Kh2 Ne5 28. Ra3 Rd2 29. Kg2 Nd3 30. Ra7+ Kf6 31. Ra6+ Kf5 32. Ra5+ Ke4 33. Ra4+ Kd5 34. Ra5+ Kd4 35. Ra4+ Kc3 36. Ra3+ Kc4 -+

maybe so can make this diagram to replay, dont know how to do this

greetings, Peer

update: ups its not exactly the same position, sry for that
but it doesnt matter much, believe me...the ending is won

I would very much like to believe you, but I am not able to reproduce the win.

[d]8/8/4k1p1/7p/5RnP/6P1/1r3PK1/8 b - - 0 7

this seems to be a fortress.

white plays with rook to f3,f4,f7,f8, etc., only way black can try to do something is play the rook to f7 to take the f line, but then the white rook starts checking from the side, and it seems white has sufficient space and spare moves(Kg1/g2) to avoid zugzwang.

but no one could be certain, those positions on the brink between 2 game outcomes are simply so hard for a simple verdict.

[Lyudmil Tsvetkov]

...2 clear lines that should help you (and SF also) to see that its winning

[d]8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - - 0 1

[pgn]
[Event "?"]
[Site "?"]
[Date "2017.08.08"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - -"]

1. Kg1 Rc2 2. Kf1 Ng4 3. Rf8 Ke6 4. Kg1 Ne5 5. Re8+ Kf6
6. Ra8 Nd3 7. Rd8 Rd2 8. Re8 Kf5 9. Rf8+ Ke6 10. Rf3
(10. Kf1 Ra2 11. Rf3 Ra1+ 12. Ke2 Ne5 13. Rf4 Ke7 14. Re4
Kf6 15. Re3 Kf5) 10... Rd1+ 11. Kh2 Ne5 12. Re3 Rd2 13. Kg1
Kd5 14. Ra3 Nd3 15. Ra8 Ke6 16. Re8+ Kf6 17. Rf8+ Kg7
18. Rf3 Rd1+ 19. Kh2 Ne5 20. Ra3 Rd2 21. Kg2 Nd3 22. Ra7+
Kf6 23. Ra6+ Kf5 24. Ra5+ Ke4 25. Ra4+ Kd5 26. Ra5+ Kd4
27. Ra4+ Kc3 28. Ra3+ Kc4 $17 *
[/pgn]

or:

[pgn]
[Event "?"]
[Site "?"]
[Date "2017.08.08"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - -"]

1. Kg1 Rc2 2. Rb6 Ke4 3. Rb4+ Rc4 4. Rb6 Ng4 5. Rb3 Rc1+
6. Kg2 Rc2 7. Rf3 Ke5 8. Rf8 Ke6 9. Rf3 Rd2 10. Kg1 Ne5
11. Rf4 Nd3 12. Rf3 Rd1+ 13. Kh2 Ne5 14. Re3 Rd2 15. Kg2
(15. Kg1 Kd5) 15... Kf5 $17 *
[/pgn]

in the first line, why not 20.Rf4, staying on the f file, instead of Ra3?

[Lyudmil Tsvetkov]

...2 clear lines that should help you (and SF also) to see that its winning

[d]8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - - 0 1

[pgn]
[Event "?"]
[Site "?"]
[Date "2017.08.08"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - -"]

1. Kg1 Rc2 2. Kf1 Ng4 3. Rf8 Ke6 4. Kg1 Ne5 5. Re8+ Kf6
6. Ra8 Nd3 7. Rd8 Rd2 8. Re8 Kf5 9. Rf8+ Ke6 10. Rf3
(10. Kf1 Ra2 11. Rf3 Ra1+ 12. Ke2 Ne5 13. Rf4 Ke7 14. Re4
Kf6 15. Re3 Kf5) 10... Rd1+ 11. Kh2 Ne5 12. Re3 Rd2 13. Kg1
Kd5 14. Ra3 Nd3 15. Ra8 Ke6 16. Re8+ Kf6 17. Rf8+ Kg7
18. Rf3 Rd1+ 19. Kh2 Ne5 20. Ra3 Rd2 21. Kg2 Nd3 22. Ra7+
Kf6 23. Ra6+ Kf5 24. Ra5+ Ke4 25. Ra4+ Kd5 26. Ra5+ Kd4
27. Ra4+ Kc3 28. Ra3+ Kc4 $17 *
[/pgn]

Unfortunately at least my version of Stockfish is so bad at this that it loses the solution when going upwards again, Especially if you leave it too long in a position with White to move. Part of this is Lazy SMP that will pick the highest scoring line. If this line is bad and deep at the same time it will start overwriting everything with new scores and can erase your whole solution. At least that is what I found. Up to here from the above pgn everything is still okay and it sees that Black is winning:

[D]5R2/6k1/6p1/7p/7P/3n2P1/3r1P2/6K1 w - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

28 1:03 -5.06 18.Rf3 Rd1+ 19.Kh2 Ne5 20.Ra3 Ng4+
21.Kg2 Rd2 22.Rf3 Ne5 23.Ra3 Nd3
24.Ra7+ Kf6 25.Ra6+ Kf5 26.Ra5+ Ke4
27.Ra4+ Kd5 28.Ra5+ Kd4 29.Ra4+ Kc3
30.g4 Rxf2+ 31.Kg3 (993.313.152) 15669

28 1:03 -7.25 18.Rc8 Nxf2 19.Kg2 Ng4+ 20.Kf3 Rd3+
21.Kf4 Rd4+ 22.Kf3 Rb4 23.Re8 Nh2+
24.Ke2 Rb2+ 25.Kd3 Rb3+ 26.Ke4 Rxg3
27.Rc8 Ng4 28.Rc7+ Kh6 29.Kf4 Rd3
30.Rc8 Rb3 31.Ke4 (993.313.152) 15669

27 1:03 -7.19 18.Ra8 Nxf2 19.Kf1 Rb2 20.Ra6 Kh6
21.Ra8 Ng4 22.Ra3 Rb7 23.Kg1 Re7
24.Ra8 Re3 25.Kh1 Rf3 26.Kg2 Rb3
27.Rc8 Rb2+ 28.Kh1 Rh2+ 29.Kg1 Rd2
30.Rh8+ Kg7 31.Rb8 (993.313.152) 15669

(This is backtracked from the positon after 28... Kc4)

This is still good

5R2/8/5kp1/7p/7P/3n2P1/3r1P2/6K1 b - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

34 8:44 -5.12 17...Kg7 18.Rf3 Rd1+ 19.Kh2 Ne5
20.Ra3 Ng4+ 21.Kg2 Rd2 22.Rf3 Ne5
23.Ra3 Nd3 24.Ra7+ Kf6 25.Ra6+ Kf5
26.Ra5+ Ke4 27.Ra4+ Kd5 28.Ra5+ Kd4
29.Ra4+ Kc3 30.g4 Rxf2+ (8.726.257.294) 16630

34 8:44 -2.24 17...Ke7 18.Rf3 Rd1+ 19.Kh2 Ke6
20.Re3+ Kd5 21.Rf3 Ne5 22.Rf8 Rb1
23.Rd8+ Ke6 24.Re8+ Kf6 25.Ra8 Ng4+
26.Kg2 Rb2 27.Kg1 Kf5 28.Rf8+ Ke6
29.Kg2 Kd5 30.Rf4 Rd2 (8.726.257.294) 16630

34 8:44 -2.24 17...Ke6 18.Kf1 Ra2 19.Kg1 Ke7
20.Rf3 Rd2 21.Kf1 Ne5 22.Rf4 Ke6
23.Rf8 Ra2 24.Re8+ Kf6 25.Kg1 Ng4
26.Rf8+ Kg7 27.Rf3 Rb2 28.Kg2 Rd2
29.Kg1 Ne5 30.Re3 Kf6 (8.726.257.294) 16630

One half move up

[D]4R3/8/5kp1/7p/7P/3n2P1/3r1P2/6K1 w - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

37 259:52 -4.93 17.Ra8 Ke7 18.Ra6 Nxf2 19.Rxg6 Ng4
20.Rc6 Nf6 21.Kf1 Rb2 22.Rc5 Ke6
23.Rc3 Nd5 24.Rc8 Ke5 25.Re8+ Kf5
26.Kg1 Rc2 27.Kf1 Ra2 28.Re1 Nf6
29.Re7 Rc2 30.Kg1 (259.309.940.484) 16630

37 259:52 -5.12 17.Rf8+ Kg7 18.Rf3 Rd1+ 19.Kh2 Ne5
20.Ra3 Rd2 21.Kg2 Nd3 22.Ra7+ Kf6
23.Ra6+ Kf5 24.Ra5+ Ke4 25.Ra4+ Kd5
26.Ra5+ Kd4 27.Ra4+ Kc3 28.g4 Rxf2+
29.Kg3 Rf1 30.gxh5 (259.309.940.484) 16630

37 259:52 -6.87 17.f3 Ne5 18.Kf1 Ra2 19.Rc8 Ra3
20.Kg2 Kg7 21.g4 hxg4 22.fxg4 Nxg4
23.Rc6 Rb3 24.Rc8 Ne3+ 25.Kf3 Nf5+
26.Kf2 Rb2+ 27.Kf3 Nxh4+ 28.Ke3 Nf5+
29.Kf3 Rb4 30.Rc7+ (259.309.940.484) 16630

But I waited too long and now it loses it one ply deeper in the same position...

4R3/8/5kp1/7p/7P/3n2P1/3r1P2/6K1 w - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

38 386:20 -2.26 {wrong} 17.Rf8+ Ke7 {wrong} 18.Rf3 Rd1+ 19.Kh2 Ke6
20.Re3+ Kd5 21.Rf3 Rd2 22.Kg1 Ke6
23.Rf8 Ke5 24.Rf3 Nc5 25.Ra3 Ne4
26.Re3 Kd4 27.Rf3 Ra2 28.Rf8 Ke5
29.Kg2 Nf6 30.Kg1 (387.226.768.361) 16704

38 386:20 -6.29 17.Ra8 Kf7 18.Ra7+ Ke6 19.Ra6+ Kf5
20.f3 Ne1 21.Ra5+ Kf6 22.Ra6+ Kg7
23.Ra7+ Kh6 24.f4 Nf3+ 25.Kf1 Rb2
26.f5 Nh2+ 27.Ke1 gxf5 28.Ra5 Kg6
29.Ra6+ Kg7 30.Ra7+ (387.226.768.361) 16704

37 386:20 -6.87 17.f3 Ne5 18.Kf1 Ra2 19.Rc8 Ra3
20.Kg2 Kg7 21.g4 hxg4 22.fxg4 Nxg4
23.Rc6 Rb3 24.Rc8 Ne3+ 25.Kf3 Nf5+
26.Kf2 Rb2+ 27.Kf3 Nxh4+ 28.Ke3 Nf5+
29.Kf3 Rb4 30.Rc7+ (387.226.768.361) 16704

I could still repair that going one ply up to a Black position, but a little bit further up it lost it again.


[D]R7/8/6p1/3k3p/7P/3n2P1/3r1P2/6K1 b - -

Engine: Ancalagon HT (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G.

40 18:12 -2.19 {wrong} 15...Ke6 16.Rf8 Rc2 17.Rf3 Rc1+
18.Kh2 Rd1 19.Re3+ Kd5 20.Rf3 Rd2
21.Kg1 Ne5 22.Rf8 Ke6 23.Re8+ Kf6
24.Kf1 Ra2 25.Kg1 Rc2 26.Ra8 Ng4
27.Rb8 Rd2 28.Re8 Kf7 (17.261.854.047) 15793

40 18:12 -2.19 {possibly wrong but not the main line} 15...Ke4 16.Rf8 Ra2 17.Rf6 Ne5
18.Rf4+ Kd5 19.Kg2 Re2 20.Rf8 Ke6
21.Rb8 Kd6 22.Re8 Rb2 23.Ra8 Ng4
24.Rf8 Ke7 25.Rf3 Rd2 26.Kg1 Ke6
27.Rf8 Ne5 28.Rb8 Kd5 (17.261.854.047) 15793

40 18:12 -2.19 {possibly wrong but not the main line} 15...Ke5 16.Re8+ Kd6 17.Rg8 Ne5
18.Re8 Ra2 19.Kg2 Ng4 20.Rf8 Ke6
21.Rf3 Ne5 22.Rb3 Nc4 23.Rf3 Nd6
24.Re3+ Kf5 25.Rf3+ Ke5 26.Rf4 Rc2
27.Rf3 Ke6 28.Re3+ Kf6 (17.261.854.047) 15793

or:

[pgn]
[Event "?"]
[Site "?"]
[Date "2017.08.08"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "8/8/5Rp1/3kn2p/7P/6P1/1r3P2/5K2 w - -"]

1. Kg1 Rc2 2. Rb6 Ke4 3. Rb4+ Rc4 4. Rb6 Ng4 5. Rb3 Rc1+
6. Kg2 Rc2 7. Rf3 Ke5 8. Rf8 Ke6 9. Rf3 Rd2 10. Kg1 Ne5
11. Rf4 Nd3 12. Rf3 Rd1+ 13. Kh2 Ne5 14. Re3 Rd2 15. Kg2
(15. Kg1 Kd5) 15... Kf5 $17 *
[/pgn]

this is absurd.

why not 14.Rf4 instead of Re3?

de aventouren van Siske en Wiske.

[Spliffjiffer]

because black king protects f8f7f6 and there is a fork on f3...i spended hours and hours to analyze it and you came up with such an unanalysed move...you are really funny

[Lyudmil Tsvetkov]

because black king protects f8f7f6 and there is a fork on f3...i spended hours and hours to analyze it and you came up with such an unanalysed move...you are really funny

I really don't understand what you mean by this?

that I made a mistake, or something else?

in what position the black king protects f7/f8/f6?

[Spliffjiffer]

dear Lyudmil...in my opinion ure not willing to analyse sth in depth at any point ...if u had spend 1 sec to analyse what ure talking about with 20. rf4 u would get it...how to say best for u?...hm..ok, 20. rf4 looses to rd4.