It takes a while to prove this one using Chest319

[Dann Corbit]

[d]r1b2rk1/pp1n1pp1/1b5p/3B4/2P4P/2B2Nq1/PPQ3P1/1K1R3R w - - acn 3518850511; acs 514295; bm Ng5; ce 32748; dm 10; pv Ng5 Qxg5 hxg5 Ne5 gxh6 Bh3 Bxe5 Rfd8 hxg7 Rxd5 Rxh3 f5 Rh8+ Kf7 Qxf5+ Ke7 Qf6+ Kd7 Rxd5#;

Upon analyzing alternative moves deeply, it seems that there may be other winning moves, but the wins will be a lot deeper:
[d]r1b2rk1/pp1n1pp1/1b5p/3B4/2P4P/2B2Nq1/PPQ3P1/1K3R1R b - - acd 49/80; acs 2652; acn 60435247k; ce -948; bm Bd8; pv Bd8 Ng5 Bxg5 hxg5 Qxg5 Bd2 Qg3 Bf4 Qg4 Bxh6 Nf6 Bg5 Nh5 Be7 g6 Bxf8 Bf5 Rxf5 Qxf5 Qxf5 gxf5 Rxh5 Rxf8 Bxb7 f4 Rg5+ Kh8 Rf5 Rb8 Bf3 Kg7 Rxf4 Rc8 Kc2 a5 Kc3 Rc5 Rd4 Re5 Rd5 Re1 Kc2 Re7 c5 Rc7 Kc3 Rc8 b3 Rb8 c6 Rc8 Rd4 Rb8 Kc4 Re8 Kb5 Rc8 Kb6 Kf6 Kb7 Rxc6 Kxc6 Ke5 Rd3 Kf4;
[d]r1b2rk1/pp1n1pp1/1b5p/3B4/2P4P/2B2Nq1/PPQ3P1/1K1RR3 b - - acd 56/98; acs 2763; acn 54789038k; ce -509; bm Qg4; pv Qg4 Rf1 Nf6 Ng5 hxg5 Rxf6 Be6 Bxe6 fxe6 Rd7 gxh4 Qg6 Qxg6+ Rxg6 Rf7 Rxf7 Kxf7 Rxg7+ Kf8 Rxb7 Rc8 b3 Ke8 Rh7 Rc5 Rxh4 Rg5 g4 Bf2 Rh8+ Kf7 Rh7+ Kg8 Rh2 Be3 Rg2 Kf7 Kc2 Bc5 Kd3 Ba3 Bd2 Rg7 g5 Kg6 Ke4 Rf7 Ke5 Rf5+ Kxe6 Rf3 Kd7 Rf5 Kc6 Rc5+ Kb7 Re5 Kxa7 Re7+ Ka6 Bd6 Ba5;

[Dann Corbit]

More info
r1b2rk1/pp1n1pp1/1b5p/3B4/2P4P/2B2Nq1/PPQ3P1/1K1R3R w - - acd 999; bm Ng5; ce 32748; c0 "Jaan Ehlvest vs Sten Safholm, Stockholm, 1997"; c1 "#10"; c2 "Source:wtharvey.com"; c3 "Ng5"; c4 "Ehlvest, Jaan; pm=Ng5; 1-0;"; dm 10; pm Ng5; pv Ng5 Qxg5 hxg5 Ne5 gxh6 Bh3 Bxe5 Rfd8 hxg7 Rxd5 Rxh3 f5 Rh8+ Kf7 Qxf5+ Ke7 Qf6+ Kd7 Rxd5#;

[peter]

No problem for SF dev. (30 threads of 16x3.5GHz CPU)

r1b2rk1/pp1n1pp1/1b5p/3B4/2P4P/2B2Nq1/PPQ3P1/1K1R3R w - - 0 1

Analysis by Stockfish dev-20231105-442c294a:

Code: Select all

1.Rh3 
  White is clearly better: +- (1.80)  Depth: 1/1   00:00:01  58kN
1.Ng5 
  White has a decisive advantage: +- (2.93)  Depth: 2/2   00:00:01  348kN
1.Ng5 Nf6 
  White is clearly winning: +- (8.87)  Depth: 3/2   00:00:01  349kN
1.Ng5 Nf6 2.Bxf6 Qxg5 
  White is clearly winning: +- (8.87)  Depth: 4/4   00:00:01  349kN
1.Ng5 Nf6 2.Bxf6 Qxg5 3.hxg5 
  White is clearly winning: +- (8.87)  Depth: 5/5   00:00:01  350kN
1.Ng5 Nf6 2.Bxf6 Qxg5 3.hxg5 Be6 
  White is clearly winning: +- (9.04)  Depth: 6/6   00:00:01  351kN
1.Ng5 Nf6 2.Bxf6 Qxg5 3.hxg5 Be6 4.Bxe6 
  White is clearly winning: +- (9.04)  Depth: 7/6   00:00:01  352kN
...
1.Ng5 Qxg5 2.hxg5 Ne5 3.gxh6 Bd4 4.hxg7 Bh3 5.Rxh3 Kxg7 6.Qh7+ Kf6 7.Qh4+ Kf5 8.Rf1+ Bf2 9.Rxf2+ Nf3 10.Rfxf3+ Kg6 11.Qh6# 
  White mates: +- (#11)  Depth: 31/22   00:00:02  19431kN
1.Ng5 Qxg5 2.hxg5 Ne5 3.gxh6 Bd4 4.Bxd4 Rd8 5.hxg7 Kxg7 6.Bxe5+ Kf8 7.Bf6 Bh3 8.Rxh3 Ke8 9.Bc6+ bxc6 10.Rh8# 
  White mates: +- (#10)  Depth: 32/22   00:00:02  22955kN
...
1.Ng5 Qxg5 2.hxg5 Ne5 3.gxh6 Bd4 4.Bxd4 Rd8 5.hxg7 Kxg7 6.Bxe5+ Kf8 7.Bf6 Bh3 8.Rxh3 Ke8 9.Bc6+ bxc6 10.Rh8# 
  White mates: +- (#10)  Depth: 39/20   00:00:07  87976kN

BTW, Lc0 is at 1.Ng5 as best move at once too, giving DTM also, but a little too short a line because of the one and another one suboptimal moves for Black
3070ti GPU, net 2860M:

r1b2rk1/pp1n1pp1/1b5p/3B4/2P4P/2B2Nq1/PPQ3P1/1K1R3R w - - 0 1

Analysis by Lc0 v0.31.0-dag+git.ee08b3b1:

1.Ng5 Nf6 2.Bxf6 hxg5 3.hxg5 Qh3 4.Rxh3 Bxh3 5.Qg6 Bf5+ 6.Qxf5 Rfe8 7.Qg6 Rf8 8.Qxg7#
+- (18.12) Depth: 10/20 00:00:08 30kN
White is clearly winning

(1...Nf6?!, 6....Rfe8?!),

Berserk 12 shows #10 within seconds too, regards

[Dann Corbit]

I realize many engines can solve it. But only Chest or Gustav, Popeye or other provers can prove it.

[peter]

I realize many engines can solve it. But only Chest or Gustav, Popeye or other provers can prove it.

Code: Select all

*****  Gustav 4.2 e   64 Bit  *****    Berechnung beendet am  8.11.2023  20:50

Matt in 10 Zügen, Weiß beginnt, ,   

Lösung gefunden bei folgenden Parametern: 
Fluchtfelder: 2
alle Lösungen suchen / keine Duale in Kurzvarianten anzeigen
Zugzahl unbekannt / Hashtabelle: 8192 MB

Lösezeit:  :           10 : 03,47 min      Zaehler: 602.246797

1.Sf3-g5 Dg3-g5 2.Bh4-g5 Sd7-e5 3.Th1-h6 Tf8-d8 4.Dc2-h7 Kg8-f8 5.Lc3-b4 Lb6-c5 6.Lb4-c5 Td8-d6 7.Th6-d6 Lc8-f5 8.Ld5-e4 Se5-d7 9.Td6-e6 Sd7-c5 10.Dh7-h8#

Only main line of solution, full one's too big, I'd have to upload or send by mail. The flights defined by two was my setting, even if I thought them to be at least 3, I simply gave it a try. So here we are at a point to doubt even special program's results, the parameters you set could influence the results too.
With automatic parameters Gustav's about just as fast here yet too anyhow:

Code: Select all

*****  Gustav 4.2 e   64 Bit  *****    Berechnung beendet am  8.11.2023  21:30

Matt in 10 Zügen, Weiß beginnt, ,   

Lösung gefunden bei folgenden Parametern: 
automatisch eingestellt: Fluchtfelder: 2
sonstige Parameter: alle Lösungen suchen / keine Duale in Kurzvarianten anzeigen
                    Zugzahl bekannt / Hashtabelle: 8192 MB
Lösezeit:  :           10 : 06,08 min      Zaehler: 600.684713

1.Sf3-g5 Dg3-g5 2.Bh4-g5 Sd7-e5 3.Th1-h6 Tf8-d8 4.Dc2-h7 Kg8-f8 5.Lc3-b4 Lb6-c5 6.Lb4-c5 Td8-d6 7.Th6-d6 Lc8-f5 8.Ld5-e4 Se5-d7 9.Td6-e6 Sd7-c5 10.Dh7-h8#

[peter]

With automatic parameters Gustav's about just as fast here yet too anyhow

And Brute Force Gustav takes not quite 5 hours for full solution of all possible lines, ChestUCI 5.2 didn't come to end of computing overnight with 3 KingSquares, regards