3-move checkmate puzzle that many engines have trouble with

[Chessnut1071]

FEN[5810] = "8/1n3Np1/1N4Q1/1bkP4/p1p2p2/P1P2R2/3P2PK/B2R4 w - - 0 1 "; // Frank Healey (1861) -3-move mate [Rh1]

This problem uses a tricky "check" with "ent passant capture". Nice way to verify your ent passant code.

[Ajedrecista]

Hello Bill:

FEN[5810] = "8/1n3Np1/1N4Q1/1bkP4/p1p2p2/P1P2R2/3P2PK/B2R4 w - - 0 1 "; // Frank Healey (1861) -3-move mate [Rh1]

This problem uses a tricky "check" with "ent passant capture". Nice way to verify your ent passant code.

Sorry, but I do not see the tricky check with en passant capture. Can you explain it, please? Thank you in advance.

https://yacpdb.org/#50486

Regards from Spain.

Ajedrecista.

[Chessnut1071]

Hello Bill:

FEN[5810] = "8/1n3Np1/1N4Q1/1bkP4/p1p2p2/P1P2R2/3P2PK/B2R4 w - - 0 1 "; // Frank Healey (1861) -3-move mate [Rh1]

This problem uses a tricky "check" with "ent passant capture". Nice way to verify your ent passant code.

Sorry, but I do not see the tricky check with en passant capture. Can you explain it, please? Thank you in advance.

https://yacpdb.org/#50486

Regards from Spain.

Ajedrecista.

[Chessnut1071]

Hello Bill:

FEN[5810] = "8/1n3Np1/1N4Q1/1bkP4/p1p2p2/P1P2R2/3P2PK/B2R4 w - - 0 1 "; // Frank Healey (1861) -3-move mate [Rh1]

This problem uses a tricky "check" with "ent passant capture". Nice way to verify your ent passant code.

Sorry, but I do not see the tricky check with en passant capture. Can you explain it, please? Thank you in advance.

https://yacpdb.org/#50486

Regards from Spain.

Ajedrecista.

The tricky move t saw that fooled an engine was the Pg4 and PxP ep+, The engine didn't anticipate the check capture after ep and missed the 3-move mate. I'm not sure how the engine got into that sequence, but some engines don't look for check after e.p.

[MOBMAT]

I don't know what you are referring to. White doesn't need to move a pawn to win in 3.

Code: Select all

FEN: 8/1n3Np1/1N4Q1/1bkP4/p1p2p2/P1P2R2/3P2PK/B2R4 w - - 0 1
 5	00:00	 115,279	2,561,755	+M3	1.Rh1 Bd7 2.Qb1 Nd6 3.Qb4+
 6	00:00	 200,954	2,450,658	+M3	1.Rh1 Bd7 2.Qb1 Nd6 3.Qb4+
 7	00:00	 1,417,278	2,940,410	+M3	1.Rh1 Bd7 2.Qb1 Nd6 3.Qb4+
 8	00:01	 3,324,621	2,873,484	+M3	1.Rh1 Bd7 2.Qb1 Nd6 3.Qb4+

The Rh1 moves opens up the g1a7 diagonal for the queen after it moves to b1 on move 2.

It is a super simple mate in 3. Some humans might miss it, though.

[Chessnut1071]

I don't know what you are referring to. White doesn't need to move a pawn to win in 3.

Code: Select all

FEN: 8/1n3Np1/1N4Q1/1bkP4/p1p2p2/P1P2R2/3P2PK/B2R4 w - - 0 1
 5	00:00	 115,279	2,561,755	+M3	1.Rh1 Bd7 2.Qb1 Nd6 3.Qb4+
 6	00:00	 200,954	2,450,658	+M3	1.Rh1 Bd7 2.Qb1 Nd6 3.Qb4+
 7	00:00	 1,417,278	2,940,410	+M3	1.Rh1 Bd7 2.Qb1 Nd6 3.Qb4+
 8	00:01	 3,324,621	2,873,484	+M3	1.Rh1 Bd7 2.Qb1 Nd6 3.Qb4+

The Rh1 moves opens up the g1a7 diagonal for the queen after it moves to b1 on move 2.

It is a super simple mate in 3. Some humans might miss it, though.

The point I was referring to was the check after the e.p. Some engines fail to look for check after e.p. It's very rare that it has any effect, like 1:15,000. You are correct, Pg4 has no bearing on the 3-move mate.

[hgm]

Fairy-Max only finds mate in 4. Is this a problem that involves zugzwang?

[Andrew]

Fairy-Max only finds mate in 4. Is this a problem that involves zugzwang?

Yes , the yacpdb link mentions zugzwang after the first move. I have seen many engines give a 4 move mate that
doesn't involve any en passant lines. Probably pruning too much (null move related).

Andrew

[hgm]

Indeed, without verification zugzwang cannot be exploited as long as null move is used.

[MOBMAT]

[/quote]
The point I was referring to was the check after the e.p. Some engines fail to look for check after e.p. It's very rare that it has any effect, like 1:15,000. You are correct, Pg4 has no bearing on the 3-move mate.
[/quote]

I doubt any engine will miss seeing a check after an e.p. or else their PERFT test would fail.