Mate in one: unsolved by Fritz & Sjeng !

[Jouni]

[d]5K2/8/2qk4/2nPp3/3r4/6B1/B7/3R4 w - e6 0 1

Analysis by Fritz 10:

1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd2 5.Rxc5+ Kxc5 6.Be6 Rd1
= (-0.11) Depth: 9/22 00:00:00 66kN, tb=4
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd8+ 5.Ke7 Rd7+ 6.Kf6 Rd2 7.Rxc5+ Kxc5 8.Be6
= (-0.10) Depth: 10/24 00:00:00 148kN, tb=23
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd8+ 5.Ke7 Rd7+ 6.Kf6 Rd2 7.Rxc5+ Kxc5 8.Be6
= (-0.10) Depth: 11/19 00:00:00 180kN, tb=24
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd8+ 5.Ke7 Rd7+ 6.Kf6 Rd2 7.Rxc5+ Kxc5 8.Be6
= (-0.10) Depth: 12/25 00:00:00 337kN, tb=31
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Rd2 4.Bc4 Kxc6 5.Kf7 Rd7+ 6.Kf6 Rd4 7.Re6+ Kc7 8.Re7+ Kb6 9.Re6+ Ka5
= (-0.11) Depth: 13/27 00:00:00 752kN, tb=84

Analysis by Deep Sjeng 2.5:

1.dxc6 Rxd1 2.Bc4
-+ (-1.49) Depth: 3/12 00:00:00
1.Bxe5+
µ (-1.38) Depth: 3/12 00:00:00
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6
= (-0.08) Depth: 3/12 00:00:00
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6
= (-0.08) Depth: 4/13 00:00:00
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Be6
= (-0.05) Depth: 5/14 00:00:00
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Be6 Kd6 5.Bf7 Nd3 6.Rb1
= (-0.04) Depth: 6/19 00:00:00 26kN, tb=4
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Be6 Kd6 5.Bf5 Rf4 6.Rd1+ Ke5
= (-0.04) Depth: 7/23 00:00:00 54kN, tb=6
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Bf7 Kd6 5.Kg7 Rg4+ 6.Kf6 Rf4+ 7.Kg7 Nd3 8.Re8
= (-0.04) Depth: 8/23 00:00:00 100kN, tb=14
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Bf7 Kd6 5.Kg7 Rg4+ 6.Kf6 Rf4+ 7.Kg7 Nd3 8.Re8
= (-0.04) Depth: 9/26 00:00:00 175kN, tb=41
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Ke7 Nd3 5.Re6+ Kc5 6.Re2 Rb4 7.Rc2+ Kd4 8.Be6 Ke5 9.Ra2
= (-0.03) Depth: 10/27 00:00:01 431kN, tb=142
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Ke7 Nd3 5.Re6+ Kc5 6.Re2 Nb4 7.Be6 Nd5+ 8.Bxd5 Kxd5 9.Rc2 Rf4 10.Rc8
= (-0.02) Depth: 11/29 00:00:02 989kN, tb=405

Too difficult

Jouni

[F. Bluemers]

Crafty 19.1 finds it.
Ktulu 7.1 too
rybka 1.0 beta too
shredder 9 too
spike 1.2 too

CT 2007.1 native not .
CT2007.1 uci in shredder gui does find it however.
CT2007.1 UCI in lokasoft gui doesn't start analysis unless a move is done

btw the native finds it after playing what has been the last black move, so would sjeng and fritz(at least i assume they will .

Best
Fonzy

[Uri Blass]

This is not correct that the mate is not solved by Fritz

New game
5K2/8/2qk4/2nPp3/3r4/6B1/B7/3R4 w - e6 0 1

Analysis by Fritz 8:

1.d5xe6#
+- (#1) Depth: 2/4 00:00:00

(, 02.05.2007)

New game
5K2/8/2qk4/2nPp3/3r4/6B1/B7/3R4 w - e6 0 1

Analysis by Fritz 6 Light:

1.d5xc6!
-+ (-2.09) Depth: 1/4 00:00:00
1.d5xe6#!
+- (#1) Depth: 1/7 00:00:00
1.d5xe6#
+- (#1) Depth: 2/2 00:00:00

(, 02.05.2007)

[chessfurby]

This is not too difficult. The problem might be the fact that it is an en passant move which leads to mate. If you setup a position from blank using "edit board" some engines might not understand that capture en passant is a valid move in this constructed position.

[Mike S.]

Yes, in fact it is a test for the e.p. part of the EPD, only. It could also be an interface issue if this isn't solved. For example, I have an old GUI which is very good in general, but ignores the e.p. square in EPD, both for copy&paste import and export. As a workaround to test for solutions which start with an en passant capture and at the same time avoid such problems, the position can be modified like this:

[d]5K2/4p3/2qk4/2nP4/3r4/6B1/B7/3R4 b - - 0 1
And then start the test by making the pawn move (here, e7-e5).

I remember a related discussion about mate solving capabilities (for chess problems), where it is expected from the solver that he figures out by himself, if the last move must have been a pawn's double step, thus allowing an e.p. capture to be taken into consideration. In other words, there seem to be rules for chess problems which contain that no information about an e.p. square is provided and the solver must figure out that, himself.

Of course, the required retroanalysis cannot be expected from normal chess engines. I wouldn't even expect or demand that from dedicated problem solving engines... maybe it would cause compatibility problems. I mean, if the engine doesn't receive an e.p. square information but would find there must be one, by a retroanalysis phase prior to the normal "forward" calculations, and further that it's the solution move: It would have to output an e.p. capture which is impossible/illegal in the position the GUI knows, which doesn't contain the e.p. square.

Although I can imagine that it would be - or is? - a useful or even required feature for a dedicated chess problem software. I didn't use such programs yet, only engines like ChestUCI in standard interfaces.

[Jouni]

Of course this is ep problem! But it's funny, that Fritz10 can't solve this, when most UCI engines under Fritz interface can.

Jouni

PS. Sorry for wrong forum

[tiger]

[d]5K2/8/2qk4/2nPp3/3r4/6B1/B7/3R4 w - e6 0 1

Analysis by Fritz 10:

1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd2 5.Rxc5+ Kxc5 6.Be6 Rd1
= (-0.11) Depth: 9/22 00:00:00 66kN, tb=4
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd8+ 5.Ke7 Rd7+ 6.Kf6 Rd2 7.Rxc5+ Kxc5 8.Be6
= (-0.10) Depth: 10/24 00:00:00 148kN, tb=23
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd8+ 5.Ke7 Rd7+ 6.Kf6 Rd2 7.Rxc5+ Kxc5 8.Be6
= (-0.10) Depth: 11/19 00:00:00 180kN, tb=24
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd8+ 5.Ke7 Rd7+ 6.Kf6 Rd2 7.Rxc5+ Kxc5 8.Be6
= (-0.10) Depth: 12/25 00:00:00 337kN, tb=31
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Rd2 4.Bc4 Kxc6 5.Kf7 Rd7+ 6.Kf6 Rd4 7.Re6+ Kc7 8.Re7+ Kb6 9.Re6+ Ka5
= (-0.11) Depth: 13/27 00:00:00 752kN, tb=84

Analysis by Deep Sjeng 2.5:

1.dxc6 Rxd1 2.Bc4
-+ (-1.49) Depth: 3/12 00:00:00
1.Bxe5+
µ (-1.38) Depth: 3/12 00:00:00
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6
= (-0.08) Depth: 3/12 00:00:00
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6
= (-0.08) Depth: 4/13 00:00:00
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Be6
= (-0.05) Depth: 5/14 00:00:00
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Be6 Kd6 5.Bf7 Nd3 6.Rb1
= (-0.04) Depth: 6/19 00:00:00 26kN, tb=4
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Be6 Kd6 5.Bf5 Rf4 6.Rd1+ Ke5
= (-0.04) Depth: 7/23 00:00:00 54kN, tb=6
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Bf7 Kd6 5.Kg7 Rg4+ 6.Kf6 Rf4+ 7.Kg7 Nd3 8.Re8
= (-0.04) Depth: 8/23 00:00:00 100kN, tb=14
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Bf7 Kd6 5.Kg7 Rg4+ 6.Kf6 Rf4+ 7.Kg7 Nd3 8.Re8
= (-0.04) Depth: 9/26 00:00:00 175kN, tb=41
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Ke7 Nd3 5.Re6+ Kc5 6.Re2 Rb4 7.Rc2+ Kd4 8.Be6 Ke5 9.Ra2
= (-0.03) Depth: 10/27 00:00:01 431kN, tb=142
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Ke7 Nd3 5.Re6+ Kc5 6.Re2 Nb4 7.Be6 Nd5+ 8.Bxd5 Kxd5 9.Rc2 Rf4 10.Rc8
= (-0.02) Depth: 11/29 00:00:02 989kN, tb=405

Too difficult

Jouni

Thank you for posting this position and point out that some programs have problems with it.

Chess Tiger 2007.1 (native version) has this bug as well and will be fixed soon.


// Christophe

[F. Bluemers]

Thank you for posting this position and point out that some programs have problems with it.

Chess Tiger 2007.1 (native version) has this bug as well and will be fixed soon.


// Christophe

maybe it would be also nice to change ct 2007 sending 0.01 as eval for a book move,repeatingly geting 0.00 some gui(. ..) take that for a draw offer (most engines send 0.01 to circumvent that..).
This game didn't happen in the cp gui and i hesitate to call it a bug

[Event "Partij met rating, 60m + 0s"]
[Site "Computers toegestaan"]
[Date "2007.05.03"]
[Round "?"]
[White "Naum Chess, Deep Sjeng 2.5"]
[Black "Fonzy, Chess Tiger 2007"]
[Result "1/2-1/2"]
[ECO "D16"]
[WhiteElo "2537"]
[PlyCount "19"]
[EventDate "2007.05.05"]
[TimeControl "3600"]

1. d4 {0.00/0 0} d5 {0.00/1 2} 2. c4 {0.00/0 0} c6 {0.00/1 0} 3. Nf3 {0.00/0 0}
Nf6 {0.00/1 0} 4. Nc3 {0.00/0 0} dxc4 {0.00/1 0} 5. a4 {0.00/0 0} Na6 {0.00/1 0
} 6. e4 {0.00/0 0} Bg4 {0.00/1 0} 7. Bxc4 {
Fonzy,Chess Tiger 2007 doet een remise aanbod 0.00/0 0} e6 {0.00/1 0} 8. Be3 {
0.00/0 0} Nb4 {0.00/1 0} 9. O-O {0.00/0 0} Be7 {0.00/1 0} 10. Be2 {0.00/0 0}
1/2-1/2

[/quote]

[tiger]

Thank you for posting this position and point out that some programs have problems with it.

Chess Tiger 2007.1 (native version) has this bug as well and will be fixed soon.


// Christophe

maybe it would be also nice to change ct 2007 sending 0.01 as eval for a book move,repeatingly geting 0.00 some gui(. ..) take that for a draw offer (most engines send 0.01 to circumvent that..).
This game didn't happen in the cp gui and i hesitate to call it a bug

[Event "Partij met rating, 60m + 0s"]
[Site "Computers toegestaan"]
[Date "2007.05.03"]
[Round "?"]
[White "Naum Chess, Deep Sjeng 2.5"]
[Black "Fonzy, Chess Tiger 2007"]
[Result "1/2-1/2"]
[ECO "D16"]
[WhiteElo "2537"]
[PlyCount "19"]
[EventDate "2007.05.05"]
[TimeControl "3600"]

1. d4 {0.00/0 0} d5 {0.00/1 2} 2. c4 {0.00/0 0} c6 {0.00/1 0} 3. Nf3 {0.00/0 0}
Nf6 {0.00/1 0} 4. Nc3 {0.00/0 0} dxc4 {0.00/1 0} 5. a4 {0.00/0 0} Na6 {0.00/1 0
} 6. e4 {0.00/0 0} Bg4 {0.00/1 0} 7. Bxc4 {
Fonzy,Chess Tiger 2007 doet een remise aanbod 0.00/0 0} e6 {0.00/1 0} 8. Be3 {
0.00/0 0} Nb4 {0.00/1 0} 9. O-O {0.00/0 0} Be7 {0.00/1 0} 10. Be2 {0.00/0 0}
1/2-1/2

[/quote]


And what about sending a +Mate score? Will "some GUI" believe that Chess Tiger has won the game?

Aha, I'm going to make some easy elo points then!


// Christophe

[Dan Honeycutt]

Of course, the required retroanalysis cannot be expected from normal chess engines.

Expecting the engine to do a retroanalysis is too much but it is not relevant here as there are other ways to arrive at the position besides the pawn advance. Everything the engine needs is in the FEN and there is no excuse for not finding the mate.

Best
Dan H.

(And yes, before I opened my fat mouth I checked to make sure that Bruja found the mate.)