Nice chess problems

[Christopher Kreuzer]

While reading through a computer chess paper, I came across a concept I had never seen before. I won't say what it is (as that will give away the problem solution below), but I got to wondering if it had been used in chess problems, and found this one:

Erich Anselm Brunner, Deutsches Wochenschach 12/1908, Selfmate In 3

8/8/5Q2/2p3Pk/5R1P/5r2/1PPKB1P1/2NRB3

It is a self-mate, but it is a lovely concept at the end when you manage to solve it!

For those not familiar with this, White moves first and forces Black to deliver mate on move 3.

I wonder if the concept showcased here has been used in 'normal' problems? Not found one yet.

[Christopher Kreuzer]

If the above is too hard, here is a "normal" chess problem that uses the same concept (though slightly differently):

Pospíšil, Karel
Humoristické Listy, 27 Feb 1885 (9/325, p. 75).

2B4K/Q5p1/3p1p2/3pk3/4p3/6p1/3P4/8

White to move. Mate in 3.

[Jon D'Souza-Eva]

I found the first one quite easy but I'm not getting anywhere with the second.

[IM Jack Rudd]

OK, let's see with the second one...

The set play is 1...d4 2.Qa5+ and then 2...d5 3.Qc7# or 2...Kf4 3.Qf5#, and 1...Kf4 2.Qe3+ Ke5 3.d4#, so we want something that meets 1...g2 while retaining those mates. So how about 1.Qa3? That retains those two lines and then we have:
1...Kd4 2.Qc3#
1...g2/g5/g6 2.Qg3+ Kd4 3.Qc3#
1...f5 2.Qxg3+ and then 2...Kd4 3.Qc3# or 2...f4/Kf6 3.Qxg7#

[Christopher Kreuzer]

I think Jack got nearly all the lines for the second one (the mate in 3), except the 1...e3 2.Qxe3# set play. It is the line ending in 3.d4# that is the clue for the first puzzle (the selfmate in 3), if people need clues for that. The logic is remorseless there once you start from which Black piece gives the mate and what White needs to do to force it.

Can anyone (giving people time to try and solve them) say what the common element is, and what is the best way to find other examples of this? The problem database I got these from does have a keyword that encompasses this, but most of the examples are not this specific sort of element (it combines a common chess tactic and a well-known chess move).

[Jon D'Souza-Eva]

The theme must be mate with a pawn that can't be taken en passant.

[IM Jack Rudd]

This selfmate is frustrating. If the white queen were a rook instead, there would be selfmate in two with 1.b3 c4 2.b4 c3#, and if the white pawn were already on b3, there would likewise be selfmate in two with 1.Qe6/d6/b6/a6 c4 2.b4 c3#. But as it is, there's a bit of a traffic jam to sort out.

[Jon D'Souza-Eva]

The way I did it was to try to work out what Black's extra move was in order to allow 3. ... c3#

[Christopher Kreuzer]

Indeed. Black's extra move must logically be an extra move with the king (as the Black rook is pinned). Once you work out where the Black king ends up (and White needs to force that king move), the rest falls into place. I'll say more about Jon's comment later. There are some interesting variations on this theme.

EDIT: It is interesting that Jack found the second one easier than the first, and the difficulty was reversed for Jon!

[Christopher Kreuzer]

Unless anyone wants to post sooner, I will post the solution to the self mate in 3 at some point tomorrow (Thursday).

[Graham Borrowdale]

Finally solved the first one - very nice!

Spoiler Alert!

White has to play b4, with the black pawn at c4, at a point when cxb3 ep is not legal, so the rook has to go to a4 and the king forced to g4, so that cxb3 ep would be check. So 1 Ra4 c4, 2 Qf7+ Kg4, 3 b4 c3#. Hope that’s right.

[Christopher Kreuzer]

That's exactly right, Graham! Congratulations.

What I was hoping for was other examples of where en passant is made illegal by the circumstances of the position on the board. I think I am right in saying that only the bishop, queen and rook can do this sort of pinning (the queen can do this both diagonally and along rows and columns). I can't think of any other way in which this can happen.

Challenges to any composers out there!

(1) An endgame study (or position fragment) where the illegality of an en passant capture is key to the solution.

(2) A chess problem (or position fragment) where the illegality of more than one en passant capture (whether the same pawn or different ones) is seen in the main lines, or the set play, or sub-variations.

(3) A chess problem (or position fragment) where all three types of pieces (rook, queen and bishop) are involved in preventing an en passant capture (not necessarily the same capture or the same pawn), but that would be ideal). Failing this, just two pieces (whether two bishops, two queens, two rooks, or rook and queen, or rook and bishop, or bishop and queen).

The presence of a disallowed en passant capture does not have to be at the point of mate. It can be elsewhere, as in the self-mate in 3 above where the lack of the en passant capture allowed (forced!) the pushing of the pawn to deliver mate.

Some of these challenges may be impossible (and others may have tried and failed)! Can the limits of this challenge be proven? Chess problem databases tend to use the keyword "anti en passant" to categorise these sort of problems.

[The differences between pinning along a row is that both pawns feature in the pin, as opposed to along a diagonal, where only one of the pawns can feature in the pin, either the one that is pushed or the one that tries to do the capture - there are some subtle variants on this, but I can't quite work them all out logically quite yet.]