12
$\begingroup$

A hypothetical game of chess reached the following position:

Your goal is to determine, conclusively, how the game would end with best play by both sides. But there is a catch:

Instead of telling you whose turn it is, you only know the following information:

  • This position was reached by making only "monochromatic" moves from the starting position (i.e., pieces never deviated from the square color they started on--for example, no player ever castled queenside, because that would change the rook's square color).
  • The entire move sequence played in the hypothetical game leading to this position is optimal, in the sense that no shorter monochromatic move sequence could have produced the same position.

Clarification: the best continuation after the position is reached no longer needs to satisfy the above conditions. Overall, this is just a chess game that happens to satisfy the retrograde conditions up until the position shown was reached.

As a bonus, can you figure out whether the move sequence satisfying the above constraints is unique? If so, prove it; if not, determine how many possible game histories there are.

$\endgroup$
5
  • 1
    $\begingroup$ There really should be a monochromatic-chess tag now :) $\endgroup$ Commented Dec 23, 2024 at 7:55
  • 1
    $\begingroup$ By the way, you only specified that the history is monochromatic. But of course you also mean the future has to be monochromatic, right? $\endgroup$ Commented Dec 23, 2024 at 7:57
  • $\begingroup$ Thanks for the clarification @BenjaminWang I actually did not mean for the future to be monochromatic (it's just a regular chess puzzle once you figure out whose turn it is) I'll add a clarification to the original post. $\endgroup$ Commented Dec 23, 2024 at 8:36
  • 1
    $\begingroup$ Are pawns also subject to the rule? $\endgroup$ Commented Dec 23, 2024 at 10:59
  • 1
    $\begingroup$ @acegikmoqsuwy2000 brilliant!! I got beaten like eggs by this puzzle. $\endgroup$ Commented Dec 23, 2024 at 20:35

2 Answers 2

7
$\begingroup$

Here is a proof game in

19.0 moves

which goes like this:

1. b4 f5 2. Bb2 Kf7 3. Bxg7 Kg6 4. Bxh8 Kh5 5. Bd4 Bg7 6. Bxa7 Bc3 7. Bxb8 Rxa2 8. dxc3 Ra6 9. Ra5 Re6 10. Rc5 b5 11. h4 Bb7 12. Rh3 Bd5 13. Qxd5 Kg4 14. Rd3 Re4 15. Qxe4+ fxe4 16. Kd2 exd3 17. Kc1 dxc2 18. Kb2 cxb1=R+ 19. Ka3 Rxf1

To see that this is optimal, we will have to take a closer look at the position (I won't use spoiler blocks from here on out):

Obviously, the black rook on f1 is the promoted f7 pawn. In order to promote, it needed to capture at least four times. Since all dark squared white pawns (in fact, all dark squared white pieces) are still on the board, all these captures needed to consume light squared pieces. Since the Bf1 was not available, there remain only five possible pieces - Rh1, Qd1, Nb1, Pa2 and Pc2. In particular, this implies that Pf7 could not capture six times and therefore had to start its journey with a double step. As a consequence, it could not possibly have captured Pa2 - in order to do so, this pawn would have had to promote itself, which is impossible, since it could capture at most three times (rook, bishop and a pawn e.p. once).

Hence, Pf7 had to capture Nb1 to promote, which fixes his journey as f7-f5xe4xRd3xc2xNb1. (The Rh1 cannot reach any other square to be captured.)

From this fragment of the history, we can already conclude that White spent at least four moves with their light squared pieces - two moves with the rook and two moves bringing the queen to e4 (the alternative, bringing the Pc2 to e4, would need at least five moves altogether since the queen cannot be captured at home).

Before tying this in with Black's moves on light squares, we will now take a look at the plays on dark squares. Here, Black is missing quite a bit of material. Clearly, the Bf8 was captured on c3 (since no other black piece could reach that square), but the two missing pawns, the Rh8 and Nb8 had to be captured by white pieces. It is straightforward enough to see that the most efficient way to go about this is by using the Bc1 to capture at least Rh8 and then Pa7 and Nb8. The missing g7 pawn could have either been captured by the Bc1 on the way to the corner, or maybe by the Ra1 after taking a double step. Either way, the capture of this pawn consumes at least one extra move for either the rook or the bishop, which leads to a total of at least 15 white moves on dark squares: Three pawn moves, four king moves, two (three) rook moves and five (six) bishop moves.

This already brings us up to the 19 allowed moves by White. In order to achieve this minimum, we cannot spend any extra moves at all - in particular this means that Black's light squared rook and bishop needed to be captured using only the very few moves on light squares that are available to White.

But these two pieces are already quite constrained for time - if we tally up all known black moves so far, we already arrive at a total of 13 moves (four by the king, two by Bf8 to reach c3, one by Pb7, five by Pf7 to promote and finally one for Rb1xBf1).

This leaves only six more moves for the two unaccounted pieces to reach convenient squares for capture AND capture White's missing Pa2 at home.

In fact, since the first move by the white queen must be vertical (Pc2 is still blocking the path at that time), the only squares where these captures can possibly occur, are d3, d5, e4 (by the queen) and h3, d3 (by the rook). Note that among these, only e4 can be reached by Ra8, which leaves no degrees of freedom on how these two pieces move:

To see why, I will first claim that the White must have already played b4 at this point. Otherwise, we run into trouble at the start of the game - before opening Bc1 with either b4 or dxc3, White can only move h4 and Rh3 (not Rd3 because it blocks the path of the queen later). But this is not enough time for the black Bf8 to serve itself up on c3. Hence, one of the first white moves must be b4, which constrains the path of Ra8 later.

Indeed, this shows that Pa2 cannot be captured by the bishop - to reach it and get out would take four moves, leaving only two for Ra8 to reach e4 - impossible.

As such, the pawn must be captured by the rook, which then still needs three more moves to reach e4 and leaves exactly two moves for the bishop, which can barely reach d5 in time.

This finishes the proof - each player had to take at least 19 moves, which are all uniquely determined after all. (I won't go through this in detail - but many of the loose earlier estimates can be tightened up once we have the whole picture - for instance, we can now conclude that Pg7 must have been captured by Bc1 and not Ra1 so as to achieve the black minimum). Nonetheless, there is still ambiguity in the move order at several places. Maybe I will look at this later ...

To answer the final question: It is now White to move and win with 20. Rg5+. Black can only move 20. ... Kf4 (Kh4 gets directly mated by Nf3#). Following up with 21. g3+ Ke4 (forced), we can play 22. Nf3 and threaten the two mates Re5 and Nd2, which cannot both be blocked. Black can only delay by giving two checks with the rook, after which mate directly follows.

This was quite a tricky puzzle and I had a lot of fun solving it! :)

$\endgroup$
2
  • 1
    $\begingroup$ Wow. I didn't think of that idea. I agree that this is a tricky puzzle, it took me hours and I still get it wrong lol. I think I will have my revenge on @acegikmoqsuwy2000 soon. $\endgroup$ Commented Dec 23, 2024 at 19:51
  • 1
    $\begingroup$ The number of sequences is at least 15, I'm not 100% sure about this though. But now I believe it is really "countable". $\endgroup$ Commented Dec 23, 2024 at 20:32
5
$\begingroup$

My continuation after reaching final position:

1.Rg5 Kf4 (1..Kh4 Nf3#) 2.g3 Ke4 3.Nf3. Nd2 and Re5 threat is unstoppable, white victory.

Below is the retrograde animation.

solution gif

PGN:

1. b4 b5 2. Bb2 Bb7 3. Bxg7 Bd5 4. Bxh8 Bb3 5. Bd4 Bg7 6. a4 Bxa4 7. Bxa7 Bb3 8. Ra3 Bd5 9. Rc3 f5 10. Rc5 Bc3 11. dxc3 Kf7 12. h4 Kg6 13. Rh3 Be4 14. Qd5 Kh5 15. Qxa8 Kg4 16. Kd2 Bd5 17. Kc1 Be4 18. Qxe4+ fxe4 19. Rd3 exd3 20. Kb2 dxc2 21. Ka3 cxb1=R 22. Bxb8 Rxf1 *

  1. Pretty sure the idea is correct now, but I have to add a "free" move in my solution to fit the timing. Error to optimal 0 or 1 move.
  2. The sequence is not unique, but counting possible solutions is still nuts, I won't do it :) But I will give an lower bound of 64 sequences based on my intuition.
$\endgroup$
6
  • 3
    $\begingroup$ Welcome to Puzzling Stack Exchange. Unfortunately your second black move a5->a4 is not monochromatic. $\endgroup$ Commented Dec 23, 2024 at 10:33
  • $\begingroup$ @BenjaminWang Pawns are not pieces. And if monochromatic rule was applied for pawn, the puzzle would become unsolvable since there would be no way to get that black rook to f1. I know this is quite unsatisfying but that's it. $\endgroup$ Commented Dec 23, 2024 at 10:38
  • 1
    $\begingroup$ @BenjaminWang You are right, gimme me a sec, I will reanswer this puzzle now :) $\endgroup$ Commented Dec 23, 2024 at 11:12
  • 1
    $\begingroup$ @QuýNhân Sorry for the confusion--for the purpose of this puzzle, pawns are subjected to the monochromatic rule. I'll update the original post to make this clarification. $\endgroup$ Commented Dec 23, 2024 at 11:25
  • 1
    $\begingroup$ Also, there's a reason I asked to compute the number of possible shortest solutions--there are very few of them, or maybe only one :-) $\endgroup$ Commented Dec 23, 2024 at 11:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.