8
$\begingroup$

I got from my children as a gift a puzzle consisting of 25 y-shaped 5-polyominoes that form a $5 \times 5 \times 5$-cube (see picture).

I'm wondering for which $n$ does a y-formed $n$-polyomino tile a $n \times n \times n$-cube?

By a y-formed $n$-polyomino I understand a generalisation of the y-formed $5$-polyomino (made of the $5$ squares $(0, 0), (1, 0), (2, 0), (3, 0), (1, 1)$), i.e., the center of the $n$ squares can be chosen as $(0,0), (1,0), (1,1), (2,0), \dotsc, (n-2,0)$. The polyomino has a thickness of 1.

For $n=5$ I have the (wooden) proof that it's possible before me, for $n=4$ it's trivial (as one can tile one $4 \times 4 \times 1$-slice of the $4 \times 4 \times 4$-cube already with four y-formed $4$-polyominoes), what about $n=6, 7 , \dotsc$?

Any ideas for patterns that work for whole sequences of $n$, any proofs that for some $n$ it's not possible?

 A 5×5×5-cube tiled by 25 y-shaped 5-polyominoes

$\endgroup$
7
  • 3
    $\begingroup$ Using Don Knuth's implementation of Algorithm X, I've verified that there are no solutions for $n=6$ and $n=7$. Also, Michael Reid has a webpage with many ways of packing various polycubes into boxes. $\endgroup$ Jan 24, 2023 at 21:52
  • $\begingroup$ I did not know what a y-shaped polyomino was except from the picture, so I have edited in a description in words, which is hopefully correct. I am confused by the reference to the centre of the squares. Do you mean to say that the generalised polyomino has the squares $(0, 0), (1, 0), \dotsc, (n - 2, 0), (i, 1)$ for some integer $i$ with $0 \le i < n - 1$? $\endgroup$
    – LSpice
    Jan 26, 2023 at 17:52
  • 2
    $\begingroup$ My interpretation was: a y-formed $n$-polyomino is given by the $n-1$ cells $(0,0)$, $(1,0)$, …, $(n-2,0)$, together with the cell $(1,1)$. $\endgroup$ Jan 26, 2023 at 18:41
  • 1
    $\begingroup$ I certainly have no objection to that formulation, if it is what was intended! However, then I didn't know how to make sense of the terminology "the center of the squares can be chosen as $(0, 0), (1, 0), \dotsc, (n - 2, 0)$"; what is the choice that is being made? (That is, what is the ‘centre’?) Anyway, all I wanted is to make sure I understood the statement. $\endgroup$
    – LSpice
    Jan 26, 2023 at 18:53
  • 1
    $\begingroup$ Many thanks for improving my formulation, and sorry that my original one was not crystal clear. $\endgroup$ Jan 27, 2023 at 15:48

1 Answer 1

4
$\begingroup$

There are no solutions for $n\ge6$. For $n=6$ one checks that the exact cover formulation of the problem has no solution, as ho boon suan did in the comment.

For $n\ge7$, the following argument works: Let $(i,j,k)$ for $1\le i,j,k\le n$ be the $n^3$ positions of the cube to be covered. Consider the subset $S$ of positions $(i,j,k)$ where $i,j,k\in\{1, 4, n\}$. Each polyomino intersects $S$ in either $0$ or $2$ positions. As $\lvert S\rvert=27$ is odd, there is no cover.

There are also choices of $S$ which are symmetric under the symmetries of the cube:

If $n$ is odd, then for $S=\{(i,j,k)\;|\;i,j,k\in\{1,(n+1)/2,n\}\}$ we get $\lvert S\rvert=27$, while $\lvert S\cap P\rvert=0$ or $2$ for each polyomino.

If $n$ is even, then for $S=\{(i,j,k)\;|\;i,j,k\in\{1,n/2,n/2+1,n\}\}$ we get $\lvert S\rvert=64$, while $\lvert S\cap P\rvert=0$ or $3$ for each polyomino.

$\endgroup$
4
  • $\begingroup$ This makes rough sense to me, but I think I'm having a little bit of trouble visualizing: what in the argument needs $n\geq7$? $\endgroup$ Jan 26, 2023 at 19:04
  • 1
    $\begingroup$ Thinking of the y-polyomino as a straight line of $n-1$ cubes, say along the $x$-axis, together with an extra protruding cube, that cube can only have $x$-coordinates in $\{2,3,n-2,n-1\}$. $\endgroup$ Jan 26, 2023 at 19:10
  • 1
    $\begingroup$ @StevenStadnicki To complement ho boon suan's comment: If $n=6$ then $n−2=4$, so it can happen that the polyomino hits S (in its "extra protruding cube") in only one point. $\endgroup$ Jan 26, 2023 at 21:42
  • $\begingroup$ @PeterMueller Thank you! That's exactly the bit that I was missing. $\endgroup$ Jan 27, 2023 at 18:12

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.