I was trying to solve the problem of construction $N$ unit cubes, and while searching I came across this sequence at OEIS.

This is exactly what I need but I could not find a method to generate the series. How do you think this can be derived?


I'm not sure it is generated by a series, because such would suppose that 10 cubes would naturally follow off nine, which follows off 8.

For example, eight cubes is 2*2*2, which leads 18 lines in each direction (or 54). Adding an extra cube to the side of this would make 62. But a 3*3*1 construction gives 16 vertical, and 8*3*2 horizontal, ie 48 (total 64). So 2*2*2 + 1*1*1 is more efficient here.

For 10, one would add an extra cube adjacent to the two already there, giving 67 (four horizontal, and 2 vertical). Likewise, 11 puts an extra four sides of the cube, and adds 5 extra edges (72), For 12, one has 2*2*3 to give 75.

So it looks like they're keeping close to the cube, adding either 8, 5, or 3 as there are either 1, 2 or 3 already completed sides. Obviously the first cube has 12 edges, since there are none new.

I tend to think it's an experimental result. It seems though the most efficient way is to make first a cubeoid, of the product of $m$ and $m+1$. The residue makes a polygon less than $m+1$, which is made into $n, n+1$, and the remainder is set to one side.

We can divide the series 12, 8, 5, 3 into "4, 3, 2, 3", where every cube requires at least 3 edges. It is then how many edge cubes are needed.

The 4 is used exactly once: it makes a square face which we extend into a cube.

The 3 is used on each side of the cube, where $m$ is the largest cube not larger than N, we have $3m + x$, where x=0 if $N=m^3$, x=1 if $N=m^3+m^2$, and x=2 if $N=m(m+1)^2$. If it lies between these ranges, add 1.

The 2 is used to form the 'faces' of the large cube. It corresponds to $(m)(m+1)$ for each complete cuboid face, and $2n+1, 2$ according to the remaining over the cuboid.

The remainder 3 is applied to all cubes, ie $3N$.

So you end up with $4+3m^2+2n+3N+x$, where $x is the residue part of the broken fraction.

  • 1
    $\begingroup$ I am sure it is not experimental because it is a question on link-eolimp and if you think 2D version of this problem (constructing squares) the solution would be $2N + ceil(2*sqrt(N))$ so it might be constructed from that observation but i could not $\endgroup$ – kadircet Dec 10 '13 at 12:12
  • $\begingroup$ When you come to add a cube to a structure, there can be one, two or three faces already present, meaning that 8, 5 or 3 new edges are needed. It seems that the most efficient way of making such a thing is to keep the assembly as cubic as possible, and try to minimise the 8 and 5 cubes. For example, for 31, one can make it as 27 (ie a product of numbers p and p+1), then the difference is made as a product of q^2, or q(q+1), and the rest added to the end. $\endgroup$ – wendy.krieger Dec 11 '13 at 7:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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