Coding puzzle

Well, more of a decision puzzle in code
current source - https://bit.ly/2qbkfxZ

For a given integer vector, there is a bit of a decision problem of which set of edge-bins to store(?) the vector in, that will be relatively optimal over a list of such vectors.

Ex. 0xff becomes the vector (16, 16), and could be stored as edgebin[x][16, 16] = true for some choice of x.

Well, more of a deciphering puzzle in English

simply edgebin /g/ro!! lel

>>3
Don't mess with me boy, I'm with the Hells Angles math club and I'll morph your anal manifold into a nonorientable surface.

>>2
0xff (16,16) is a bit trivial of an example
Ex2. 0xff could also become the vector (4, 4, 4, 4), and then could be stored as edgebin[x][1][4, 4] = true, edgebin[x][2][4, 4] = true, edgebin[x][3][4, 4] = true, again for some choice of x if x>1

More generally, it would use any fixed-size discrete-valued vector, eg binned floats [20 bins, 30 bins, 40 bins] -> edgebin[x][1](20x30), edgebin[x][2](30x40)

I have left part of the question quite open, but one goal would be accurate retrieval from a compact representation (so compression w/ not too lossy decompression) if possible

If you want us to read the code use a language known by more than 12 people.

>>7
So at the core of the program currently are these boolean matrixes that capture the edges between nodes in a data vector

[data_id] | [column_a] | [column_b] | [column_c]
1 | category 5 | category 2 | category 1 - (data vec with edges a5-b2 and b2-c1, column a node cat 5, column b node cat 2, col c cat 1)

{x=1
[boolean matrix m * n (col a:b)]
[5,2] = true;

[boolean matrix n * m2 (col b:c)]
[2,1] = true;
}

for retrieval it still probably needs the first entry, column a cat 5, plus some data for the x>1 / alternates

>>8
Is that Haskell?

>>1
This can be extended in more dimensions. Imagine how many vectors you can stuff in 5D hypercubes.

>>10
Imagine how many vectors you can stuff in my hyperanus. dubs!!!!!!!

>>9
I don't think so, it was supposed to just be pseudocode

>>10
That is true, but if each dimension is 1000 units wide, you could have a chain of 1,000,000,000 2d 1000x1000 for the same memory cost as 1x 5d hypercube of the same width. Note it only needs to be >1D to be able to start chaining these boolean tables together. Higher dim could still be good on short widths though