(where the boxed nodes in the diagrams represent the box minirows for Boxes1,2 and the 'arc labels' represent the dependence/distribuition of cell value counts between the boxes)PatmaxDaddy wrote:For my 4xC and 5xC counts, I find a set of equivalence classes (gangsters) for box 2 where box 2 permutations are in the same class if the "structure" of the mapping from box 1 rows to box 2 rows is the same. This structure is represented by a directed, labelled graph (DLG) containing R nodes.

Here are the number of box 2 gangsters for various cases:

- Code: Select all
`4x4: 28`

4x5: 53

5x3: 89

5x4: 618

5x5: 3087

Has anyone explored using symbolic algebra programs (e.g. Mathematica (combinatorics pkg) / Maple) for computing (or developing) the band permutations counts?