For 1 x n, the determinants were:
1x1: 0
1x2: -1
1x3: 0
1x4: 1
1x5: 0
1x6: -1
The pattern: for n equiv 0 (mod 2), the determinant is either -1 or 1
For 2 x n, the determinants were:
2x2: -3
2x3: 7
2x4: 5
2x5: -33
And for 3 x n, determinants:
3x3: 0
3x4: 9
3x5: 0
I'm not sure what it is for mxn yet.
But I think that this is the rule for solvability: it seems that for 1xm, if m is even it's solvable. In general, it seems that if one or both dimensions are even, then the puzzle is solvable.
I'll work on the nx1 case next time.

The image at the top is hard to interpret, taken out of context of the spreadsheet it was in -- you might want to clarify that it specifies *square* cells, arranged in *rectangular* grid, where the neighborhood included cells adjacent by both edges and vertices (i.e., the 8 surrounding squares) but did NOT include the center cell itself.
ReplyDelete