Re^2: decomposing binary matrices

[hv]

I think that approach has possibilities, though it might need some modification - there is no guarantee that the row with fewest bits is separable. Consider, say:

A:  1 1 0 0 0
B:  1 0 1 1 0
C:  1 0 1 1 0
D:  1 0 1 1 0
E:  1 1 1 1 1

.. from which {B, C, D} is a separable submatrix. I'd have to think further whether I can easily construct an example that defeats the approach on both rows and columns simultaneously - it'd certainly need a larger matrix (at least 7x7 I think), but I think the same general idea could throw up a proper counterexample.

At the risk of wasting your time ...

Not a waste by any means - many of the suggestions so far could maybe be adaptable to a full solution. It's just a question of whether the adapted version would end up any quicker than going brute force in the first place. :)

Hugo