Re: [eigen] [PATCH] Toeplitz matrix specialization

[ Thread Index | Date Index | More Archives ]

On Thu, 29 Jan 2009, Gael Guennebaud wrote:

to start the debate I put some thougths on the wiki:

Does "we assume we don't need a special class for triangular storage" mean that triangular matrices are stored as general matrices (and thus half the memory is wasted)? That strikes me as an assumption that could be regretted later. So perhaps we should make sure that a special class for triangular storage can be added later.

One other usage scenario, of which I'm wondering how difficult it would be to implement in the current setting, is in-place LU decomposition (that is, replace the matrix A by its L and U factors, so that you don't need to allocate extra memory except for the permutation vector).


Mail converted by MHonArc 2.6.19+