Re: [eigen] In-place decompositions; was: UmfPackSupport vs SuperLUSupport: input matrix

[ Thread Index | Date Index | More Archives ]

On 16.04.2013 17:50, Gael Guennebaud wrote:
On Tue, Apr 16, 2013 at 3:49 PM, Christoph Hertzberg
<chtz@xxxxxxxxxxxxxxxxxxxxxxxx> wrote:
A fixed-size use case I would have is to have a lot of rather small fixed
size matrices which are accumulated simultaneously and shall all be
decomposed, once the accumulation is done.

damn, I was not expecting use cases for small fixed size matrices, but
yours is perfectly reasonable. Then there are two approaches, either
the factorization uses the memory shared by the matrix A, or we create

I just filed a feature request for this:

API-wise this could be accomplished by declaring the decomposition as
Decomposition<Ref<MatrixXd> >.
On calling compute(mat), it would not copy the matrix, but simply re-initialize the internal reference. I assume the rest of the code can stay as it is -- unless someone also wants to store the additional data of the decomposition (like permutation matrices) into already allocated memory.


Dipl.-Inf., Dipl.-Math. Christoph Hertzberg
Cartesium 0.049
Universität Bremen
Enrique-Schmidt-Straße 5
28359 Bremen

Tel: +49 (421) 218-64252

Mail converted by MHonArc 2.6.19+