I am relatively new to Eigen and I am trying to perform an operation on sparse matrix as follows:

Code: Select all

Eigen::SparseMatrix<float> priors;
// Some code follows.
// Now, I want to do the following

prior = (prior.transpose() * prior).pruned();


Is this the right way to do this? I am hoping that there is no dense matrix being computed along the way and that it is reasonable efficient?

Thanks for your help.

Luca

yes that's the right way. .pruned() only remove explicit zeros that may appear during the computation. The implicit zeros are never created!

Hi

I'm also new to Eigen and used sparse matrix multiplication with ".pruned()".
However, the result is dense! I checked and zero values are not pruned.
I have no idea in which part, I wrote wrong.

Mehdi

How did you look at the matrix entries ? What about mat.nonZeros() ??