I was wondering if there is a routine for the multiplication of a sparse symmetric matrix with a dense matrix. There is
however it accepts the general CSR format not the symmetric one. I guess this is not directly possible, however there is a routine which does it for a sparse symmetric matrix and a dense vector, so I could make use of this to generate the columns of the resulting dense matrix in this way. I was wondering what could be the best way to accomplish this task. Since my matrices for the application I work on are symmetric, I was wondering if there are some clever tricks.
Thanks for the advice and best regards, Umut
Edit: Reading a bit more on the documentation, on the matdescra character array one can specify Symmetric option, does that mean that it extracts the symmetric triplets inside that, I am a bit confused.
Like dense BLAS symmetric matrix-matrix multiply, mkl_dcsrmmis able to extract internallythesymmetric triplets but it alsoworks if there is only CSR representation of upper or lower triangle of the matrix.