topic Your question should have in Intel® oneAPI Math Kernel Library

Sparse Matrix storage to compressed sparse row format

Ma_Khe_ — Mon, 03 Mar 2014 07:10:22 GMT

Hi :-)

I am new here and I have a question about using "PARDISO - Parallel Direct Sparse Solver Interface". As you may know in order to use this subroutine for solving a set sparse linear equations as AX = b, the A matrix should be stored in comperes sparse row format. My question is there any subroutine which can do this storage? if there is I think it would be really nice to include them here "http://software.intel.com/sites/products/documentation/hpc/mkl/mklman/GUID-431916D5-B76D-48A1-ABB5-1A0613FDC0FA.htm#GUID-431916D5-B76D-48A1-ABB5-1A0613FDC0FA"

Kind regards,

Your question should have

mecej4 — Mon, 03 Mar 2014 12:06:00 GMT

Your question should have been posted in the MKL forum.

There is a routine mkl_ddnscsr() (for double precision real matrices; similarly named routines exist for other types) in MKL for exactly this purpose. The MKL documentation should be consulted about matrix storage schemes and conversion. A recent post that involves such conversion is "How can I remove a column from a sparse matrix (CSR) using MKL"

in that case, we also

Gennady_F_Intel — Mon, 03 Mar 2014 16:28:31 GMT

in that case, we also recommend to look at the existing examples show how to use such routines -- all these examples you can find into <mklroot>\examples\spblasc\source\ directory as an example