topic Does this mean that it is in Intel® oneAPI Math Kernel Library

modifying the pardiso solver's sparse factorization

Jaron_K_ — Tue, 02 Feb 2016 17:41:28 GMT

I am attempting to implement this algorithm for efficiently computing the solution to a sparse linear system of equations when small changes are made to the input matrix A by updating only the matrix L in the LDL^T factorization of A. However, from what I have seen there is no documentation about accessing L and D in the pardiso solver. Ideally, my code would look roughly like this:

1. run solver on matrix A with phase 13

2. use solution to compute modification to matrix A

3. use the linked algorithm to directly modify L and D, still in pardiso's memory

4. run solve on matrix A with phase 33, which should use the modified L and D to instead solve the system for the modified matrix A

5. repeat steps 2 through 4

How would I go about doing this? Is it even possible with pardiso?

You may get the diagonal

Gennady_F_Intel — Tue, 02 Feb 2016 19:37:10 GMT

You may get the diagonal elements of initial or factorized matrix by using pardiso_getdiag() routine and with regard to L: you couldn't obtained and the update L part of factorized matrix.

Does this mean that it is

Ivan_K_6 — Wed, 02 Nov 2016 19:16:09 GMT

Does this mean that it is impossible to get L and D factors "exported" as a matrix?

yes. that's not possible

Gennady_F_Intel — Thu, 03 Nov 2016 03:33:52 GMT

yes. that's not possible

Hi,

Alexander_K_Intel2 — Thu, 03 Nov 2016 03:50:44 GMT

Hi,

Probably you can try this approach if you need to solve set of system with small changes in matrices.

Thanks,

Alex