Intel® oneAPI Math Kernel Library
Ask questions and share information with other developers who use Intel® Math Kernel Library.

Pardiso matrix size change over the simulation

Tabrizian__Amin
Beginner
354 Views

Dear all, 

I have just started learning PARDISO and I face a question:

How can we use PARDISO when our coefficient matrix size changes over the simulation?

I know that we can simply solve this issue by reinitializing PARDISO but maybe a special order of steps needs to be followed to avoid this inefficient using of PARDISO.

Thanks.

0 Kudos
1 Solution
Kirill_V_Intel
Employee
354 Views

Hi Amin,

You need to formulate on the level of mathematics how you would like to re-use the matrices. PARDISO is a way to factor the matrix and then solve the system. For example you can use one of the matrices as a preconditioner. Or you can use the Schur complement technique.

Best,
Kirill

View solution in original post

0 Kudos
3 Replies
Kirill_V_Intel
Employee
354 Views

Hi!

First of all, please clarify what do you mean by changing the coefficient matrix size? And how do you want to re-use the original matrix on the high-level? For example do you have an extended system where the original matrix is a sub-block (maybe with slightly changed matrix values but with the same pattern)? Or do you have a completely different new larger matrix (for example when you change the discretization step when solving numerically some PDEs)?

Best,
Kirill

0 Kudos
Tabrizian__Amin
Beginner
354 Views

Hi Kirill,

Thank you for  your reply. I mean how we can use PARDISO when size of our system of equations changes over the simulation? And about re-using the original matrix,  our original matrix may change by adding some rows(columns) or omitting some.

Thanks,

Amin

0 Kudos
Kirill_V_Intel
Employee
355 Views

Hi Amin,

You need to formulate on the level of mathematics how you would like to re-use the matrices. PARDISO is a way to factor the matrix and then solve the system. For example you can use one of the matrices as a preconditioner. Or you can use the Schur complement technique.

Best,
Kirill

0 Kudos
Reply