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

adjust dcsrilu0 parameters in intel MKL.

sellountos__euripide
1,107 Views

Hello,

 

In intel MKL dfgmres with ILUT preconditioner, i.e. dcsrilu0 is there any way to adjust the parameters : initial fill (lfil0) and initial drop (tol0)? I wanted to know if there's any way to adjust the preconditioner's dropping tolerances in order to converge with less iterations.

 

According to the documentation
https://www.intel.com/content/www/us/en/docs/onemkl/developer-reference-c/2024-0/dcsrilu0.html#GUID-E40E25D7-80D0-45FE-919C-F7DD969655E4

there's only option to adjust the diagonal

 

Thanks in advance.

 

0 Kudos
1 Solution
sellountos__euripide
944 Views

Hello,

 

Thanks for the answer. It resolved my problem.

Regards,

E.

View solution in original post

0 Kudos
3 Replies
Mahan
Moderator
1,021 Views

Hi,

I would recommend to use dcsrilut instead of dcsrilu0 as it allows you to do ILU with a threshold. It has better performance but poorer sparsity.

Please follow the link for more details:

https://www.intel.com/content/www/us/en/docs/onemkl/developer-reference-c/2024-0/dcsrilut.html


Please let me know does this answer your query.


0 Kudos
Mahan
Moderator
989 Views

Please let me know if the above resolution answer your question.


0 Kudos
sellountos__euripide
945 Views

Hello,

 

Thanks for the answer. It resolved my problem.

Regards,

E.

0 Kudos
Reply