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

mkl_sparse_?_svd implementation

Robert_E_
Beginner
464 Views

Hello,

is mkl_sparse_?_svd yet implemented and supposed to work?

source:

https://software.intel.com/en-us/onemkl-developer-reference-fortran-mkl-sparse-svd

Because after trying several things I get error 6, SPARSE_STATUS_NOT_SUPPORTED every time.
I'll give you a MWE when it's implemented.

0 Kudos
1 Solution
Gennady_F_Intel
Moderator
465 Views

hi Robert,

Yes, it should work. You could check the latest version v.2020 or give us the reproducer to check the problem on our side.

View solution in original post

4 Replies
Gennady_F_Intel
Moderator
466 Views

hi Robert,

Yes, it should work. You could check the latest version v.2020 or give us the reproducer to check the problem on our side.

Robert_E_
Beginner
465 Views

Hi,

thank you!

I tried more things and it works. The problem was that my matrix was in CSC format. The documentation does not state that only CSR format is supported, or maybe I'm blind.

If you want sugesstions on improving the documentation, read on, else I'm done here. Thanks for the great work and support!

errors in documentation:

The interface should be

stat = mkl_sparse_d_svd (whichS, whichV, pm, A, descrA, k0, k, E, XL, XR, res);

instead of

stat = mkl_sparse_d_svd (whichS, whichV, pm, A, descrA, k0, k, E, X, res);

I think the term "eigenvalues" should be replaced with the term "singular values" everywhere.

Gennady_F_Intel
Moderator
465 Views

Thanks, Robert for these issues. We will fix all of these topics. 

Gennady_F_Intel
Moderator
462 Views

The error into documentation has been fixed into version 2020 u1 which already released and available. 

Reply