functions for the conversion of matrix storage schemes

[woshiwuxin]

Hi, everyone!

There're three types of matrix storage schemes in Intel MKL. But is there any function for the conversion between them?

Thank you in advance!

[ArturGuzik]

Hi,

take a look at these (if you're referring to sparse storage schemes).

Matrix converters
mkl_?dnscsr	Converts a sparse matrix in the dense representation to the CSR format (3-array variation).
mkl_?csrcoo	Converts a sparse matrix in the CSR format (3-array variation) to the coordinate format and vice versa.
mkl_?csrbsr	Converts a sparse matrix in the CSR format to the BSR format (3-array variations) and vice versa.
mkl_?csrcsc	Converts a sparse matrix in the CSR format to the CSC and vice versa (3-array variations).
mkl_?csrdia	Converts a sparse matrix in the CSR format (3-array variation) to the diagonal format and vice versa.
mkl_?csrsky	Converts a sparse matrix in the CSR format (3-array variation) to the sky line format and vice versa.

A.

[woshiwuxin]

Hi, Artur!

I wasn't not refering to the sparse storage. Anyway, I can write the code on my own. But that can be less optimized compared with the library functions from professional programers.

[ArturGuzik]

Hi,

if you didin't refer to sparse storage, MKl uses these:

• conventional full storage (in a two-dimensional array)

• packed storage for Hermitian, symmetric, or triangular matrices (in a one-dimensional array)

• band storage for band matrices (in a two-dimensional array)

• rectangular full packed storage for symmetric, Hermitian, or triangular matrices as compact as the Packed storage while maintaining efficiency by using Level 3 BLAS/LAPACK kernels.

Also from docs:

Intel MKL provides a number of routines such as ?hfrk, ?sfrk performing BLAS operations working directly on RFP matrices, as well as some conversion routines, for instance, ?tpttf goes from the standard packed format to RFP and ?trttf goes from the full format to RFP.

Please refer to the Netlib site for more information (this is actually very good place to start if plan any sort of development).

A.

[woshiwuxin]

Thanks, Artur! I will take a look.