Ask questions and share information with other developers who use Intel® Math Kernel Library.
Announcements
FPGA community forums and blogs on community.intel.com are migrating to the new Altera Community and are read-only. For urgent support needs during this transition, please visit the FPGA Design Resources page or contact an Altera Authorized Distributor.
Success! Subscription added.
Success! Subscription removed.
Sorry, you must verify to complete this action. Please click the verification link in your email. You may re-send via your
profile.
Does Intel have any plan to include iterative methods for the sparse least square problem such as LSQR, SYMMLQ, and MINRES into MKL? (or PARADISO has already included these functions?)
MKL doesn't have any plan to include iterative methods for the sparse least square problems. PARDISO doesn't have such functioanality as well.
But you could try the following way. It can work if your matrix has full rank.It is known thatthe least square problem for A*x = fis equivalent to finding the solution of the system(A^T * A) x = A^T *f. MKL Sparse BLAS provides routines for computing product of two sparse matrices stored in the compressed sparse row format. The result matrix can be stored as a dense matrix orcompressed sparse row matrix. The vector A^T*f can be computed with the help of MKL sparse BLAS as well. After that PARDISO or LAPACK solver can be used for finding the solution(A^T * A) x = A^T *f.
But you could try the following way. It can work if your matrix has full rank.It is known thatthe least square problem for A*x = fis equivalent to finding the solution of the system(A^T * A) x = A^T *f. MKL Sparse BLAS provides routines for computing product of two sparse matrices stored in the compressed sparse row format. The result matrix can be stored as a dense matrix orcompressed sparse row matrix. The vector A^T*f can be computed with the help of MKL sparse BLAS as well. After that PARDISO or LAPACK solver can be used for finding the solution(A^T * A) x = A^T *f.
Thank you very much for your reply. Unfortunately, my problem doesn't have full rank. So I'm currently using open-source version of MINRES.
