SVDKSVD#

SVDKSVD = “ksvd” - A wrapper to the KSVD singular value solver [Sukkari et al., 2019].

Notes#

Only available for standard SVD problems.

This is a direct singular value solver, that is, the full decomposition is computed. The computation involves redistributing the matrices from PETSc storage to ScaLAPACK distribution, and vice versa (this is done automatically by SLEPc). Alternatively, the user may create the problem matrices already with type MATSCALAPACK.

The implemented method is supposed to be more scalable than the one in SVDSCALAPACK.

References#

[Suk19]

D. Sukkari, H. Ltaief, A. Esposito, and D. Keyes. A QDWH-based SVD software framework on distributed-memory manycore systems. ACM Trans. Math. Software, 2019. doi:10.1145/3309548.

See Also#

SVD: Singular Value Decomposition, SVD, SVDType, SVDSetType(), SVDSCALAPACK

Level#

beginner

Location#

src/svd/impls/external/ksvd/svdksvd.c


Index of all SVD routines Table of Contents for all manual pages Index of all manual pages