Skip to content

ZGELSD #99

Open
Open
@oamarques

Description

@oamarques

ZGELSD returns INFO>0 for a matrix of dimension 23068-by-23066, resulting from DLASD4 (root finder) exceeding its maximum number of iterations. The problem is not in DLASD4 per se, but in the merging/sorting/deflation of singular values of (two) submatrices. The subroutine that does that, DLASD7, is failing to do the sorting that is later required by DLASD4. The issue in this case is that the 23068-by-23066 matrix has a very large number (1000’s) of singular values extremely close to each other and the merging/sorting/deflation step is not working as intended. The problem is under investigation; it might be related to the issue #34 created by Jack Poulson on Aug 7, 2016.

Obs. ZGELSD computes the minimum-norm solution to a real linear least squares problem
minimize 2-norm(| b - A*x |)
using the singular value decomposition (SVD), which is computed by a divide-and-conquer algorithm.

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions