Bug 596 - SparseQR matrixQ() to plain SparseMatrix
SparseQR matrixQ() to plain SparseMatrix
Status: NEW
Product: Eigen
Classification: Unclassified
Component: Sparse
3.2
All All
: Normal Unknown
Assigned To: Desire NUENTSA
:
Depends on:
Blocks: 3.3
  Show dependency treegraph
 
Reported: 2013-05-07 15:35 UTC by Gael Guennebaud
Modified: 2013-10-18 15:41 UTC (History)
2 users (show)



Attachments

Description Gael Guennebaud 2013-05-07 15:35:00 UTC
SparseQRMatrixQReturnType should inherit EigenBase and provide an evalTo member to be converted to a plain SparseMatrix.

Also, investigate what need to be done to be able to exploit more features of the Householder module.
Comment 1 Desire NUENTSA 2013-05-22 10:50:51 UTC
An evaluation to a plain sparse matrix is provided with this changeset
https://bitbucket.org/eigen/eigen/commits/2403f1f7b8a7/
Changeset:   2403f1f7b8a7
User:        dnuentsa
Date:        2013-05-21 17:35:10
Summary:     Fix bug 596 : Recover plain SparseMatrix from SparseQR matrixQ()
Affected #:  5 files
Comment 2 Gael Guennebaud 2013-06-10 13:09:00 UTC
Let's differ the second part to 3.3:

> Also, investigate what need to be done to be able to exploit more features of
> the Householder module.
Comment 3 Gael Guennebaud 2013-10-02 22:34:14 UTC
matrixQ() currently relies on SparseQR_QProduct which is implemented with dense RHS in mind. This makes matrixQ() quite inefficient. The solution is to extend SparseQR_QProduct to handle sparse RHS using the same approach than during the factorization. There is a nice opportunity for code sharing here.
Comment 4 Gael Guennebaud 2013-10-02 22:36:07 UTC
see also: http://forum.kde.org/viewtopic.php?f=74&t=117500&p=293605#p292860

Note You need to log in before you can comment on or make changes to this bug.