Registered Member
|
Hi all,
The released version of 3.2 (as supplied via the download link on the Eigen home page) works fine for dense QR factorization, but has an assertion failure for certain inputs in the sparse case (see https://forum.kde.org/viewtopic.php?f=74&t=117474 for a similar issue). The fixed versions in the "default" and "3.2" branches on Bitbucket work for the sparse case but then fail on dense matrices. I assume the fix for one has broken the other, but am not sure. Can someone have a look?
|
Moderator
|
The dense and sparse QR classes are not related at all. The problem with dense full piv QR is fixed though:
https://bitbucket.org/eigen/eigen/commits/5d557a6213f4/ Changeset: 5d557a6213f4 User: ggael Date: 2013-11-19 12:53:46 Summary: Fix FullPivHouseholderQR ctors for non squared fixed size matrix types https://bitbucket.org/eigen/eigen/commits/6518cd75818e/ Changeset: 6518cd75818e Branch: 3.2 User: ggael Date: 2013-11-19 12:53:46 Summary: Fix FullPivHouseholderQR ctors for non squared fixed size matrix types (grafted from 5d557a6213f4a05e9bd06904edb5fa5fab519b80) |
Registered users: Bing [Bot], Google [Bot], Yahoo [Bot]