New user self-registration is disabled due to spam. Please email eigen-core-team @ lists.tuxfamily.org if you need an account.
Before reporting a bug, please make sure that your Eigen version is up-to-date!
Bug 1121 - Using SPQR as SparseQR for nullspace computation
Summary: Using SPQR as SparseQR for nullspace computation
Status: NEW
Alias: None
Product: Eigen
Classification: Unclassified
Component: Sparse (show other bugs)
Version: 3.2
Hardware: x86 - 64-bit Windows
: Normal Feature Request
Assignee: Nobody
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-11-27 21:45 UTC by Semih
Modified: 2015-11-28 20:00 UTC (History)
2 users (show)



Attachments

Description Semih 2015-11-27 21:45:40 UTC
Hi, I need to compute nullspaces of psd sparse matrices. However , I am not happy with the performance of the SParseQR and planned to try SPQR. However matrixQ of SPQR is returning a different type of variable than SparseQR. You may see the compilation error below. I need convert SPQRMatrixQReturnType to plain sparsematrix. Are they any way or any workaround or planned patch for this problem.


error C2679: binary '=' : no operator found which takes a right-hand operand of type 'Eigen::SPQRMatrixQReturnType<Eigen::SPQR<Eigen::SparseMatrix<double,0,int>>>' (or there is no acceptable conversion)
Comment 1 Gael Guennebaud 2015-11-28 20:00:27 UTC
Patch welcome! SPQR stores its factors in its own format, so first thing to do is to check SPQR documentation to see whether it offers routines to extract Q, and if so, then follow SparseQR implementation. Otherwise, maybe you don't need to explicitly extract Q, maybe all you need is to apply Q or Q^T?

Also, feel free to contribute your null space extraction routine as a SparseQR::kernel() method ;)

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