GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-06-04 18:54:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

set ColPivHouseholderQR as default preconditioner for JacobiSVD

Browse Source
This commit is contained in:
Benoit Jacob 2010-10-11 21:00:42 -04:00
parent 5c3d21693b
commit b8bb804007
3 changed files with 10 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Eigen/src/Core/util/ForwardDeclarations.h
View File

@ -170,7 +170,7 @@ template<typename MatrixType> class HouseholderQR;
template<typename MatrixType> class ColPivHouseholderQR;
template<typename MatrixType> class FullPivHouseholderQR;
template<typename MatrixType> class SVD;
template<typename MatrixType, int QRPreconditioner = FullPivHouseholderQRPreconditioner> class JacobiSVD;
template<typename MatrixType, int QRPreconditioner = ColPivHouseholderQRPreconditioner> class JacobiSVD;
template<typename MatrixType, int UpLo = Lower> class LLT;
template<typename MatrixType, int UpLo = Lower> class LDLT;
template<typename VectorsType, typename CoeffsType, int Side=OnTheLeft> class HouseholderSequence;

16
Eigen/src/SVD/JacobiSVD.h
View File

@ -205,17 +205,17 @@ struct ei_qr_preconditioner_impl<MatrixType, HouseholderQRPreconditioner, Precon
* for the R-SVD step for non-square matrices. See discussion of possible values below.
*
* The possible values for QRPreconditioner are:
* \li FullPivHouseholderQRPreconditioner (the default), is the safest and slowest. It uses full-pivoting QR.
* We make it the default so that JacobiSVD is guaranteed to be entirely, uncompromisingly safe by default.
* \li ColPivHouseholderQRPreconditioner is the default. In practice it's very safe. It uses column-pivoting QR.
* \li FullPivHouseholderQRPreconditioner, is the safest and slowest. It uses full-pivoting QR.
* Contrary to other QRs, it doesn't allow computing thin unitaries.
* \li ColPivHouseholderQRPreconditioner is faster, and in practice still very safe, although theoretically not as safe as the default
* full-pivoting preconditioner. It uses column-pivoting QR.
* \li HouseholderQRPreconditioner is even faster, and less safe and accurate than the pivoting variants. It uses non-pivoting QR.
* \li HouseholderQRPreconditioner is the fastest, and less safe and accurate than the pivoting variants. It uses non-pivoting QR.
* This is very similar in safety and accuracy to the bidiagonalization process used by bidiagonalizing SVD algorithms (since bidiagonalization
* is inherently non-pivoting).
* \li NoQRPreconditioner allows to not use a QR preconditioner at all. This is useful if you know that you will only be computing
* is inherently non-pivoting). However the resulting SVD is still more reliable than bidiagonalizing SVDs because the Jacobi-based iterarive
* process is more reliable than the optimized bidiagonal SVD iterations.
* \li NoQRPreconditioner allows not to use a QR preconditioner at all. This is useful if you know that you will only be computing
* JacobiSVD decompositions of square matrices. Non-square matrices require a QR preconditioner. Using this option will result in
* faster compilation and smaller executable code.
* faster compilation and smaller executable code. It won't significantly speed up computation, since JacobiSVD is always checking
* if QR preconditioning is needed before applying it anyway.
*
* \sa MatrixBase::jacobiSvd()
*/

2
test/jacobisvd.cpp
View File

@ -159,7 +159,7 @@ template<typename MatrixType> void jacobisvd_verify_assert(const MatrixType& m)
RhsType rhs(rows);
JacobiSVD<MatrixType, HouseholderQRPreconditioner> svd;
JacobiSVD<MatrixType> svd;
VERIFY_RAISES_ASSERT(svd.matrixU())
VERIFY_RAISES_ASSERT(svd.singularValues())
VERIFY_RAISES_ASSERT(svd.matrixV())