SVD< ValueType > Class Template Reference

Compute Singular Value Decomposition, Wrapper for super-old Numerical Recipes code. More...

#include <SVD.h>

Public Types
typedef boost::numeric::ublas::matrix < ValueType >	Matrix
typedef boost::numeric::ublas::vector < ValueType >	Vector

Public Member Functions
	SVD (Matrix &matrix)
	Compute Singular Value Decomposition. More...
const Vector &	getSingularValues () const
const Matrix &	getV () const

Protected Member Functions
void	svdcmp (Matrix &a, Vector &w, Matrix &v)

Detailed Description

template<class ValueType>
class SVD< ValueType >

Compute Singular Value Decomposition, Wrapper for super-old Numerical Recipes code.

Member Typedef Documentation

template<class ValueType >

typedef boost::numeric::ublas::matrix<ValueType> SVD< ValueType >::Matrix

template<class ValueType >

typedef boost::numeric::ublas::vector<ValueType> SVD< ValueType >::Vector

Constructor & Destructor Documentation

template<class T >

SVD< T >::SVD

(

Matrix &

matrix

)

Compute Singular Value Decomposition.

Compute the singular value decomposition of a matrix A into A=UDV'.

Parameters

[in,out]

matrix

row-major matrix, gets overwritten by matrix U of decomposition

Note

Input matrix will be destroyed!

References SVD< ValueType >::svdcmp().

Member Function Documentation

template<class ValueType >

const Vector& SVD< ValueType >::getSingularValues ( ) const

inline

Return singular values sorted from largest to smallest.

Returns

singular values sorted from largest to smallest

template<class ValueType >

const Matrix& SVD< ValueType >::getV ( ) const

inline

Return matrix V of decomposition (this is really V and not V transposed).

Returns

matrix V of decomposition

template<class T >

void SVD< T >::svdcmp ( Matrix &  a, Vector &  w, Matrix &  v  )

protected

Given a matrix a[0..m-1][0..n-1], this routine computes its singular value decomposition, A = U · W · V^T. The matrix U replaces a on output.The diagonal matrix of singular values W is output as a vector w[0..n-1]. The matrix V (not the transpose V^T) is output as v[0..n-1][0..n-1].

References FMAX, IMIN, and SIGN.

Referenced by SVD< ValueType >::SVD().