vnl_levenberg_marquardt Class Reference

#include <vnl_levenberg_marquardt.h>

Inheritance diagram for vnl_levenberg_marquardt:

[legend]Collaboration diagram for vnl_levenberg_marquardt:

[legend]List of all members.

Public Methods
	vnl_levenberg_marquardt (vnl_least_squares_function &f)
	vnl_levenberg_marquardt (vnl_least_squares_function &f, vnl_vector< double > &x)
	~vnl_levenberg_marquardt ()
bool	minimize_without_gradient (vnl_vector< double > &x)
bool	minimize_using_gradient (vnl_vector< double > &x)
bool	minimize (vnl_vector< double > &x)
void	diagnose_outcome () const
void	diagnose_outcome (vcl_ostream &) const
vnl_matrix< double > const &	get_JtJ ()
Protected Methods
void	init (vnl_least_squares_function *f)
Static Protected Methods
int	lmdif_lsqfun (int *m, int *n, const double *x, double *fx, int *iflag)
int	lmder_lsqfun (int *m, int *n, const double *x, double *fx, double *fJ, int &, int *iflag)
Protected Attributes
vnl_least_squares_function *	f_
vnl_matrix< double > *	fdjac_
vnl_vector< int > *	ipvt_
vnl_matrix< double > *	covariance_
bool	set_covariance_
Friends
class	vnl_levenberg_marquardt_Activate

---

Detailed Description

Levenberg Marquardt nonlinear least squares. vnl_levenberg_marquardt is an interface to the MINPACK routine lmdif, and implements Levenberg Marquardt nonlinear fitting. The function to be minimized is passed as a vnl_least_squares_function object, which may or may not wish to provide derivatives. If derivatives are not supplied, they are calculated by forward differencing, which costs one function evaluation per dimension, but is perfectly accurate. (See Hartley in ``Applications of Invariance in Computer Vision'' for example).

Definition at line 33 of file vnl_levenberg_marquardt.h.

---

Constructor & Destructor Documentation

vnl_levenberg_marquardt::vnl_levenberg_marquardt (  vnl_least_squares_function &    f )  [inline]

Initialize with the function object that is to be minimized. Definition at line 38 of file vnl_levenberg_marquardt.h. References init(), and minimize().

vnl_levenberg_marquardt::vnl_levenberg_marquardt (  vnl_least_squares_function &    f, vnl_vector< double > &    x )  [inline]

Initialize as above, and then run minimization. Definition at line 42 of file vnl_levenberg_marquardt.h.

vnl_levenberg_marquardt::~vnl_levenberg_marquardt (    )

---

Member Function Documentation

void vnl_levenberg_marquardt::diagnose_outcome (  vcl_ostream &    )  const

void vnl_levenberg_marquardt::diagnose_outcome (    )  const

Provide an ASCII diagnosis of the last minimization on ostream.

vnl_matrix<double> const& vnl_levenberg_marquardt::get_JtJ (    )

Return J'*J computed at last minimum.

void vnl_levenberg_marquardt::init (  vnl_least_squares_function *    f )  [protected]

Referenced by vnl_levenberg_marquardt().

int vnl_levenberg_marquardt::lmder_lsqfun (  int *    m, int *    n, const double *    x, double *    fx, double *    fJ, int &   , int *    iflag )  [static, protected]

int vnl_levenberg_marquardt::lmdif_lsqfun (  int *    m, int *    n, const double *    x, double *    fx, int *    iflag )  [static, protected]

bool vnl_levenberg_marquardt::minimize (  vnl_vector< double > &    x )  [inline]

Definition at line 58 of file vnl_levenberg_marquardt.h. References covariance_, f_, fdjac_, ipvt_, and set_covariance_. Referenced by vnl_levenberg_marquardt().

bool vnl_levenberg_marquardt::minimize_using_gradient (  vnl_vector< double > &    x )

bool vnl_levenberg_marquardt::minimize_without_gradient (  vnl_vector< double > &    x )

Minimize the function supplied in the constructor until convergence. or failure. On return, x is such that f(x) is the lowest value achieved. Returns true for convergence, false for failure.

---

Friends And Related Function Documentation

friend class vnl_levenberg_marquardt_Activate [friend]

Definition at line 83 of file vnl_levenberg_marquardt.h.

---

Member Data Documentation

vnl_matrix<double>* vnl_levenberg_marquardt::covariance_ [protected]

Definition at line 77 of file vnl_levenberg_marquardt.h. Referenced by minimize().

vnl_least_squares_function* vnl_levenberg_marquardt::f_ [protected]

Definition at line 73 of file vnl_levenberg_marquardt.h. Referenced by minimize().

vnl_matrix<double>* vnl_levenberg_marquardt::fdjac_ [protected]

Definition at line 74 of file vnl_levenberg_marquardt.h. Referenced by minimize().

vnl_vector<int>* vnl_levenberg_marquardt::ipvt_ [protected]

Definition at line 75 of file vnl_levenberg_marquardt.h. Referenced by minimize().

bool vnl_levenberg_marquardt::set_covariance_ [protected]

Definition at line 78 of file vnl_levenberg_marquardt.h. Referenced by minimize().

---

The documentation for this class was generated from the following file:

• vnl_levenberg_marquardt.h

---