A globally convergent Newton's method for multidimensional root solving. Solves for func(x)=0. More...

#include <Newton.h>

Inheritance diagram for Optimization::NewtonRoot:

Public Member Functions
	NewtonRoot (VectorFieldFunction *func)

bool	GlobalSolve (int &iters, ConvergenceResult *res=NULL)

ConvergenceResult	Solve (int &iters)

ConvergenceResult	Solve_Sparse (int &iters)

bool	LineMinimization (const Vector &g, const Vector &p, Real *f)

Real	MaxDistance (const Vector &x)

virtual bool	SolveUnderconstrainedLS (const Matrix &A, const Vector &b, Vector &x)

virtual bool	SolveUnderconstrainedLS (const SparseMatrix &A, const Vector &b, Vector &x)

virtual Real	Merit ()

Public Attributes
Vector	x

VectorFieldFunction *	func

Real	tolf

Real	tolmin

Real	tolx

Real	stepMax
	maximum distance to step

Real	lambda
	damped-least-squares constant

Vector	bmin

Vector	bmax
	optional bound constraints

bool	sparse
	set to true if should use a sparse least-squares solver.

Vector	bias
	set this to a bias vector to solve for min \|\|x-bias\|\| s.t. func(x)=0

int	verbose

int	debug

RobustSVD< Real >	svd

Vector	fx

Vector	g

Vector	p

Vector	xold

Matrix	fJx

Detailed Description

A globally convergent Newton's method for multidimensional root solving. Solves for func(x)=0.

Can use sparse methods for solving for a SparseVectorFieldFunction if sparse = true. However, cannot use the bias-vector option.

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

Public Member Functions