BFGSOpt Namespace Reference

Functions
void	linearSearch (unsigned int dim, double *oldPt, double oldVal, double *grad, double *dir, double *newPt, double &newVal, double(*func)(double *), double maxStep, int &resCode)
	Do a Quasi-Newton minimization along a line.
int	minimize (unsigned int dim, double *pos, double gradTol, unsigned int &numIters, double &funcVal, double(*func)(double *), void(*gradFunc)(double *, double *), double funcTol=TOLX, unsigned int maxIts=MAXITS)
	Do a BFGS minimization of a function.
Variables
const double	FUNCTOL = 1e-4
	Default tolerance for function convergence in the minimizer.
const double	MOVETOL = 1e-7
	Default tolerance for x changes in the minimizer.
const int	MAXITS = 200
	Default maximum number of iterations.
const double	EPS = 3e-8
	Default gradient tolerance in the minimizer.
const double	TOLX = 4.*EPS
	Default direction vector tolerance in the minimizer.
const double	MAXSTEP = 100.0
	Default maximim step size in the minimizer.

---

Function Documentation

void BFGSOpt::linearSearch	(	unsigned int	dim,
		double *	oldPt,
		double	oldVal,
		double *	grad,
		double *	dir,
		double *	newPt,
		double &	newVal,
		double(*)(double *)	func,
		double	maxStep,
		int &	resCode
	)

Do a Quasi-Newton minimization along a line.

See Numerical Recipes in C, Section 9.7 for a description of the algorithm.

Parameters:

	dim	the dimensionality of the space.
	oldPt	the current position, as an array.
	oldVal	the current function value.
	grad	the value of the function gradient at oldPt
	dir	the minimization direction
	newPt	used to return the final position
	newVal	used to return the final function value
	func	the function to minimize
	maxStep	the maximum allowable step size
	resCode	used to return the results of the search.

Possible values for resCode are on return are:

• 0: success
• 1: the stepsize got too small. This probably indicates success.
• -1: the direction is bad (orthogonal to the gradient)

int BFGSOpt::minimize	(	unsigned int	dim,
		double *	pos,
		double	gradTol,
		unsigned int &	numIters,
		double &	funcVal,
		double(*)(double *)	func,
		void(*)(double *, double *)	gradFunc,
		double	funcTol = TOLX,
		unsigned int	maxIts = MAXITS
	)

Do a BFGS minimization of a function.

See Numerical Recipes in C, Section 10.7 for a description of the algorithm.

Parameters:

	dim	the dimensionality of the space.
	pos	the starting position, as an array.
	gradTol	tolerance for gradient convergence
	numIters	used to return the number of iterations required
	funcVal	used to return the final function value
	func	the function to minimize
	gradFunc	calculates the gradient of func
	funcTol	tolerance for changes in the function value for convergence.
	maxIts	maximum number of iterations allowed

Returns:

a flag indicating success (or type of failure). Possible values are:

• 0: success
• 1: too many iterations were required

---

Variable Documentation

const double BFGSOpt::EPS = 3e-8

Default gradient tolerance in the minimizer.

Definition at line 11 of file BFGSOpt.h.

const double BFGSOpt::FUNCTOL = 1e-4

Default tolerance for function convergence in the minimizer.

Definition at line 8 of file BFGSOpt.h.

const int BFGSOpt::MAXITS = 200

Default maximum number of iterations.

Definition at line 10 of file BFGSOpt.h.

const double BFGSOpt::MAXSTEP = 100.0

Default maximim step size in the minimizer.

Definition at line 13 of file BFGSOpt.h.

const double BFGSOpt::MOVETOL = 1e-7

Default tolerance for x changes in the minimizer.

Definition at line 9 of file BFGSOpt.h.

const double BFGSOpt::TOLX = 4.*EPS

Default direction vector tolerance in the minimizer.

Definition at line 12 of file BFGSOpt.h.