Performs minimization of multivariate functions by the LBFGS method. More...

Inheritance diagram for NumLib.MinimizerLbfgs:

Collaboration diagram for NumLib.MinimizerLbfgs:

Public Member Functions
	MinimizerLbfgs (FunctionWithGradient function, double[] initialGuess, int maxIterations, double tolGrad, double tolValue, double tolParameters)
	Constructs a new function mnimizer. More...

override void	InitializeIteration ()
	Initializes the solution iteration. More...

override void	NextIteration ()
	Optimization of quadratic functions. More...

override void	Minimize ()
	Optimization of quadratic functions. More...

Protected Attributes
int	_maxIterationsLbfgs = DefaultMaxIterationsLbfgs

lbfgs.lbfgsstate	LbfgsState
	LBFGS algorithm's state struct. More...

lbfgs.lbfgsreport	LbfgsReport
	LBFGS algorithm's results. More...

Protected Attributes inherited from NumLib.MinimizerBase
int	_outputLevel = DefaultOutputLevel

int	_maxIterations = DefaultMaxIterations

double	_toleranceFunctionValue = DefaultToleranceFunctionValue

double	_toleranceParameterNorm = DefaultToleranceParameterNorm

double	_toleranceGradientNorm = DefaultToleranceGradientNorm

FunctionWithGradient	_function

double[]	_initialGuess

int	_numParameters

double[]	_currentGuess

double	_currentValue

double[]	_currentGradient

double[]	_solutionParameters

double	_solutionValue

double[]	_solutionGradient

int	_iterationCount = 0

int	_evalutaionCount = 0

bool	_isConvergenceCriteriaMet = false

Static Protected Attributes
static int	_defaultMaxIterationsLbfgs = 500

Static Protected Attributes inherited from NumLib.MinimizerBase
static int	_defaultOutputLevel = 0

static int	_defaultMaxiIterations = 1000

static double	_defaultToleranceFunctionValue = -1

static double	_defaultToleranceParemeterNorm = -1

static double	_defaultToleranceGradientNorm = 1.0e-6

Properties
static int	DefaultMaxIterationsLbfgs `[get, set]`
	Default value for the maximal number of iterations for the Lbfgs method. More...

override int	MaxIterations `[get, set]`
	Maximal number of iterations. More...

Properties inherited from NumLib.MinimizerBase
static int	DefaultOutputLevel `[get, set]`
	Default output level for nonlinear equation solvers. More...

static int	DefaultMaxIterations `[get, set]`
	Default value for the maximal number of iterations. More...

static double	DefaultToleranceFunctionValue `[get, set]`
	Default tolerance on the error of the value of the minimized function in solution. More...

static double	DefaultToleranceParameterNorm `[get, set]`
	Default tolerance on norm of the errors in parameters in the solution. More...

static double	DefaultToleranceGradientNorm `[get, set]`
	Default tolerance on norm of the function gradient in the solution. More...

virtual int	OutputLevel `[get, set]`
	Level of console output for nonlinear equation solvers. More...

virtual int	MaxIterations `[get, set]`
	Maximal number of iterations. More...

virtual double	ToleranceFunctionValue `[get, set]`
	Tolerance on absolute error of function value of the function in solution. More...

virtual double	ToleranceParameters `[get, set]`
	Tolerance on norm of error in parameters in solution. More...

virtual double	ToleranceFunctionGradient `[get, set]`
	Tolerance on absolute error of function value of the function in solution. More...

virtual FunctionWithGradient	Function `[get, set]`
	Function to be minimized. More...

virtual double[]	InitialGuess `[get, set]`
	Initial guess. More...

virtual int	NumParameters `[get, protected set]`
	Number of parameters. More...

virtual double[]	CurrentGuess `[get, set]`
	Current guess. More...

virtual double	CurrentValue `[get, set]`
	Current function value. More...

virtual double[]	CurrentGradient `[get, set]`
	Current function gradient. More...

virtual double[]	SolutionParameters `[get, set]`
	Solution of the prioblem - parameter values in minimum. More...

virtual double	SolutionValue `[get, set]`
	Function value in solution. More...

virtual double[]	SolutionGradient `[get, set]`
	Function gradient in solution. More...

virtual int	IterationCount `[get, protected set]`
	Current iteration number. More...

virtual int	Evaluationcount `[get, protected set]`
	Number of function evaluations performed up to now (usually within the single minimization procedure). More...

virtual bool	IsConvergenceCriteriaMet `[get]`
	Returna a flag telling whether convergence criteria are met. More...

Private Attributes
int	numHessianCorrections = 2
	Number of Hessian corrections. More...

int	flags = 0
	Flags used for the LBFGS algorithm. More...

Detailed Description

Performs minimization of multivariate functions by the LBFGS method.

Constructor & Destructor Documentation

NumLib.MinimizerLbfgs.MinimizerLbfgs	(	FunctionWithGradient	function,
		double[]	initialGuess,
		int	maxIterations,
		double	tolGrad,
		double	tolValue,
		double	tolParameters
	)

inline

Constructs a new function mnimizer.

Parameters

function	Function to be minimized.
initialGuess	Initial guess.
maxIterations	Maximal number of iterations.
tolGrad	Tolerance on norm of function gradient.
tolValue	Tolerance on error in function value.
tolParameters	Tolerance on norm of error in parameters.