Levenberg-Marquardt algorithm for least squares optimization. More...

#include <LevenbergMarquardt.hpp>

Inheritance diagram for sgpp::optimization::optimizer::LevenbergMarquardt:

Public Member Functions
void	clone (std::unique_ptr< LeastSquaresOptimizer > &clone) const override

double	getAcceptanceThreshold () const

double	getEffectivenessThreshold () const

double	getInitialDamping () const

base::VectorFunctionGradient &	getPhiGradient () const

double	getTolerance () const

	LevenbergMarquardt (const base::VectorFunction &phi, const base::VectorFunctionGradient &phiGradient, size_t maxItCount, double tolerance, double initialDamping, double acceptanceThreshold, double effectivenessThreshold, const base::sle_solver::SLESolver &sleSolver)
	Constructor.

	LevenbergMarquardt (const base::VectorFunction &phi, const base::VectorFunctionGradient &phiGradient, size_t maxItCount=DEFAULT_N, double tolerance=DEFAULT_TOLERANCE, double initialDamping=DEFAULT_INITIAL_DAMPING, double acceptanceThreshold=DEFAULT_ACCEPTANCE_THRESHOLD, double effectivenessThreshold=DEFAULT_EFFECTIVENESS_THRESHOLD)
	Constructor.

	LevenbergMarquardt (const LevenbergMarquardt &other)
	Copy constructor.

void	optimize () override
	Pure virtual method for optimization of the objective function.

void	setAcceptanceThreshold (double acceptanceThreshold)

void	setEffectivenessThreshold (double effectivenessThreshold)

void	setInitialDamping (double initialDamping)

void	setTolerance (double tolerance)

	~LevenbergMarquardt () override
	Destructor.

Public Member Functions inherited from sgpp::optimization::optimizer::LeastSquaresOptimizer
const base::DataMatrix &	getHistoryOfOptimalPoints () const

const base::DataVector &	getHistoryOfOptimalValues () const

size_t	getN () const

const base::DataVector &	getOptimalPoint () const

double	getOptimalValue () const

base::VectorFunction &	getPhiFunction () const

const base::DataVector &	getStartingPoint () const

	LeastSquaresOptimizer (const base::VectorFunction &phi, size_t N=DEFAULT_N)
	Constructor.

	LeastSquaresOptimizer (const LeastSquaresOptimizer &other)
	Copy constructor.

void	setN (size_t N)

void	setPhiFunction (const base::VectorFunction &phi)

void	setStartingPoint (const base::DataVector &startingPoint)

virtual	~LeastSquaresOptimizer ()
	Destructor.

Static Public Attributes
static constexpr double	DEFAULT_ACCEPTANCE_THRESHOLD = 0.3
	default acceptance threshold

static constexpr double	DEFAULT_EFFECTIVENESS_THRESHOLD = 0.9
	default effectiveness threshold

static constexpr double	DEFAULT_INITIAL_DAMPING = 1.0
	default initial damping

static constexpr double	DEFAULT_TOLERANCE = 1e-6
	default tolerance

Static Public Attributes inherited from sgpp::optimization::optimizer::LeastSquaresOptimizer
static const size_t	DEFAULT_N = 1000
	default maximal number of iterations or function evaluations

Protected Attributes
double	beta0
	acceptance threshold

double	beta1
	effectiveness threshold

const base::sle_solver::GaussianElimination	defaultSleSolver
	default linear solver

double	mu0
	initial damping

std::unique_ptr< base::VectorFunctionGradient >	phiGradient
	phi gradient

const base::sle_solver::SLESolver &	sleSolver
	linear solver

double	tol
	tolerance

Protected Attributes inherited from sgpp::optimization::optimizer::LeastSquaresOptimizer
base::DataVector	fHist
	search history vector (optimal values)

double	fOpt
	result of optimization (optimal function value)

size_t	N
	maximal number of iterations or function evaluations

std::unique_ptr< base::VectorFunction >	phi
	phi function

base::DataVector	x0
	starting point

base::DataMatrix	xHist
	search history matrix (optimal points)

base::DataVector	xOpt
	result of optimization (location of optimum)

Detailed Description

Levenberg-Marquardt algorithm for least squares optimization.

Constructor & Destructor Documentation

◆ LevenbergMarquardt() [1/3]

sgpp::optimization::optimizer::LevenbergMarquardt::LevenbergMarquardt	(	const base::VectorFunction &	phi,
		const base::VectorFunctionGradient &	phiGradient,
		size_t	maxItCount = `DEFAULT_N`,
		double	tolerance = `DEFAULT_TOLERANCE`,
		double	initialDamping = `DEFAULT_INITIAL_DAMPING`,
		double	acceptanceThreshold = `DEFAULT_ACCEPTANCE_THRESHOLD`,
		double	effectivenessThreshold = `DEFAULT_EFFECTIVENESS_THRESHOLD`
	)

Constructor.

By default, GaussianElimination is used to solve the linear systems.

Parameters

phi	base function
phiGradient	Jacobian of phi
maxItCount	maximal number of function evaluations
tolerance	tolerance
initialDamping	initial damping
acceptanceThreshold	acceptance threshold
effectivenessThreshold	effectiveness threshold

References sgpp::optimization::optimizer::LeastSquaresOptimizer::clone(), and phiGradient.

◆ LevenbergMarquardt() [2/3]

sgpp::optimization::optimizer::LevenbergMarquardt::LevenbergMarquardt	(	const base::VectorFunction &	phi,
		const base::VectorFunctionGradient &	phiGradient,
		size_t	maxItCount,
		double	tolerance,
		double	initialDamping,
		double	acceptanceThreshold,
		double	effectivenessThreshold,
		const base::sle_solver::SLESolver &	sleSolver
	)

Constructor.

Do not destruct the solver before this object!

Parameters

phi	phi function
phiGradient	Jacobian of phi
maxItCount	maximal number of function evaluations
tolerance	tolerance
initialDamping	initial damping
acceptanceThreshold	acceptance threshold
effectivenessThreshold	effectiveness threshold
sleSolver	reference to linear solver for solving the linear systems

References phiGradient.

◆ LevenbergMarquardt() [3/3]

sgpp::optimization::optimizer::LevenbergMarquardt::LevenbergMarquardt ( const LevenbergMarquardt & other )

Copy constructor.

Parameters

other optimizer to be copied

References phiGradient.

◆ ~LevenbergMarquardt()

sgpp::optimization::optimizer::LevenbergMarquardt::~LevenbergMarquardt ( )

override

Destructor.

Member Function Documentation

◆ clone()

void sgpp::optimization::optimizer::LevenbergMarquardt::clone ( std::unique_ptr< LeastSquaresOptimizer > & clone ) const

overridevirtual

Parameters

[out] clone pointer to cloned object

Implements sgpp::optimization::optimizer::LeastSquaresOptimizer.

References clone().

Referenced by clone().

◆ getAcceptanceThreshold()

double sgpp::optimization::optimizer::LevenbergMarquardt::getAcceptanceThreshold ( ) const

Returns: acceptanceThreshold

References beta0.

◆ getEffectivenessThreshold()

double sgpp::optimization::optimizer::LevenbergMarquardt::getEffectivenessThreshold ( ) const

Returns: effectiveness threshold

References beta1.

◆ getInitialDamping()

double sgpp::optimization::optimizer::LevenbergMarquardt::getInitialDamping ( ) const

Returns: initial damping

References mu0.

◆ getPhiGradient()

base::VectorFunctionGradient & sgpp::optimization::optimizer::LevenbergMarquardt::getPhiGradient ( ) const

Returns: phi gradient

References phiGradient.

◆ getTolerance()

double sgpp::optimization::optimizer::LevenbergMarquardt::getTolerance ( ) const

Returns: tolerance

References tol.

◆ optimize()

void sgpp::optimization::optimizer::LevenbergMarquardt::optimize ( )

overridevirtual

Pure virtual method for optimization of the objective function.

The result of the optimization process can be obtained by member functions, e.g., getOptimalPoint() and getOptimalValue().

Implements sgpp::optimization::optimizer::LeastSquaresOptimizer.

References sgpp::base::DataVector::append(), sgpp::base::DataMatrix::appendRow(), beta0, beta1, sgpp::base::Printer::disableStatusPrinting(), sgpp::base::Printer::enableStatusPrinting(), sgpp::optimization::optimizer::LeastSquaresOptimizer::fHist, sgpp::optimization::optimizer::LeastSquaresOptimizer::fOpt, sgpp::base::Printer::getInstance(), sgpp::base::Printer::isStatusPrintingEnabled(), m, mu, mu0, sgpp::base::DataMatrix::mult(), sgpp::optimization::optimizer::LeastSquaresOptimizer::N, sgpp::optimization::optimizer::LeastSquaresOptimizer::phi, phiGradient, sgpp::base::Printer::printStatusBegin(), sgpp::base::Printer::printStatusEnd(), sgpp::base::Printer::printStatusUpdate(), sgpp::base::DataMatrix::resize(), sleSolver, sgpp::base::sle_solver::SLESolver::solve(), tol, sgpp::base::DataVector::toString(), sgpp::optimization::optimizer::LeastSquaresOptimizer::x0, sgpp::optimization::optimizer::LeastSquaresOptimizer::xHist, and sgpp::optimization::optimizer::LeastSquaresOptimizer::xOpt.

◆ setAcceptanceThreshold()

void sgpp::optimization::optimizer::LevenbergMarquardt::setAcceptanceThreshold ( double acceptanceThreshold )

Parameters

acceptanceThreshold acceptance threshold

References beta0.

◆ setEffectivenessThreshold()

void sgpp::optimization::optimizer::LevenbergMarquardt::setEffectivenessThreshold ( double effectivenessThreshold )

Parameters

effectivenessThreshold effectiveness threshold

References beta1.

◆ setInitialDamping()

void sgpp::optimization::optimizer::LevenbergMarquardt::setInitialDamping ( double initialDamping )

Parameters

initialDamping initial damping

References mu0.

◆ setTolerance()

void sgpp::optimization::optimizer::LevenbergMarquardt::setTolerance ( double tolerance )

Parameters

tolerance tolerance

References tol.

Member Data Documentation

◆ beta0

double sgpp::optimization::optimizer::LevenbergMarquardt::beta0

protected

acceptance threshold

Referenced by getAcceptanceThreshold(), optimize(), and setAcceptanceThreshold().

◆ beta1

double sgpp::optimization::optimizer::LevenbergMarquardt::beta1

protected

effectiveness threshold

Referenced by getEffectivenessThreshold(), optimize(), and setEffectivenessThreshold().

◆ DEFAULT_ACCEPTANCE_THRESHOLD

constexpr double sgpp::optimization::optimizer::LevenbergMarquardt::DEFAULT_ACCEPTANCE_THRESHOLD = 0.3

staticconstexpr

default acceptance threshold

◆ DEFAULT_EFFECTIVENESS_THRESHOLD

constexpr double sgpp::optimization::optimizer::LevenbergMarquardt::DEFAULT_EFFECTIVENESS_THRESHOLD = 0.9

staticconstexpr

default effectiveness threshold

◆ DEFAULT_INITIAL_DAMPING

constexpr double sgpp::optimization::optimizer::LevenbergMarquardt::DEFAULT_INITIAL_DAMPING = 1.0

staticconstexpr

default initial damping

◆ DEFAULT_TOLERANCE

constexpr double sgpp::optimization::optimizer::LevenbergMarquardt::DEFAULT_TOLERANCE = 1e-6

staticconstexpr

default tolerance

◆ defaultSleSolver

const base::sle_solver::GaussianElimination sgpp::optimization::optimizer::LevenbergMarquardt::defaultSleSolver

protected

default linear solver

◆ mu0

double sgpp::optimization::optimizer::LevenbergMarquardt::mu0

protected

initial damping

Referenced by getInitialDamping(), optimize(), and setInitialDamping().

◆ phiGradient

std::unique_ptr<base::VectorFunctionGradient> sgpp::optimization::optimizer::LevenbergMarquardt::phiGradient

protected

phi gradient

Referenced by getPhiGradient(), LevenbergMarquardt(), LevenbergMarquardt(), LevenbergMarquardt(), and optimize().

◆ sleSolver

const base::sle_solver::SLESolver& sgpp::optimization::optimizer::LevenbergMarquardt::sleSolver

protected

linear solver

Referenced by optimize().

◆ tol

double sgpp::optimization::optimizer::LevenbergMarquardt::tol

protected

tolerance

Referenced by getTolerance(), optimize(), and setTolerance().

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

optimization/src/sgpp/optimization/optimizer/least_squares/LevenbergMarquardt.hpp
optimization/src/sgpp/optimization/optimizer/least_squares/LevenbergMarquardt.cpp

Public Member Functions

Static Public Attributes

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ LevenbergMarquardt() [1/3]

◆ LevenbergMarquardt() [2/3]

◆ LevenbergMarquardt() [3/3]

◆ ~LevenbergMarquardt()

Member Function Documentation

◆ clone()

◆ getAcceptanceThreshold()

◆ getEffectivenessThreshold()

◆ getInitialDamping()

◆ getPhiGradient()

◆ getTolerance()

◆ optimize()

◆ setAcceptanceThreshold()

◆ setEffectivenessThreshold()

◆ setInitialDamping()

◆ setTolerance()

Member Data Documentation

◆ beta0

◆ beta1

◆ DEFAULT_ACCEPTANCE_THRESHOLD

◆ DEFAULT_EFFECTIVENESS_THRESHOLD

◆ DEFAULT_INITIAL_DAMPING

◆ DEFAULT_TOLERANCE

◆ defaultSleSolver

◆ mu0

◆ phiGradient

◆ sleSolver

◆ tol