dart::optimizer::ModularFunction Class Reference

ModularFunction uses C++11 std::function to allow you to easily swap out the cost function, gradient function, and Hessian function during runtime for an optimizer::Function instance. More...

#include <Function.hpp>

Inheritance diagram for dart::optimizer::ModularFunction:

Public Member Functions
	ModularFunction (const std::string &_name="modular_function")
	Constructor. More...
	~ModularFunction () override
	Destructor. More...
double	eval (const Eigen::VectorXd &_x) override
	eval() will now call whatever CostFunction you set using setCostFunction() More...
void	evalGradient (const Eigen::VectorXd &_x, Eigen::Map< Eigen::VectorXd > _grad) override
	evalGradient() will now call whatever GradientFunction you set using setGradientFunction() More...
void	evalHessian (const Eigen::VectorXd &_x, Eigen::Map< Eigen::VectorXd, Eigen::RowMajor > _Hess) override
	evalHessian() will now call whatever HessianFunction you set using setHessianFunction() More...
void	setCostFunction (CostFunction _cost)
	Set the function that gets called by eval() More...
void	clearCostFunction (bool _printWarning=true)
	Replace the cost function with a constant-zero function. More...
void	setGradientFunction (GradientFunction _gradient)
	Set the function that gets called by evalGradient() More...
void	clearGradientFunction ()
	Replace the gradient function with the default evalGradient() of the base Function class. More...
void	setHessianFunction (HessianFunction _hessian)
	Set the function that gets called by evalHessian() More...
void	clearHessianFunction ()
	Replace the Hessian function with the default evalHessian() of the base Function class. More...
virtual void	setName (const std::string &newName)
	Sets the name of this Function. More...
const std::string &	getName () const
	Returns the name of this Function. More...
void	evalGradient (const Eigen::VectorXd &_x, Eigen::VectorXd &_grad)
	Evaluates and return the objective function at the point x. More...

Protected Attributes
CostFunction	mCostFunction
	Storage for the cost function. More...
GradientFunction	mGradientFunction
	Storage for the gradient function. More...
HessianFunction	mHessianFunction
	Storage for the Hessian function. More...
std::string	mName
	Name of this function. More...

Detailed Description

ModularFunction uses C++11 std::function to allow you to easily swap out the cost function, gradient function, and Hessian function during runtime for an optimizer::Function instance.

Constructor & Destructor Documentation

ModularFunction()

dart::optimizer::ModularFunction::ModularFunction ( const std::string &  _name = "modular_function" )

explicit

Constructor.

~ModularFunction()

dart::optimizer::ModularFunction::~ModularFunction ( )

override

Destructor.

Member Function Documentation

clearCostFunction()

void dart::optimizer::ModularFunction::clearCostFunction

(

bool

_printWarning = true

)

Replace the cost function with a constant-zero function.

Passing in true will cause a warning to be printed out whenever eval() is called.

clearGradientFunction()

void dart::optimizer::ModularFunction::clearGradientFunction

(

)

Replace the gradient function with the default evalGradient() of the base Function class.

A warning will be printed whenever evalGradient() gets called.

clearHessianFunction()

void dart::optimizer::ModularFunction::clearHessianFunction

(

)

Replace the Hessian function with the default evalHessian() of the base Function class.

A warning will be printed whenever evalHessian() gets called.

eval()

double dart::optimizer::ModularFunction::eval ( const Eigen::VectorXd &  _x )

overridevirtual

eval() will now call whatever CostFunction you set using setCostFunction()

Implements dart::optimizer::Function.

evalGradient() [1/2]

void dart::optimizer::ModularFunction::evalGradient ( const Eigen::VectorXd &  _x, Eigen::Map< Eigen::VectorXd >  _grad  )

overridevirtual

evalGradient() will now call whatever GradientFunction you set using setGradientFunction()

Reimplemented from dart::optimizer::Function.

evalGradient() [2/2]

void dart::optimizer::Function::evalGradient ( const Eigen::VectorXd &  _x, Eigen::VectorXd &  _grad  )

inherited

Evaluates and return the objective function at the point x.

If you have a raw array that the gradient will be passed in, then use evalGradient(const Eigen::VectorXd&, Eigen::Map<Eigen::VectorXd>) for better performance.

evalHessian()

void dart::optimizer::ModularFunction::evalHessian ( const Eigen::VectorXd &  _x, Eigen::Map< Eigen::VectorXd, Eigen::RowMajor >  _Hess  )

overridevirtual

evalHessian() will now call whatever HessianFunction you set using setHessianFunction()

Reimplemented from dart::optimizer::Function.

getName()

const std::string & dart::optimizer::Function::getName ( ) const

inherited

Returns the name of this Function.

setCostFunction()

void dart::optimizer::ModularFunction::setCostFunction

(

CostFunction

_cost

)

Set the function that gets called by eval()

setGradientFunction()

void dart::optimizer::ModularFunction::setGradientFunction

(

GradientFunction

_gradient

)

Set the function that gets called by evalGradient()

setHessianFunction()

void dart::optimizer::ModularFunction::setHessianFunction

(

HessianFunction

_hessian

)

Set the function that gets called by evalHessian()

setName()

void dart::optimizer::Function::setName ( const std::string &  newName )

virtualinherited

Sets the name of this Function.

Member Data Documentation

mCostFunction

CostFunction dart::optimizer::ModularFunction::mCostFunction

protected

Storage for the cost function.

mGradientFunction

GradientFunction dart::optimizer::ModularFunction::mGradientFunction

protected

Storage for the gradient function.

mHessianFunction

HessianFunction dart::optimizer::ModularFunction::mHessianFunction

protected

Storage for the Hessian function.

mName

std::string dart::optimizer::Function::mName

protectedinherited

Name of this function.