roboptim::finiteDifferenceGradientPolicies::Policy< T > Class Template Reference

Interface for the finite difference gradient policies. More...

#include <roboptim/core/decorator/finite-difference-gradient.hh>

Inheritance diagram for roboptim::finiteDifferenceGradientPolicies::Policy< T >:

Collaboration diagram for roboptim::finiteDifferenceGradientPolicies::Policy< T >:

Public Member Functions
	ROBOPTIM_DIFFERENTIABLE_FUNCTION_FWD_TYPEDEFS_ (GenericDifferentiableFunction< T >)
	Policy (const GenericFunction< T > &adaptee)
virtual	~Policy ()
	Virtual destructor.
virtual void	computeGradient (value_type epsilon, gradient_ref gradient, const_argument_ref argument, size_type idFunction, argument_ref xEps) const =0
virtual void	computeColumn (value_type epsilon, gradient_ref column, const_argument_ref argument, size_type colIdx, argument_ref xEps) const =0
virtual void	computeJacobian (value_type epsilon, jacobian_ref jacobian, const_argument_ref argument, argument_ref xEps) const
void	computeJacobian (value_type epsilon, jacobian_ref jacobian, const_argument_ref argument, argument_ref xEps) const

Protected Attributes
const GenericFunction< T > &	adaptee_
	Wrapped function.
vector_t	column_
	Vector storing temporary Jacobian column.
gradient_t	gradient_
	Vector storing temporary Jacobian row.

Detailed Description

template<typename T>
class roboptim::finiteDifferenceGradientPolicies::Policy< T >

Interface for the finite difference gradient policies.

Constructor & Destructor Documentation

Policy()

template<typename T>

roboptim::finiteDifferenceGradientPolicies::Policy< T >::Policy ( const GenericFunction< T > & adaptee )

inlineexplicit

References adaptee_, column_, and gradient_.

Referenced by roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >::FivePointsRule(), and roboptim::finiteDifferenceGradientPolicies::Simple< T >::Simple().

~Policy()

template<typename T>

virtual roboptim::finiteDifferenceGradientPolicies::Policy< T >::~Policy ( )

inlinevirtual

Virtual destructor.

Member Function Documentation

computeColumn()

template<typename T>

virtual void roboptim::finiteDifferenceGradientPolicies::Policy< T >::computeColumn ( value_type epsilon, gradient_ref column, const_argument_ref argument, size_type colIdx, argument_ref xEps ) const

pure virtual

Implemented in roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >, roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >, roboptim::finiteDifferenceGradientPolicies::Simple< T >, and roboptim::finiteDifferenceGradientPolicies::Simple< T >.

Referenced by computeJacobian().

computeGradient()

template<typename T>

virtual void roboptim::finiteDifferenceGradientPolicies::Policy< T >::computeGradient ( value_type epsilon, gradient_ref gradient, const_argument_ref argument, size_type idFunction, argument_ref xEps ) const

pure virtual

Implemented in roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >, roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >, roboptim::finiteDifferenceGradientPolicies::Simple< T >, and roboptim::finiteDifferenceGradientPolicies::Simple< T >.

computeJacobian() [1/2]

template<typename T>

void roboptim::finiteDifferenceGradientPolicies::Policy< T >::computeJacobian ( value_type epsilon, jacobian_ref jacobian, const_argument_ref argument, argument_ref xEps ) const

virtual

Reimplemented in roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >, roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >, and roboptim::finiteDifferenceGradientPolicies::Simple< T >.

References adaptee_, column_, and computeColumn().

Referenced by roboptim::finiteDifferenceGradientPolicies::Simple< T >::computeJacobian().

computeJacobian() [2/2]

void roboptim::finiteDifferenceGradientPolicies::Policy< EigenMatrixSparse >::computeJacobian ( value_type epsilon, jacobian_ref jacobian, const_argument_ref argument, argument_ref xEps ) const

inline

ROBOPTIM_DO_NOT_CHECK_ALLOCATION

ROBOPTIM_DO_NOT_CHECK_ALLOCATION

References adaptee_, computeColumn(), roboptim::is_malloc_allowed(), and roboptim::set_is_malloc_allowed().

ROBOPTIM_DIFFERENTIABLE_FUNCTION_FWD_TYPEDEFS_()

template<typename T>

roboptim::finiteDifferenceGradientPolicies::Policy< T >::ROBOPTIM_DIFFERENTIABLE_FUNCTION_FWD_TYPEDEFS_

(

GenericDifferentiableFunction< T >

)

Member Data Documentation

adaptee_

template<typename T>

const GenericFunction<T>& roboptim::finiteDifferenceGradientPolicies::Policy< T >::adaptee_

protected

Wrapped function.

Referenced by roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >::compute_deriv(), roboptim::finiteDifferenceGradientPolicies::Simple< T >::computeColumn(), roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >::computeGradient(), roboptim::finiteDifferenceGradientPolicies::Simple< T >::computeGradient(), roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >::computeJacobian(), computeJacobian(), roboptim::finiteDifferenceGradientPolicies::Simple< T >::computeJacobian(), and Policy().

column_

template<typename T>

vector_t roboptim::finiteDifferenceGradientPolicies::Policy< T >::column_

mutableprotected

Vector storing temporary Jacobian column.

Referenced by computeJacobian(), and Policy().

gradient_

template<typename T>

gradient_t roboptim::finiteDifferenceGradientPolicies::Policy< T >::gradient_

mutableprotected

Vector storing temporary Jacobian row.

Referenced by roboptim::finiteDifferenceGradientPolicies::FivePointsRule< T >::computeJacobian(), and Policy().