roboptim::Problem< T > Class Template Reference

#include <roboptim/core/problem.hh>

Public Types
typedef GenericFunction< T >	function_t
	Function type. More...
typedef boost::mpl::vector < function_t >	constraintsList_t
	Constraints types list. More...
typedef boost::shared_ptr < function_t >	constraint_t
	Constraint's type. More...
typedef function_t::value_type	value_type
	Import function's value_type type. More...
typedef function_t::vector_t	vector_t
	Vector type. More...
typedef function_t::argument_t	argument_t
	Argument type. More...
typedef function_t::result_t	result_t
	Result type. More...
typedef function_t::size_type	size_type
	Size type. More...
typedef std::vector< constraint_t >	constraints_t
	Constraints are represented as a vector of constraints. More...
typedef boost::optional < argument_t >	startingPoint_t
	Optional vector defines a starting point. More...
typedef function_t::interval_t	interval_t
	Interval type (e.g. for bounds). More...
typedef function_t::intervals_t	intervals_t
	Intervals type. More...
typedef std::vector< value_type >	scaling_t
	Scaling vector. More...
typedef scaling_t	scales_t
	Scaling vector (deprecated typedef) More...
typedef function_t::names_t	names_t
	Vector of names (e.g. for arguments). More...
typedef std::vector< intervals_t >	intervalsVect_t
	Vector of interval vectors. More...
typedef std::vector< scaling_t >	scalingVect_t
	Vector of scaling vectors. More...
typedef scalingVect_t	scalesVect_t
	Vector of scaling vectors (deprecated typedef). More...
typedef GenericFunctionTraits < T >::jacobian_t	jacobian_t
	Jacobian matrix type. More...
typedef GenericFunctionTraits < T >::const_argument_ref	const_argument_ref
	Constant reference to an argument vector. More...

Public Member Functions
std::ostream &	print (std::ostream &o) const
	Display the problem on the specified output stream. More...
template<>
Problem< EigenMatrixSparse > ::jacobian_t	jacobian (const_argument_ref x) const
template<int NORM>
Problem< T >::value_type	constraintsViolation (const_argument_ref x) const
Constructors and destructors.
	Problem (const boost::shared_ptr< const function_t > &cost)
	Constructor taking a shared_ptr to a cost function. More...
	Problem (const function_t &cost)
	Deprecated constructor taking a reference to a cost function. More...
	Problem (const Problem< T > &pb)
	Copy constructor. More...
virtual	~Problem ()
	Virtual destructor. More...
Cost function.
const function_t &	function () const
	Retrieve cost function. More...
intervals_t &	argumentBounds ()
	Retrieve arguments bounds. More...
const intervals_t &	argumentBounds () const
	Retrieve arguments bounds. More...
scaling_t &	objectiveScaling ()
	Retrieve objective scaling. More...
const scaling_t &	objectiveScaling () const
	Retrieve objective scaling. More...
scaling_t &	argumentScaling ()
	Retrieve arguments scaling. More...
const scaling_t &	argumentScaling () const
	Retrieve arguments scaling. More...
scaling_t &	argumentScales ()
	Retrieve arguments scaling (deprecated version). More...
const scaling_t &	argumentScales () const
	Retrieve arguments scaling (deprecated version). More...
names_t &	argumentNames ()
	Retrieve arguments names. More...
const names_t &	argumentNames () const
	Retrieve arguments names. More...
Constraints.
const constraints_t &	constraints () const
	Retrieve constraints. More...
void	addConstraint (boost::shared_ptr< function_t > constraint, interval_t interval, value_type scale=1.)
	Add a constraint to the problem. More...
void	addConstraint (boost::shared_ptr< function_t > constraint, intervals_t intervals, scaling_t scaling)
	Add a constraint to the problem. More...
const intervalsVect_t &	boundsVector () const
	Retrieve constraints bounds vector. More...
intervalsVect_t &	boundsVector ()
	Retrieve constraints bounds vector. More...
const scalingVect_t &	scalingVector () const
	Retrieve constraints scaling vector. More...
const scalingVect_t &	scalesVector () const
	Retrieve constraints scaling vector (deprecated version). More...
size_type	constraintsOutputSize () const
	Return the output size of the problem's constraints. More...
size_type	differentiableConstraintsOutputSize () const
	Return the output size of the problem's differentiable constraints. More...
void	clearConstraints ()
	Clear the constraints from the problem. More...
Starting point (initial guess).
startingPoint_t &	startingPoint ()
	Set the initial guess. More...
const startingPoint_t &	startingPoint () const
	Get the initial guess. More...
Helper methods.
jacobian_t	jacobian (const_argument_ref x) const
	Evaluate the Jacobian matrix of the problem for a given x. More...
jacobian_t	scaledJacobian (const_argument_ref x) const
	Evaluate the scaled Jacobian matrix of the problem for a given x. More...
result_t	constraintsViolationVector (const_argument_ref x) const
	Evaluate the vector of constraints violation for a given x. More...
template<int NORM>
value_type	constraintsViolation (const_argument_ref x) const
	Evaluate the constraint violation for a given x. More...

Detailed Description

template<typename T>
class roboptim::Problem< T >

Examples:

problem-cc.cc.

Member Typedef Documentation

template<typename T>

typedef function_t::argument_t roboptim::Problem< T >::argument_t

Argument type.

template<typename T>

typedef GenericFunctionTraits<T>::const_argument_ref roboptim::Problem< T >::const_argument_ref

Constant reference to an argument vector.

template<typename T>

typedef boost::shared_ptr<function_t> roboptim::Problem< T >::constraint_t

Constraint's type.

template<typename T>

typedef std::vector<constraint_t> roboptim::Problem< T >::constraints_t

Constraints are represented as a vector of constraints.

template<typename T>

typedef boost::mpl::vector<function_t> roboptim::Problem< T >::constraintsList_t

Constraints types list.

template<typename T>

typedef GenericFunction<T> roboptim::Problem< T >::function_t

Function type.

This has to be either Function or one of its sub-classes.

template<typename T>

typedef function_t::interval_t roboptim::Problem< T >::interval_t

Interval type (e.g. for bounds).

template<typename T>

typedef function_t::intervals_t roboptim::Problem< T >::intervals_t

Intervals type.

template<typename T>

typedef std::vector<intervals_t> roboptim::Problem< T >::intervalsVect_t

Vector of interval vectors.

This type is used to take into account the fact that constraints can have output values in \(\mathbb{C}^{m}\).

If \(m=1\), then the associated interval vector contains only one element of interval_t type.

template<typename T>

typedef GenericFunctionTraits<T>::jacobian_t roboptim::Problem< T >::jacobian_t

Jacobian matrix type.

template<typename T>

typedef function_t::names_t roboptim::Problem< T >::names_t

Vector of names (e.g. for arguments).

template<typename T>

typedef function_t::result_t roboptim::Problem< T >::result_t

Result type.

template<typename T>

typedef scaling_t roboptim::Problem< T >::scales_t

Scaling vector (deprecated typedef)

Deprecated:

Deprecated since version 3.1. Use Problem<T>::scaling_t instead.

template<typename T>

typedef scalingVect_t roboptim::Problem< T >::scalesVect_t

Vector of scaling vectors (deprecated typedef).

Deprecated:

Deprecated since version 3.1. Use Problem<T>::scalingVect_t instead.

template<typename T>

typedef std::vector<value_type> roboptim::Problem< T >::scaling_t

Scaling vector.

If \(g_u\) is the constraint as seen by the user, \(g_a\) the constraint as seen by the solver's algorithm, and \(s\) the scaling factor, then we have: \(g_a = s * g_u\)

template<typename T>

typedef std::vector<scaling_t> roboptim::Problem< T >::scalingVect_t

Vector of scaling vectors.

This type is used to take into account the fact that constraints can have output values in \(\mathbb{R}^m\).

If \(m=1\), then the associated scaling vector contains only one element of scaling_t type.

template<typename T>

typedef function_t::size_type roboptim::Problem< T >::size_type

Size type.

template<typename T>

typedef boost::optional<argument_t> roboptim::Problem< T >::startingPoint_t

Optional vector defines a starting point.

template<typename T>

typedef function_t::value_type roboptim::Problem< T >::value_type

Import function's value_type type.

template<typename T>

typedef function_t::vector_t roboptim::Problem< T >::vector_t

Vector type.

Constructor & Destructor Documentation

template<typename T >

roboptim::Problem< T >::Problem ( const boost::shared_ptr< const function_t > &  cost )

explicit

Constructor taking a shared_ptr to a cost function.

Precondition

costfunction \(\mathbb{R}^n \rightarrow \mathbb{R}\)

Parameters

cost	cost function.

template<typename T >

roboptim::Problem< T >::Problem ( const function_t &  cost )

explicit

Deprecated constructor taking a reference to a cost function.

This legacy version meant that we simply kept a const reference to the cost function, which could reference stack variables... This prepares the transition to something safer (shared_ptr).

Parameters

cost	cost function.

Deprecated:

Deprecated since version 3.2. Use the shared_ptr constructor instead.

template<typename T>

roboptim::Problem< T >::Problem ( const Problem< T > &  pb )

explicit

Copy constructor.

Parameters

pb	problem to copy.

template<typename T >

roboptim::Problem< T >::~Problem ( )

virtual

Virtual destructor.

One may want to inherit from Problem to provide extra information to a solver plugin.

Member Function Documentation

template<typename T >

void roboptim::Problem< T >::addConstraint	(	boost::shared_ptr< function_t >	constraint,
		interval_t	interval,
		value_type	scale = 1.
	)

Add a constraint to the problem.

Useful only when \(m=1\), use addConstraint (boost::shared_ptr<C> constraint, intervals_t intervals, scaling_t scaling) instead.

Parameters

constraint	the constraint that will be added
interval	interval in which the constraint is satisfied
scale	constraint scale

Exceptions

std::runtime_error

template<typename T >

void roboptim::Problem< T >::addConstraint	(	boost::shared_ptr< function_t >	constraint,
		intervals_t	intervals,
		scaling_t	scaling
	)

Add a constraint to the problem.

The constraint can be multidimensional.

Parameters

constraint	the constraint that will be added
intervals	interval vector in which the constraint is satisfied
scaling	constraint scaling

Exceptions

std::runtime_error

References ROBOPTIM_DEBUG_ONLY.

template<typename T >

Problem< T >::intervals_t & roboptim::Problem< T >::argumentBounds

(

)

Retrieve arguments bounds.

Arguments bounds define in which interval each argument is valid.

Returns

arguments bounds

template<typename T >

const Problem< T >::intervals_t & roboptim::Problem< T >::argumentBounds

(

)

const

Retrieve arguments bounds.

Arguments bounds define in which interval each argument is valid.

Returns

arguments bounds

template<typename T >

Problem< T >::names_t & roboptim::Problem< T >::argumentNames

(

)

Retrieve arguments names.

Arguments names define a name for each argument. This is particularly useful when logging data.

Returns

arguments names

template<typename T >

const Problem< T >::names_t & roboptim::Problem< T >::argumentNames

(

)

const

Retrieve arguments names.

Arguments names define a name for each argument. This is particularly useful when logging data.

Returns

arguments names

template<typename T >

Problem< T >::scaling_t & roboptim::Problem< T >::argumentScales

(

)

Retrieve arguments scaling (deprecated version).

Deprecated:

Deprecated since version 3.1. Use argumentScaling() instead.

template<typename T >

const Problem< T >::scaling_t & roboptim::Problem< T >::argumentScales

(

)

const

Retrieve arguments scaling (deprecated version).

Deprecated:

Deprecated since version 3.1. Use argumentScaling() instead.

template<typename T >

Problem< T >::scaling_t & roboptim::Problem< T >::argumentScaling

(

)

Retrieve arguments scaling.

Arguments scaling define which scale factor is applied for each argument.

Returns

arguments scaling

template<typename T >

const Problem< T >::scaling_t & roboptim::Problem< T >::argumentScaling

(

)

const

Retrieve arguments scaling.

Arguments scaling define which scale factor is applied for each argument.

Returns

arguments scaling

template<typename T >

const Problem< T >::intervalsVect_t & roboptim::Problem< T >::boundsVector

(

)

const

Retrieve constraints bounds vector.

Returns

constraints bounds vector

template<typename T >

Problem< T >::intervalsVect_t & roboptim::Problem< T >::boundsVector

(

)

Retrieve constraints bounds vector.

Returns

constraints bounds vector

template<typename T >

void roboptim::Problem< T >::clearConstraints

(

)

Clear the constraints from the problem.

template<typename T >

const Problem< T >::constraints_t & roboptim::Problem< T >::constraints

(

)

const

Retrieve constraints.

Returns

constraints

Referenced by roboptim::ResultAnalyzer< T >::checkNullGradient(), and roboptim::ScalingHelper< T >::ScalingHelper().

template<typename T >

Problem< T >::size_type roboptim::Problem< T >::constraintsOutputSize

(

)

const

Return the output size of the problem's constraints.

template<typename T>

template<int NORM>

value_type roboptim::Problem< T >::constraintsViolation

(

const_argument_ref

x

)

const

Evaluate the constraint violation for a given x.

This takes into account both argument bounds and constraint bounds.

Parameters

x	evaluation point.

Returns

constraint violation at x.

Template Parameters

ALG	Eigen norm used for the reduction, e.g. 1 or Eigen::Infinity.

template<typename T>

template<int NORM>

Problem<T>::value_type roboptim::Problem< T >::constraintsViolation

(

const_argument_ref

x

)

const

template<typename T >

Problem< T >::result_t roboptim::Problem< T >::constraintsViolationVector

(

const_argument_ref

x

)

const

Evaluate the vector of constraints violation for a given x.

This takes into account both argument bounds and constraint bounds. If the output value is lower than the lower bound, the violation is negative. If the output value is higher than the upper bound, the violation is positive. If the output value is within the bounds, the violation is null.

Parameters

x	evaluation point.

Returns

vector of constraint violation at x.

References roboptim::GenericFunction< T >::infinity().

Referenced by roboptim::ResultAnalyzer< T >::checkKKT().

template<typename T >

Problem< T >::size_type roboptim::Problem< T >::differentiableConstraintsOutputSize

(

)

const

Return the output size of the problem's differentiable constraints.

template<typename T >

const Problem< T >::function_t & roboptim::Problem< T >::function

(

)

const

Retrieve cost function.

Returns

cost function

template<typename T >

Problem< T >::jacobian_t roboptim::Problem< T >::jacobian

(

const_argument_ref

x

)

const

Evaluate the Jacobian matrix of the problem for a given x.

Note: this is a helper method, and is not supposed to be used in any critical loop.

Parameters

x	evaluation point.

Returns

Jacobian matrix evaluated at x.

References roboptim::GenericFunction< T >::inputSize().

template<>

Problem< EigenMatrixSparse >::jacobian_t roboptim::Problem< EigenMatrixSparse >::jacobian ( const_argument_ref  x ) const

inline

References roboptim::GenericFunction< T >::inputSize().

template<typename T >

Problem< T >::scaling_t & roboptim::Problem< T >::objectiveScaling

(

)

Retrieve objective scaling.

Objective scaling defines the scaling factors applied to the objective function.

Returns

objective scaling

template<typename T >

const Problem< T >::scaling_t & roboptim::Problem< T >::objectiveScaling

(

)

const

Retrieve objective scaling.

Objective scaling defines the scaling factors applied to the objective function.

Returns

objective scaling

template<typename T >

std::ostream & roboptim::Problem< T >::print

(

std::ostream &

o

)

const

Display the problem on the specified output stream.

Parameters

o	output stream used for display

Returns

output stream

References roboptim::decindent(), roboptim::fg::fail(), roboptim::iendl(), roboptim::incendl(), roboptim::fg::ok(), roboptim::fg::reset(), and roboptim::fg::warn().

template<typename T >

Problem< T >::jacobian_t roboptim::Problem< T >::scaledJacobian

(

const_argument_ref

x

)

const

Evaluate the scaled Jacobian matrix of the problem for a given x.

Note: this is a helper method, and is not supposed to be used in any critical loop. Both constraint and argument scaling parameters are applied.

Parameters

x	evaluation point.

Returns

scaled Jacobian matrix evaluated at x.

template<typename T >

const Problem< T >::scalingVect_t & roboptim::Problem< T >::scalesVector

(

)

const

Retrieve constraints scaling vector (deprecated version).

Deprecated:

Deprecated since version 3.1. Use scalingVector() instead.

template<typename T >

const Problem< T >::scalingVect_t & roboptim::Problem< T >::scalingVector

(

)

const

Retrieve constraints scaling vector.

Returns

constraints scaling vector

template<typename T >

Problem< T >::startingPoint_t & roboptim::Problem< T >::startingPoint

(

)

Set the initial guess.

Returns

reference on the initial guess

Exceptions

std::runtime_error

template<typename T >

const Problem< T >::startingPoint_t & roboptim::Problem< T >::startingPoint

(

)

const

Get the initial guess.

Returns

reference on the initial guess

Exceptions

std::runtime_error