Optimization problems

Optimization problem. More...

Classes
class	roboptim::Problem< T >
class	roboptim::ResultWithWarnings
	Warnings have been merged to Result to simplify the API. More...
class	roboptim::Result
	Represents the solution of an optimization problem. More...
class	roboptim::SolverError
	Base exception class for solving errors. More...
class	roboptim::SolverFactory< S >
	Define a solver factory that instanciate the plug-ins. More...
struct	roboptim::StateParameter< F >
	Solver state parameters type. More...
class	roboptim::SolverState< P >
	State of the solver. More...
class	roboptim::SolverWarning
	Exception used for non-critical errors during optimization. More...
struct	roboptim::Parameter
	Solver parameter type. More...
class	roboptim::Solver< T >
	Solver for a specific problem class. More...

Functions
template<typename F >
std::ostream &	roboptim::operator<< (std::ostream &o, const StateParameter< F > &parameter)
	Override operator<< to display ``parameters'' objects. More...
template<typename P >
std::ostream &	roboptim::operator<< (std::ostream &o, const SolverState< P > &state)
	Override operator<< to display ``parameters'' objects. More...
ROBOPTIM_CORE_DLLAPI std::ostream &	roboptim::operator<< (std::ostream &o, const SolverWarning &w)
	Override operator<< to handle warning display. More...
ROBOPTIM_CORE_DLLAPI std::ostream &	roboptim::operator<< (std::ostream &o, const Parameter &parameter)
	Override operator<< to display ``parameters'' objects. More...

Detailed Description

Optimization problem.

An optimization problem is defined as:

• a cost function ( \(\mathbb{R}^n \rightarrow \mathbb{R}\))
• one or more constraints functions, ( \(\mathbb{R}^n \rightarrow \mathbb{R}^m\)) associated with an interval and a scale,
• a set of intervals and scale factors for arguments.

The goal of the optimization process is finding a point which minimizes the cost function and which respects the constraints (i.e. the result of some functions is inside of specific interval).

To use the class, one has to instantiate a problem with a reference to a cost function. Then, constraints can be added through the addConstraint method: a shared pointer to a function and an interval is needed.

The cost function is immutable.

Constraints are stored in a vector of smart pointers to function objects.

To add a constraint, use the following syntax:

problem.addConstraint (boost::make_shared<MyFunction> (...), ...);

where MyFunction is the function type.

Unlike other classes which just copy functions, pointers are used here in order to allow sub-classes of constraints to be inserted in the problem. For instance, a twice-differentiable function can be inserted in a problem which expects a differentiable function.

Template Parameters

T	matrix type

Function Documentation

std::ostream & roboptim::operator<<	(	std::ostream &	o,
		const SolverWarning &	w
	)

Override operator<< to handle warning display.

Parameters

o	output stream used for display
w	warning to be displayed

Returns

output stream

References roboptim::SolverWarning::print().

std::ostream & roboptim::operator<<	(	std::ostream &	o,
		const Parameter &	parameter
	)

Override operator<< to display ``parameters'' objects.

Parameters

o	output stream used for display
parameter	parameter

Returns

output stream

References roboptim::Parameter::description, and roboptim::Parameter::value.

template<typename F >

std::ostream & roboptim::operator<<	(	std::ostream &	o,
		const StateParameter< F > &	parameter
	)

Override operator<< to display ``parameters'' objects.

Parameters

o	output stream used for display
parameter	parameter to display

Template Parameters

F	function type.

Returns

output stream

template<typename P >

std::ostream & roboptim::operator<<	(	std::ostream &	o,
		const SolverState< P > &	state
	)

Override operator<< to display ``parameters'' objects.

Parameters

o	output stream used for display
state	solver state to display

Template Parameters

P	problem type.

Returns

output stream