an object that defines a scalar objective function to be optimized. More...

#include <cl_Objective.hpp>

Inheritance diagram for belfem::opt::Objective:

Public Member Functions
	Objective (const uint aDimension)
virtual	~Objective ()=default
virtual real	compute_objective (const Vector< real > &aX, Vector< real > &aGradient)=0
	evaluate the objective at the design point aX.
uint	dimension () const
	return the number of design variables

Detailed Description

an object that defines a scalar objective function to be optimized.

This base class is agnostic of the optimization algorithm: the user derives from it and implements compute_objective(). It plays the same role for the Optimizer that ode::ODE plays for the ode::Integrator.

Constructor & Destructor Documentation

◆ Objective()

belfem::opt::Objective::Objective ( const uint aDimension )

inline

◆ ~Objective()

virtual belfem::opt::Objective::~Objective ( )

virtualdefault

Member Function Documentation

◆ compute_objective()

virtual real belfem::opt::Objective::compute_objective	(	const Vector< real > &	aX,
		Vector< real > &	aGradient )

pure virtual

evaluate the objective at the design point aX.

Parameters

aX	design vector, length dimension()
aGradient	gradient output. For derivative-free algorithms this vector has length zero and must be ignored. For gradient-based algorithms it has length dimension() and the implementation must fill it with d(objective)/d(aX).

Returns: the scalar objective value at aX

Implemented in Rosenbrock.

◆ dimension()

uint belfem::opt::Objective::dimension ( ) const