gsFunctionAdaptor< T > Class Template Reference

Detailed Description

template<typename T>
class gismo::gsFunctionAdaptor< T >

Adaptor to see a given gsFunction as (the objective of) an unconstrained optimization problem.

Inheritance diagram for gsFunctionAdaptor< T >:

Collaboration diagram for gsFunctionAdaptor< T >:

Public Member Functions
virtual void	computeJacStructure ()
	Computes the sparsity pattern of the constraint Jacobian matrix. More...
void	evalCon_into (const gsAsConstVector< T > &u, gsAsVector< T > &result) const
	Returns values of the constraints at design value u.
T	evalObj (const gsAsConstVector< T > &u) const
	Returns the gradient value of the objective function at design value u. More...
void	gradObj_into (const gsAsConstVector< T > &u, gsAsVector< T > &result) const
	Returns the gradient of the objective function at design value u By default it uses finite differences, overriding it should provide exact gradient.
	gsFunctionAdaptor (const gsFunction< T > &obj)
virtual void	hessLagr_into (const gsAsConstVector< T > &, gsAsVector< T > &) const
	Returns Hessian Lagrangian of the constraints at design value.
void	jacobCon_into (const gsAsConstVector< T > &u, gsAsVector< T > &result) const
	Returns Jacobian of the constraints at design value u. Format of result is sparse, complying to m_conJacRows and m_conJacCols.
int	numDesignVars () const
	Callback function is executed after every iteration. Returning false causes premature termination of the optimization.

Protected Attributes
std::vector< index_t >	m_conJacCols
	Constraint Jacobian non-zero entries columns.
std::vector< index_t >	m_conJacRows
	Constraint Jacobian non-zero entries rows.
gsVector< T >	m_conLowerBounds
	Lower bounds for the constraints.
gsVector< T >	m_conUpperBounds
	Upper bounds for the constraints.
gsMatrix< T >	m_curDesign
	Current design variables (and starting point )
gsVector< T >	m_desLowerBounds
	Lower bounds for the design variables.
gsVector< T >	m_desUpperBounds
	Upper bounds for the design variables.
gsMatrix< T >	m_lambda
	Lagrange multipliers (set in the finalize_solution method)
int	m_numConJacNonZero
	Number of nonzero entries in the Constraint Jacobian.
int	m_numConstraints
	Number of constraints.
int	m_numDesignVars
	Number of design variables.

Constructor & Destructor Documentation

gsFunctionAdaptor ( const gsFunction< T > &  obj )

inline

bound_relax_factor

Member Function Documentation

virtual void computeJacStructure ( )

inlinevirtualinherited

Computes the sparsity pattern of the constraint Jacobian matrix.

Number of constraints and design variables need to be set before calling this. By default the constraint Jacobian is set to full. Override this function to provide sparsity.

T evalObj ( const gsAsConstVector< T > &  ) const

inlinevirtual

Returns the gradient value of the objective function at design value u.

default constructor default destructor

Reimplemented from gsOptProblem< T >.