This base class is an abstraction/implementation for numerical method solving line search optimization problem. More...

#include <linesearch.h>

Inheritance diagram for oofem::LineSearchNM:

Collaboration diagram for oofem::LineSearchNM:

Public Types
enum	LS_status { ls_ok, ls_failed }

Public Member Functions
	LineSearchNM (Domain d, EngngModel m)
	Constructor. More...

virtual NM_Status	solve (FloatArray &r, FloatArray &dr, FloatArray &F, FloatArray &R, FloatArray R0, IntArray &eqnmask, double lambda, double &etaValue, LS_status &status, TimeStep tStep)
	Solves the line search optimization problem in the form of $g(r)=0; r_{new}=r_{old}+\eta\delta r; 0 < \eta < 1$ , The aim is to find $\eta$ so that the has decreased sufficiently. More...

virtual IRResultType	initializeFrom (InputRecord *ir)

virtual const char *	giveClassName () const

Public Member Functions inherited from oofem::NumericalMethod
	NumericalMethod (Domain d, EngngModel m)
	Constructor. More...

virtual	~NumericalMethod ()
	Destructor. More...

EngngModel *	giveEngngModel ()

virtual void	reinitialize ()
	Reinitializes the receiver. More...

virtual void	setDomain (Domain *d)

virtual contextIOResultType	saveContext (DataStream &stream, ContextMode mode, void *obj=NULL)

virtual contextIOResultType	restoreContext (DataStream &stream, ContextMode mode, void *obj=NULL)

Public Attributes
int	max_iter

double	ls_tolerance

double	amplifFactor

double	maxEta

double	minEta

FloatArray	eta

FloatArray	prod

Protected Member Functions
void	search (int istep, FloatArray &prod, FloatArray &eta, double amp, double maxeta, double mineta, int &status)

Additional Inherited Members
Protected Attributes inherited from oofem::NumericalMethod
Domain *	domain
	Pointer to domain. More...

EngngModel *	engngModel
	Pointer to engineering model. More...

Detailed Description

This base class is an abstraction/implementation for numerical method solving line search optimization problem.

Definition at line 60 of file linesearch.h.

Member Enumeration Documentation

enum oofem::LineSearchNM::LS_status

Enumerator
ls_ok
ls_failed

Definition at line 63 of file linesearch.h.

Constructor & Destructor Documentation

oofem::LineSearchNM::LineSearchNM	(	Domain *	d,
		EngngModel *	m
	)

Constructor.

Definition at line 44 of file linesearch.C.

References amplifFactor, ls_tolerance, max_iter, maxEta, and minEta.

Member Function Documentation

virtual const char* oofem::LineSearchNM::giveClassName ( ) const

inlinevirtual

Definition at line 95 of file linesearch.h.

IRResultType oofem::LineSearchNM::initializeFrom ( InputRecord * ir )

virtual

Reimplemented from oofem::NumericalMethod.

Definition at line 247 of file linesearch.C.

References _IFT_LineSearchNM_lsearchamp, _IFT_LineSearchNM_lsearchmaxeta, _IFT_LineSearchNM_lsearchtol, amplifFactor, IR_GIVE_OPTIONAL_FIELD, oofem::IRRT_OK, ls_tolerance, and maxEta.

Referenced by oofem::NRSolver::initializeFrom().

void oofem::LineSearchNM::search	(	int	istep,
		FloatArray &	prod,
		FloatArray &	eta,
		double	amp,
		double	maxeta,
		double	mineta,
		int &	status
	)

protected

Definition at line 160 of file linesearch.C.

References oofem::FloatArray::at(), and oofem::max().

Referenced by solve().

NM_Status oofem::LineSearchNM::solve	(	FloatArray &	r,
		FloatArray &	dr,
		FloatArray &	F,
		FloatArray &	R,
		FloatArray *	R0,
		IntArray &	eqnmask,
		double	lambda,
		double &	etaValue,
		LS_status &	status,
		TimeStep *	tStep
	)

virtual

Solves the line search optimization problem in the form of $g(r)=0; r_{new}=r_{old}+\eta\delta r; 0 < \eta < 1$ , The aim is to find $\eta$ so that the $ g(r) $ has decreased sufficiently.

The total solution vector is updated at exit as well as InternalRhs vector.

Parameters

r	Old total solution.
dr	Increment of solution.
F	Old InternalRhs (real internal forces).
R	Reference incremental Rhs (incremental load).
R0	Initial Rhs (initial load).
eqnmask	Equation numbers to mask out from computations.
lambda	Scaling of R.
etaValue	Reached eta value.
status	Linesearch status
tStep	Time step.