lineSearch Class Reference

Abstract base class for line search methods. More...

Inheritance diagram for lineSearch:

Collaboration diagram for lineSearch:

Public Member Functions
	TypeName ("lineSearch")
	Runtime type information. More...
	declareRunTimeSelectionTable (autoPtr, lineSearch, dictionary,(const dictionary &dict, const Time &time),(dict, time))
	lineSearch (const dictionary &dict, const Time &time)
	Construct from components. More...
virtual	~lineSearch ()=default
	Destructor. More...
virtual void	setDeriv (const scalar deriv)
	Set objective derivative. More...
void	setDirection (const scalarField &direction)
	Set direction. More...
void	setNewMeritValue (const scalar value)
	Set new objective value. More...
void	setOldMeritValue (const scalar value)
	Set old objective value. More...
virtual void	reset ()
	Reset step to initial value. More...
label	maxIters () const
	Get max number of iterations. More...
scalar	step () const
	Get current step. More...
virtual bool	converged ()=0
	Return the correction of the design variables. More...
virtual void	updateStep ()=0
	Update the line search step based on the specific line search strategy, e.g. bisection, quadratic fit, etc. More...
virtual void	updateStep (const scalar newStep)
	Update the step using the supplied value. More...
virtual lineSearch &	operator++ ()
	Increment iteration number and store merit value corresponding to the previous optimisation cycle. More...
virtual lineSearch &	operator++ (int)
	Postfix increment. Necessary for compilation. More...

Static Public Member Functions
static autoPtr< lineSearch >	New (const dictionary &dict, const Time &time)
	Return a reference to the selected turbulence model. More...

Protected Member Functions
const dictionary &	coeffsDict ()
	Optional coeffs dict. More...

Protected Attributes
const dictionary	dict_
	Subdict within updateMethod. More...
IOdictionary	lineSearchDict_
	IOdictionary under time/uniform for continuation. More...
scalar	directionalDeriv_
	Directional derivative of the merit function. More...
scalarField	direction_
	Update direction. More...
scalar	oldMeritValue_
	Old merit value from this opt cycle. More...
scalar	newMeritValue_
	New merit value from this opt cycle. More...
scalar	prevMeritDeriv_
	Merit directional deriv from the previous opt cycle. More...
scalar	initialStep_
	Correction multiplier at the first step of line search. More...
scalar	minStep_
	Minimum allowed correction multiplier. More...
scalar	step_
	Correction multiplier. More...
label	iter_
	Inner line search iteration. More...
label	maxIters_
	Maximum line search iterations. More...
bool	extrapolateInitialStep_
	Whether to extrapolate the correction multiplier for this optimisation cycle based on the previous ones. More...
autoPtr< stepUpdate >	stepUpdate_
	Mechanism to update method if line search conditions are not set. More...

Detailed Description

Abstract base class for line search methods.

Source files

• lineSearch.H
• lineSearch.C

Definition at line 52 of file lineSearch.H.

Constructor & Destructor Documentation

lineSearch()

lineSearch	(	const dictionary &	dict,
		const Time &	time
	)

Construct from components.

Definition at line 45 of file lineSearch.C.

~lineSearch()

virtual ~lineSearch ( )

virtualdefault

Destructor.

Member Function Documentation

coeffsDict()

const Foam::dictionary & coeffsDict ( )

protected

Optional coeffs dict.

Definition at line 37 of file lineSearch.C.

References lineSearch::dict_, dictionary::optionalSubDict(), and Foam::type().

Here is the call graph for this function:

TypeName()

TypeName

(

"lineSearch"

)

Runtime type information.

declareRunTimeSelectionTable()

declareRunTimeSelectionTable	(	autoPtr	,
		lineSearch	,
		dictionary	,
		(const dictionary &dict, const Time &time)	,
		(dict, time)
	)

New()

Foam::autoPtr< Foam::lineSearch > New ( const dictionary &  dict, const Time &  time  )

static

Return a reference to the selected turbulence model.

Definition at line 89 of file lineSearch.C.

References dict, Foam::endl(), Foam::exit(), Foam::FatalIOError, FatalIOErrorInLookup, dictionary::getOrDefault(), Foam::Info, and autoPtr< T >::reset().

Here is the call graph for this function:

setDeriv()

void setDeriv ( const scalar  deriv )

virtual

Set objective derivative.

Definition at line 129 of file lineSearch.C.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the caller graph for this function:

setDirection()

void setDirection

(

const scalarField &

direction

)

Set direction.

Definition at line 136 of file lineSearch.C.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the caller graph for this function:

setNewMeritValue()

void setNewMeritValue

(

const scalar

value

)

Set new objective value.

Definition at line 142 of file lineSearch.C.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the caller graph for this function:

setOldMeritValue()

void setOldMeritValue

(

const scalar

value

)

Set old objective value.

Definition at line 149 of file lineSearch.C.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the caller graph for this function:

reset()

void reset ( )

virtual

Reset step to initial value.

Definition at line 156 of file lineSearch.C.

References Foam::clamp(), Foam::endl(), and Foam::Info.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the call graph for this function:

Here is the caller graph for this function:

maxIters()

Foam::label maxIters

(

)

const

Get max number of iterations.

Definition at line 177 of file lineSearch.C.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the caller graph for this function:

step()

Foam::scalar step

(

)

const

Get current step.

Definition at line 183 of file lineSearch.C.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the caller graph for this function:

converged()

virtual bool converged ( )

pure virtual

Return the correction of the design variables.

Implemented in ArmijoConditions.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the caller graph for this function:

updateStep() [1/2]

virtual void updateStep ( )

pure virtual

Update the line search step based on the specific line search strategy, e.g. bisection, quadratic fit, etc.

Implemented in ArmijoConditions.

Referenced by steadyOptimisation::lineSearchUpdate().

Here is the caller graph for this function:

updateStep() [2/2]

void updateStep ( const scalar  newStep )

virtual

Update the step using the supplied value.

Definition at line 189 of file lineSearch.C.

operator++() [1/2]

Foam::lineSearch & operator++ ( )

virtual

Increment iteration number and store merit value corresponding to the previous optimisation cycle.

Definition at line 195 of file lineSearch.C.

References IOstreamOption::ASCII.

operator++() [2/2]

Foam::lineSearch & operator++ ( int  )

virtual

Postfix increment. Necessary for compilation.

Definition at line 211 of file lineSearch.C.

Member Data Documentation

dict_

const dictionary dict_

protected

Subdict within updateMethod.

Definition at line 61 of file lineSearch.H.

Referenced by lineSearch::coeffsDict().

lineSearchDict_

IOdictionary lineSearchDict_

protected

IOdictionary under time/uniform for continuation.

Definition at line 66 of file lineSearch.H.

directionalDeriv_

scalar directionalDeriv_

protected

Directional derivative of the merit function.

Definition at line 71 of file lineSearch.H.

Referenced by ArmijoConditions::converged().

direction_

scalarField direction_

protected

Update direction.

Definition at line 76 of file lineSearch.H.

oldMeritValue_

scalar oldMeritValue_

protected

Old merit value from this opt cycle.

Definition at line 81 of file lineSearch.H.

Referenced by ArmijoConditions::converged().

newMeritValue_

scalar newMeritValue_

protected

New merit value from this opt cycle.

Definition at line 86 of file lineSearch.H.

Referenced by ArmijoConditions::converged().

prevMeritDeriv_

scalar prevMeritDeriv_

protected

Merit directional deriv from the previous opt cycle.

Definition at line 91 of file lineSearch.H.

initialStep_

scalar initialStep_

protected

Correction multiplier at the first step of line search.

Definition at line 96 of file lineSearch.H.

minStep_

scalar minStep_

protected

Minimum allowed correction multiplier.

Definition at line 101 of file lineSearch.H.

step_

scalar step_

protected

Correction multiplier.

Definition at line 106 of file lineSearch.H.

Referenced by ArmijoConditions::converged().

iter_

label iter_

protected

Inner line search iteration.

Definition at line 111 of file lineSearch.H.

maxIters_

label maxIters_

protected

Maximum line search iterations.

Definition at line 116 of file lineSearch.H.

extrapolateInitialStep_

bool extrapolateInitialStep_

protected

Whether to extrapolate the correction multiplier for this optimisation cycle based on the previous ones.

Useful for non-quasi Newton methods

Definition at line 124 of file lineSearch.H.

stepUpdate_

autoPtr<stepUpdate> stepUpdate_

protected

Mechanism to update method if line search conditions are not set.

Definition at line 129 of file lineSearch.H.

---

The documentation for this class was generated from the following files:

• src/optimisation/adjointOptimisation/adjoint/optimisation/lineSearch/lineSearch/lineSearch.H
• src/optimisation/adjointOptimisation/adjoint/optimisation/lineSearch/lineSearch/lineSearch.C