Base Class for quasi-Newton acceleration schemes. More...

#include <BaseQNAcceleration.hpp>

Inheritance diagram for precice::acceleration::BaseQNAcceleration:

Collaboration diagram for precice::acceleration::BaseQNAcceleration:

Public Member Functions
	BaseQNAcceleration (double initialRelaxation, bool forceInitialRelaxation, int maxIterationsUsed, int timeWindowsReused, int filter, double singularityLimit, std::vector< int > dataIDs, impl::PtrPreconditioner preconditioner, bool reducedTimeGrid)

	~BaseQNAcceleration () override
	Destructor, empty.

std::vector< int >	getPrimaryDataIDs () const final override
	Returns all IQN involved data IDs.

void	initialize (const DataMap &cplData) final override
	Initializes the acceleration.

void	performAcceleration (DataMap &cplData, double windowStart, double windowEnd) final override
	Performs one acceleration step.

void	iterationsConverged (const DataMap &cplData, double windowStart) final override
	Marks a iteration sequence as converged.

void	exportState (io::TXTWriter &writer) final override
	Exports the current state of the acceleration to a file.

void	importState (io::TXTReader &reader) final override
	Imports the last exported state of the acceleration from file.

int	getDeletedColumns () const
	how many QN columns were deleted in this time window

int	getDroppedColumns () const
	how many QN columns were dropped (went out of scope) in this time window

int	getLSSystemCols () const
	: computes number of cols in least squares system, i.e, number of cols in _matrixV, _matrixW, _qrV, etc.. This is only necessary if some procs do not have any nodes on the coupling interface. In this case, the matrices are not constructed and we have no information about the number of cols. This info is needed for intra-participant communication. Number of its =! _cols in general.

int	getMaxUsedIterations () const
	Get the maximum number of reused iterations.

int	getMaxUsedTimeWindows () const
	Get the maximum number of reused time windows.

Public Member Functions inherited from precice::acceleration::Acceleration
virtual	~Acceleration ()=default

Protected Member Functions
int	getLSSystemRows () const

int	getPrimaryLSSystemRows () const

virtual void	specializedIterationsConverged (const DataMap &cplData)=0
	Marks a iteration sequence as converged.

virtual void	updateDifferenceMatrices (const DataMap &cplData)
	Updates the V, W matrices (as well as the matrices for the secondary data)

virtual void	updateCouplingData (const DataMap &cplData, double windowStart)
	Splits up QN system vector back into the waveforms in coupling data.

virtual void	applyFilter ()
	Applies the filter method for the least-squares system, defined in the configuration.

virtual void	computeQNUpdate (Eigen::VectorXd &xUpdate)=0
	Computes the quasi-Newton update using the specified pp scheme (IQNIMVJ, IQNILS)

virtual void	removeMatrixColumn (int columnIndex)
	Removes one iteration from V,W matrices and adapts _matrixCols.

void	writeInfo (const std::string &s, bool allProcs=false)
	Writes info to the _infostream (also in parallel)

Protected Member Functions inherited from precice::acceleration::Acceleration
void	checkDataIDs (const DataMap &cplData) const
	Checks if all dataIDs are contained in cplData.

Protected Attributes
logging::Logger	_log {"acceleration::BaseQNAcceleration"}

impl::PtrPreconditioner	_preconditioner
	Preconditioner for least-squares system if vectorial system is used.

const double	_initialRelaxation
	Constant relaxation factor used for first iteration.

const int	_maxIterationsUsed
	Maximum number of old data iterations kept.

const int	_timeWindowsReused
	Maximum number of old time windows (with data values) kept.

const std::vector< int >	_primaryDataIDs
	Data IDs of primary data to be involved in the IQN coefficient computation.

std::vector< int >	_dataIDs
	Data IDs of all coupling data.

bool	_firstIteration = true
	Indicates the first iteration, where constant relaxation is used.

bool	_firstTimeWindow = true

bool	_hasNodesOnInterface = true

bool	_forceInitialRelaxation

bool	_resetLS = false
	If true, the LS system has been modified (reset or recomputed) in such a way, that mere updating of matrices _Wtil, Q, R etc.. is not feasible any more and need to be recomputed.

Eigen::VectorXd	_oldPrimaryXTilde
	Solver output regarding primary data from last iteration.

Eigen::VectorXd	_oldXTilde
	Solver output from last iteration.

Eigen::VectorXd	_primaryResiduals
	Current iteration residuals of primary IQN data. Temporary.

Eigen::VectorXd	_residuals
	Current iteration residuals of IQN data. Temporary.

Eigen::MatrixXd	_matrixV
	Stores residual deltas.

Eigen::MatrixXd	_matrixW
	Stores x tilde deltas, where x tilde are values computed by solvers.

const bool	_reducedTimeGrid
	if _reducedTimeGrid = false uses the full QN-WI and if _reducedTimeGrid = true uses rQN-WI form the paper https://onlinelibrary.wiley.com/doi/10.1002/nme.6443

impl::QRFactorization	_qrV
	Stores the current QR decomposition ov _matrixV, can be updated via deletion/insertion of columns.

const int	_filter
	filter method that is used to maintain good conditioning of the least-squares system Either of two types: QR1FILTER or QR2Filter

const double	_singularityLimit
	Determines sensitivity when two matrix columns are considered equal.

std::deque< int >	_matrixCols
	Indices (of columns in W, V matrices) of 1st iterations of time windows.

std::vector< int >	_dimOffsets
	Stores the local dimensions, i.e., the offsets in _invJacobian for all processors.

std::vector< int >	_dimOffsetsPrimary
	Stores the local dimensions regarding primary data,.

std::ostringstream	_infostringstream
	write some debug/acceleration info to file

std::fstream	_infostream

int	its = 0

int	tWindows = 0

Private Member Functions
void	initializeVectorsAndPreconditioner (const DataMap &cplData, double windowStart)
	Initializes the vectors, matrices and preconditioner This has to be done after the first iteration of the first time window, since everything in the QN-algorithm is sampled to the timegrid of the first waveform.

virtual void	specializedInitializeVectorsAndPreconditioner (const DataMap &cplData)=0
	handles the initialization of matrices and vectors in the sub-classes

void	concatenateCouplingData (Eigen::VectorXd &data, Eigen::VectorXd &oldData, const DataMap &cplData, std::vector< int > dataIDs, precice::time::TimeGrids timeGrids, double windowStart) const
	Samples and concatenates the data and old data in cplData into a long vector.

Private Attributes
std::optional< time::TimeGrids >	_timeGrids
	Stores the time grids to which the primary and secondary data involved in the QN system will be interpolated to.

std::optional< time::TimeGrids >	_primaryTimeGrids
	Stores the time grids to which the primary data involved in the QN system will be interpolated to. If _reducedTimeGrids is true then this will only contain the last time stamp of the time window, see https://doi.org/10.1002/nme.6443.

Eigen::VectorXd	_primaryValues
	Concatenation of all primary data involved in the QN system.

Eigen::VectorXd	_oldPrimaryValues
	Concatenation of all old primary data involved in the QN system.

Eigen::VectorXd	_oldPrimaryResiduals
	Difference between solver input and output from last time window regarding primary data.

Eigen::VectorXd	_values
	Concatenation of all primary and secondary data involved in the QN system.

Eigen::VectorXd	_oldValues
	Concatenation of all old primary and secondary data involved in the QN system.

Eigen::MatrixXd	_matrixVBackup
	backup of the V,W and matrixCols data structures. Needed for the skipping of initial relaxation, if previous time window converged within one iteration i.e., V and W are empty – in this case restore V and W with time window t-2.

Eigen::MatrixXd	_matrixWBackup

std::deque< int >	_matrixColsBackup

int	_nbDelCols = 0
	Number of filtered out columns in this time window.

int	_nbDropCols = 0
	Number of dropped columns in this time window (old time window out of scope)

Additional Inherited Members
Public Types inherited from precice::acceleration::Acceleration
using	DataMap = std::map<int, cplscheme::PtrCouplingData>
	Map from data ID to data values.

Static Public Attributes inherited from precice::acceleration::Acceleration
static const int	NOFILTER = 0

static const int	QR1FILTER = 1

static const int	QR1FILTER_ABS = 2

static const int	QR2FILTER = 3

static const int	PODFILTER = 4

static const int	QR3FILTER = 5

Static Protected Member Functions inherited from precice::acceleration::Acceleration
static void	applyRelaxation (double omega, DataMap &cplData, double windowStart)
	performs a relaxation given a relaxation factor omega

Detailed Description

Base Class for quasi-Newton acceleration schemes.

Definition at line 64 of file BaseQNAcceleration.hpp.

Constructor & Destructor Documentation

◆ BaseQNAcceleration()

precice::acceleration::BaseQNAcceleration::BaseQNAcceleration	(	double	initialRelaxation,
		bool	forceInitialRelaxation,
		int	maxIterationsUsed,
		int	timeWindowsReused,
		int	filter,
		double	singularityLimit,
		std::vector< int >	dataIDs,
		impl::PtrPreconditioner	preconditioner,
		bool	reducedTimeGrid )

Definition at line 34 of file BaseQNAcceleration.cpp.

◆ ~BaseQNAcceleration()

precice::acceleration::BaseQNAcceleration::~BaseQNAcceleration ( )

inlineoverride

Destructor, empty.

Definition at line 80 of file BaseQNAcceleration.hpp.

Member Function Documentation

◆ applyFilter()

void precice::acceleration::BaseQNAcceleration::applyFilter ( )

protectedvirtual

Applies the filter method for the least-squares system, defined in the configuration.

Definition at line 364 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ computeQNUpdate()

virtual void precice::acceleration::BaseQNAcceleration::computeQNUpdate ( Eigen::VectorXd & xUpdate )

protectedpure virtual

Computes the quasi-Newton update using the specified pp scheme (IQNIMVJ, IQNILS)

Implemented in precice::acceleration::IQNILSAcceleration, and precice::acceleration::IQNIMVJAcceleration.

◆ concatenateCouplingData()

void precice::acceleration::BaseQNAcceleration::concatenateCouplingData	(	Eigen::VectorXd &	data,
		Eigen::VectorXd &	oldData,
		const DataMap &	cplData,
		std::vector< int >	dataIDs,
		precice::time::TimeGrids	timeGrids,
		double	windowStart ) const

private

Samples and concatenates the data and old data in cplData into a long vector.

Definition at line 633 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ exportState()

void precice::acceleration::BaseQNAcceleration::exportState ( io::TXTWriter & writer )

finaloverridevirtual

Exports the current state of the acceleration to a file.

Reimplemented from precice::acceleration::Acceleration.

Definition at line 554 of file BaseQNAcceleration.cpp.

◆ getDeletedColumns()

int precice::acceleration::BaseQNAcceleration::getDeletedColumns ( ) const

how many QN columns were deleted in this time window

Definition at line 564 of file BaseQNAcceleration.cpp.

◆ getDroppedColumns()

int precice::acceleration::BaseQNAcceleration::getDroppedColumns ( ) const

how many QN columns were dropped (went out of scope) in this time window

Definition at line 569 of file BaseQNAcceleration.cpp.

◆ getLSSystemCols()

int precice::acceleration::BaseQNAcceleration::getLSSystemCols ( ) const

: computes number of cols in least squares system, i.e, number of cols in _matrixV, _matrixW, _qrV, etc.. This is only necessary if some procs do not have any nodes on the coupling interface. In this case, the matrices are not constructed and we have no information about the number of cols. This info is needed for intra-participant communication. Number of its =! _cols in general.

Definition at line 574 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ getLSSystemRows()

int precice::acceleration::BaseQNAcceleration::getLSSystemRows ( ) const

protected

Definition at line 598 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ getMaxUsedIterations()

int precice::acceleration::BaseQNAcceleration::getMaxUsedIterations ( ) const

Get the maximum number of reused iterations.

Definition at line 588 of file BaseQNAcceleration.cpp.

◆ getMaxUsedTimeWindows()

int precice::acceleration::BaseQNAcceleration::getMaxUsedTimeWindows ( ) const

Get the maximum number of reused time windows.

Definition at line 593 of file BaseQNAcceleration.cpp.

◆ getPrimaryDataIDs()

std::vector< int > precice::acceleration::BaseQNAcceleration::getPrimaryDataIDs ( ) const

inlinefinaloverridevirtual

Returns all IQN involved data IDs.

Implements precice::acceleration::Acceleration.

Definition at line 93 of file BaseQNAcceleration.hpp.

◆ getPrimaryLSSystemRows()

int precice::acceleration::BaseQNAcceleration::getPrimaryLSSystemRows ( ) const

protected

Definition at line 606 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ importState()

void precice::acceleration::BaseQNAcceleration::importState ( io::TXTReader & reader )

finaloverridevirtual

Imports the last exported state of the acceleration from file.

Is empty at the moment!!!

Reimplemented from precice::acceleration::Acceleration.

Definition at line 559 of file BaseQNAcceleration.cpp.

◆ initialize()

void precice::acceleration::BaseQNAcceleration::initialize ( const DataMap & cplData )

finaloverridevirtual

Initializes the acceleration.

initialize()

@brief: Initializes all the needed variables and data

Implements precice::acceleration::Acceleration.

Definition at line 79 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ initializeVectorsAndPreconditioner()

void precice::acceleration::BaseQNAcceleration::initializeVectorsAndPreconditioner	(	const DataMap &	cplData,
		double	windowStart )

private

Initializes the vectors, matrices and preconditioner This has to be done after the first iteration of the first time window, since everything in the QN-algorithm is sampled to the timegrid of the first waveform.

make dimensions public to all procs, last entry _dimOffsets[IntraComm::getSize()] holds the global dimension, global,n

provide vertex offset information for all processors mesh->getVertexOffsets() provides an array that stores the number of mesh vertices on each processor This information needs to be gathered for all meshes. To get the number of respective unknowns of a specific processor we need to multiply the number of vertices with the dimensionality of the vector-valued data for each coupling data.

Definition at line 657 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ iterationsConverged()

void precice::acceleration::BaseQNAcceleration::iterationsConverged	(	const DataMap &	cplData,
		double	windowStart )

finaloverridevirtual

Marks a iteration sequence as converged.

: Is called when the convergence criterion for the coupling is fulfilled and finalizes the quasi Newton acceleration. Stores new differences in F and C, clears or

Since convergence measurements are done outside the acceleration, this method has to be used to signalize convergence to the acceleration.

iterationsConverged()

updates F and C according to the number of reused time windows

Sample all the data to the corresponding time grid in _timeGrids and concatenate everything into a long vector timeGrids are stored using std::optional, thus the .value() to get the actual object

pending deletion (after first iteration of next time window Using the matrices from the old time window for the first iteration is better than doing underrelaxation as first iteration of every time window

Implements precice::acceleration::Acceleration.

Definition at line 422 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ performAcceleration()

void precice::acceleration::BaseQNAcceleration::performAcceleration	(	DataMap &	cplData,
		double	windowStart,
		double	windowEnd )

finaloverridevirtual

Performs one acceleration step.

Has to be called after every implicit coupling iteration.

performAcceleration()

@brief: performs one iteration of the quasi Newton acceleration.

Sample all the data to the corresponding time grid in _timeGrids and concatenate everything into a long vector timeGrids are stored using std::optional, thus the .value() to get the actual object

update the difference matrices V,W includes: scaling of values computation of residuals appending the difference matrices

=== update and apply preconditioner ===

The preconditioner is only applied to the matrix V and the columns that are inserted into the QR-decomposition of V.

compute quasi-Newton update PRECONDITION: All objects are unscaled, except the matrices within the QR-dec of V. Thus, the pseudo inverse needs to be reverted before using it.

Implements precice::acceleration::Acceleration.

Definition at line 212 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ removeMatrixColumn()

void precice::acceleration::BaseQNAcceleration::removeMatrixColumn ( int columnIndex )

protectedvirtual

Removes one iteration from V,W matrices and adapts _matrixCols.

removeMatrixColumn()

@brief: removes a column from the least squares system, i. e., from the matrices F and C

Reimplemented in precice::acceleration::IQNILSAcceleration, and precice::acceleration::IQNIMVJAcceleration.

Definition at line 526 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ specializedInitializeVectorsAndPreconditioner()

virtual void precice::acceleration::BaseQNAcceleration::specializedInitializeVectorsAndPreconditioner ( const DataMap & cplData )

privatepure virtual

handles the initialization of matrices and vectors in the sub-classes

called by the initializeVectorsAndPreconditioner method in the BaseQNAcceleration class

Implemented in precice::acceleration::IQNILSAcceleration, and precice::acceleration::IQNIMVJAcceleration.

◆ specializedIterationsConverged()

virtual void precice::acceleration::BaseQNAcceleration::specializedIterationsConverged ( const DataMap & cplData )

protectedpure virtual

Marks a iteration sequence as converged.

called by the iterationsConverged() method in the BaseQNAcceleration class handles the acceleration specific action after the convergence of one iteration

Implemented in precice::acceleration::IQNILSAcceleration, and precice::acceleration::IQNIMVJAcceleration.

◆ updateCouplingData()

void precice::acceleration::BaseQNAcceleration::updateCouplingData	(	const DataMap &	cplData,
		double	windowStart )

protectedvirtual

Splits up QN system vector back into the waveforms in coupling data.

Definition at line 386 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ updateDifferenceMatrices()

void precice::acceleration::BaseQNAcceleration::updateDifferenceMatrices ( const DataMap & cplData )

protectedvirtual

Updates the V, W matrices (as well as the matrices for the secondary data)

: computes the current residual and stores it, computes the differences and

updateDifferenceMatrices()

updates the difference matrices F and C.

Reimplemented in precice::acceleration::IQNILSAcceleration, and precice::acceleration::IQNIMVJAcceleration.

Definition at line 114 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

◆ writeInfo()

void precice::acceleration::BaseQNAcceleration::writeInfo	(	const std::string &	s,
		bool	allProcs = false )

protected

Writes info to the _infostream (also in parallel)

Definition at line 614 of file BaseQNAcceleration.cpp.

Here is the call graph for this function:

Member Data Documentation

◆ _dataIDs

std::vector<int> precice::acceleration::BaseQNAcceleration::_dataIDs

protected

Data IDs of all coupling data.

Definition at line 174 of file BaseQNAcceleration.hpp.

◆ _dimOffsets

std::vector<int> precice::acceleration::BaseQNAcceleration::_dimOffsets

protected

Stores the local dimensions, i.e., the offsets in _invJacobian for all processors.

Definition at line 249 of file BaseQNAcceleration.hpp.

◆ _dimOffsetsPrimary

std::vector<int> precice::acceleration::BaseQNAcceleration::_dimOffsetsPrimary

protected

Stores the local dimensions regarding primary data,.

Definition at line 253 of file BaseQNAcceleration.hpp.

◆ _filter

const int precice::acceleration::BaseQNAcceleration::_filter

protected

filter method that is used to maintain good conditioning of the least-squares system Either of two types: QR1FILTER or QR2Filter

Definition at line 227 of file BaseQNAcceleration.hpp.

◆ _firstIteration

bool precice::acceleration::BaseQNAcceleration::_firstIteration = true

protected

Indicates the first iteration, where constant relaxation is used.

Definition at line 177 of file BaseQNAcceleration.hpp.

◆ _firstTimeWindow

bool precice::acceleration::BaseQNAcceleration::_firstTimeWindow = true

protected

Definition at line 182 of file BaseQNAcceleration.hpp.

◆ _forceInitialRelaxation

bool precice::acceleration::BaseQNAcceleration::_forceInitialRelaxation

protected

Definition at line 193 of file BaseQNAcceleration.hpp.

◆ _hasNodesOnInterface

bool precice::acceleration::BaseQNAcceleration::_hasNodesOnInterface = true

protected

Definition at line 187 of file BaseQNAcceleration.hpp.

◆ _infostream

std::fstream precice::acceleration::BaseQNAcceleration::_infostream

protected

Definition at line 257 of file BaseQNAcceleration.hpp.

◆ _infostringstream

std::ostringstream precice::acceleration::BaseQNAcceleration::_infostringstream

protected

write some debug/acceleration info to file

Definition at line 256 of file BaseQNAcceleration.hpp.

◆ _initialRelaxation

const double precice::acceleration::BaseQNAcceleration::_initialRelaxation

protected

Constant relaxation factor used for first iteration.

Definition at line 162 of file BaseQNAcceleration.hpp.

◆ _log

logging::Logger precice::acceleration::BaseQNAcceleration::_log {"acceleration::BaseQNAcceleration"}

protected

Definition at line 156 of file BaseQNAcceleration.hpp.

◆ _matrixCols

std::deque<int> precice::acceleration::BaseQNAcceleration::_matrixCols

protected

Indices (of columns in W, V matrices) of 1st iterations of time windows.

When old time windows are reused (_timeWindowsReused > 0), the indices of the first iteration of each time window needs to be stored, such that, e.g., all iterations of the last time window, or one specific iteration that leads to a singular matrix in the QR decomposition can be removed and tracked.

Definition at line 244 of file BaseQNAcceleration.hpp.

◆ _matrixColsBackup

std::deque<int> precice::acceleration::BaseQNAcceleration::_matrixColsBackup

private

Definition at line 333 of file BaseQNAcceleration.hpp.

◆ _matrixV

Eigen::MatrixXd precice::acceleration::BaseQNAcceleration::_matrixV

protected

Stores residual deltas.

Definition at line 213 of file BaseQNAcceleration.hpp.

◆ _matrixVBackup

Eigen::MatrixXd precice::acceleration::BaseQNAcceleration::_matrixVBackup

private

backup of the V,W and matrixCols data structures. Needed for the skipping of initial relaxation, if previous time window converged within one iteration i.e., V and W are empty – in this case restore V and W with time window t-2.

Definition at line 331 of file BaseQNAcceleration.hpp.

◆ _matrixW

Eigen::MatrixXd precice::acceleration::BaseQNAcceleration::_matrixW

protected

Stores x tilde deltas, where x tilde are values computed by solvers.

Definition at line 216 of file BaseQNAcceleration.hpp.

◆ _matrixWBackup

Eigen::MatrixXd precice::acceleration::BaseQNAcceleration::_matrixWBackup

private

Definition at line 332 of file BaseQNAcceleration.hpp.

◆ _maxIterationsUsed

const int precice::acceleration::BaseQNAcceleration::_maxIterationsUsed

protected

Maximum number of old data iterations kept.

Definition at line 165 of file BaseQNAcceleration.hpp.

◆ _nbDelCols

int precice::acceleration::BaseQNAcceleration::_nbDelCols = 0

private

Number of filtered out columns in this time window.

Definition at line 336 of file BaseQNAcceleration.hpp.

◆ _nbDropCols

int precice::acceleration::BaseQNAcceleration::_nbDropCols = 0

private

Number of dropped columns in this time window (old time window out of scope)

Definition at line 339 of file BaseQNAcceleration.hpp.

◆ _oldPrimaryResiduals

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_oldPrimaryResiduals

private

Difference between solver input and output from last time window regarding primary data.

Definition at line 319 of file BaseQNAcceleration.hpp.

◆ _oldPrimaryValues

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_oldPrimaryValues

private

Concatenation of all old primary data involved in the QN system.

Definition at line 316 of file BaseQNAcceleration.hpp.

◆ _oldPrimaryXTilde

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_oldPrimaryXTilde

protected

Solver output regarding primary data from last iteration.

Definition at line 201 of file BaseQNAcceleration.hpp.

◆ _oldValues

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_oldValues

private

Concatenation of all old primary and secondary data involved in the QN system.

Definition at line 325 of file BaseQNAcceleration.hpp.

◆ _oldXTilde

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_oldXTilde

protected

Solver output from last iteration.

Definition at line 204 of file BaseQNAcceleration.hpp.

◆ _preconditioner

impl::PtrPreconditioner precice::acceleration::BaseQNAcceleration::_preconditioner

protected

Preconditioner for least-squares system if vectorial system is used.

Definition at line 159 of file BaseQNAcceleration.hpp.

◆ _primaryDataIDs

const std::vector<int> precice::acceleration::BaseQNAcceleration::_primaryDataIDs

protected

Data IDs of primary data to be involved in the IQN coefficient computation.

Definition at line 171 of file BaseQNAcceleration.hpp.

◆ _primaryResiduals

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_primaryResiduals

protected

Current iteration residuals of primary IQN data. Temporary.

Definition at line 207 of file BaseQNAcceleration.hpp.

◆ _primaryTimeGrids

std::optional<time::TimeGrids> precice::acceleration::BaseQNAcceleration::_primaryTimeGrids

private

Stores the time grids to which the primary data involved in the QN system will be interpolated to. If _reducedTimeGrids is true then this will only contain the last time stamp of the time window, see https://doi.org/10.1002/nme.6443.

Definition at line 310 of file BaseQNAcceleration.hpp.

◆ _primaryValues

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_primaryValues

private

Concatenation of all primary data involved in the QN system.

Definition at line 313 of file BaseQNAcceleration.hpp.

◆ _qrV

impl::QRFactorization precice::acceleration::BaseQNAcceleration::_qrV

protected

Stores the current QR decomposition ov _matrixV, can be updated via deletion/insertion of columns.

Definition at line 222 of file BaseQNAcceleration.hpp.

◆ _reducedTimeGrid

const bool precice::acceleration::BaseQNAcceleration::_reducedTimeGrid

protected

if _reducedTimeGrid = false uses the full QN-WI and if _reducedTimeGrid = true uses rQN-WI form the paper https://onlinelibrary.wiley.com/doi/10.1002/nme.6443

Definition at line 219 of file BaseQNAcceleration.hpp.

◆ _resetLS

bool precice::acceleration::BaseQNAcceleration::_resetLS = false

protected

If true, the LS system has been modified (reset or recomputed) in such a way, that mere updating of matrices _Wtil, Q, R etc.. is not feasible any more and need to be recomputed.

Definition at line 198 of file BaseQNAcceleration.hpp.

◆ _residuals

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_residuals

protected

Current iteration residuals of IQN data. Temporary.

Definition at line 210 of file BaseQNAcceleration.hpp.

◆ _singularityLimit

const double precice::acceleration::BaseQNAcceleration::_singularityLimit

protected

Determines sensitivity when two matrix columns are considered equal.

When during the QR decomposition of the V matrix a pivot element smaller than the singularity limit is found, the matrix is considered to be singular and the corresponding (older) iteration is removed.

Definition at line 235 of file BaseQNAcceleration.hpp.

◆ _timeGrids

std::optional<time::TimeGrids> precice::acceleration::BaseQNAcceleration::_timeGrids

private

Stores the time grids to which the primary and secondary data involved in the QN system will be interpolated to.

Definition at line 306 of file BaseQNAcceleration.hpp.

◆ _timeWindowsReused

const int precice::acceleration::BaseQNAcceleration::_timeWindowsReused

protected

Maximum number of old time windows (with data values) kept.

Definition at line 168 of file BaseQNAcceleration.hpp.

◆ _values

Eigen::VectorXd precice::acceleration::BaseQNAcceleration::_values

private

Concatenation of all primary and secondary data involved in the QN system.

Definition at line 322 of file BaseQNAcceleration.hpp.

◆ its

int precice::acceleration::BaseQNAcceleration::its = 0

protected

Definition at line 288 of file BaseQNAcceleration.hpp.

◆ tWindows

int precice::acceleration::BaseQNAcceleration::tWindows = 0

protected

Definition at line 288 of file BaseQNAcceleration.hpp.

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

src/acceleration/BaseQNAcceleration.hpp
src/acceleration/BaseQNAcceleration.cpp

Public Member Functions

Protected Member Functions

Protected Attributes

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ BaseQNAcceleration()

◆ ~BaseQNAcceleration()

Member Function Documentation

◆ applyFilter()

◆ computeQNUpdate()

◆ concatenateCouplingData()

◆ exportState()

◆ getDeletedColumns()

◆ getDroppedColumns()

◆ getLSSystemCols()

◆ getLSSystemRows()

◆ getMaxUsedIterations()

◆ getMaxUsedTimeWindows()

◆ getPrimaryDataIDs()

◆ getPrimaryLSSystemRows()

◆ importState()

◆ initialize()

@brief: Initializes all the needed variables and data

◆ initializeVectorsAndPreconditioner()

◆ iterationsConverged()

updates F and C according to the number of reused time windows

◆ performAcceleration()

@brief: performs one iteration of the quasi Newton acceleration.

◆ removeMatrixColumn()

@brief: removes a column from the least squares system, i. e., from the matrices F and C

◆ specializedInitializeVectorsAndPreconditioner()

◆ specializedIterationsConverged()

◆ updateCouplingData()

◆ updateDifferenceMatrices()

updates the difference matrices F and C.

◆ writeInfo()

Member Data Documentation

◆ _dataIDs

◆ _dimOffsets

◆ _dimOffsetsPrimary

◆ _filter

◆ _firstIteration

◆ _firstTimeWindow

◆ _forceInitialRelaxation

◆ _hasNodesOnInterface

◆ _infostream

◆ _infostringstream

◆ _initialRelaxation

◆ _log

◆ _matrixCols

◆ _matrixColsBackup

◆ _matrixV

◆ _matrixVBackup

◆ _matrixW

◆ _matrixWBackup

◆ _maxIterationsUsed

◆ _nbDelCols

◆ _nbDropCols

◆ _oldPrimaryResiduals

◆ _oldPrimaryValues

◆ _oldPrimaryXTilde

◆ _oldValues

◆ _oldXTilde

◆ _preconditioner

◆ _primaryDataIDs

◆ _primaryResiduals

◆ _primaryTimeGrids

◆ _primaryValues

◆ _qrV

◆ _reducedTimeGrid

◆ _resetLS

◆ _residuals

◆ _singularityLimit

◆ _timeGrids

◆ _timeWindowsReused

◆ _values

◆ its