#include <SVD.h>

Inheritance diagram for CAROM::SVD:

Public Member Functions
	SVD (Options options)
	Constructor. More...

	~SVD ()

virtual bool	takeSample (double *u_in, double time, bool add_without_increase)=0
	Collect the new sample, u_in at supplied time. More...

int	getDim () const
	Returns the dimension of the system on this processor. More...

virtual const Matrix *	getSpatialBasis ()=0
	Returns the basis vectors for the current time interval. More...

virtual const Matrix *	getTemporalBasis ()=0
	Returns the temporal basis vectors for the current time interval. More...

virtual const Vector *	getSingularValues ()=0
	Returns the singular values for the current time interval. More...

virtual const Matrix *	getSnapshotMatrix ()=0
	Returns the singular values for the current time interval. More...

int	getNumBasisTimeIntervals () const
	Returns the number of time intervals on which different sets of basis vectors are defined. More...

double	getBasisIntervalStartTime (int which_interval) const
	Returns the start time for the requested time interval. More...

bool	isNewTimeInterval () const
	Returns true if the next sample will result in a new time interval. More...

void	increaseTimeInterval ()
	Increase the number of time intervals by one.

int	getNumSamples () const
	Get the number of samples taken.

Protected Attributes
int	d_dim
	Dimension of the system.

int	d_num_samples
	Number of samples stored for the current time interval.

int	d_num_rows_of_W
	Number of rows in right singular matrix.

const int	d_samples_per_time_interval
	The maximum number of samples to be collected for a time interval.

const int	d_max_time_intervals
	The maximum number of time intervals.

Matrix *	d_basis
	The globalized basis vectors for the current time interval. More...

Matrix *	d_basis_right
	The globalized right basis vectors for the current time interval. More...

Matrix *	d_U
	The matrix U which is large. More...

Matrix *	d_W
	The matrix U which is large. More...

Vector *	d_S
	The vector S which is small. More...

Matrix *	d_snapshots
	The globalized snapshot vectors for the current time interval. More...

std::vector< double >	d_time_interval_start_times
	The simulation time at which each time interval starts.

bool	d_debug_algorithm
	Flag to indicate if results of algorithm should be printed for debugging purposes.

Detailed Description

Class SVD defines the interface to the generic SVD algorithm. The API is intentionally small. One may collect the samples, compute the SVD, and get the dimension of the system.

Constructor & Destructor Documentation

CAROM::SVD::SVD ( Options options )

Constructor.

Parameters

[in] options The struct containing the options for this abstract SVD class.

See Also: Options

CAROM::SVD::~SVD ( )

Destructor.

Member Function Documentation

double CAROM::SVD::getBasisIntervalStartTime ( int which_interval ) const

inline

Returns the start time for the requested time interval.

Precondition: 0 <= which_interval; which_interval < getNumBasisTimeIntervals()

Parameters

[in] which_interval The time interval of interest.

Returns: The start time for the requested time interval.

int CAROM::SVD::getDim ( ) const

inline

Returns the dimension of the system on this processor.

Returns: The dimension of the system on this processor.

int CAROM::SVD::getNumBasisTimeIntervals ( ) const

inline

Returns the number of time intervals on which different sets of basis vectors are defined.

Returns: The number of time intervals on which there are basis vectors.

virtual const Vector* CAROM::SVD::getSingularValues ( )

pure virtual

Returns the singular values for the current time interval.

Returns: The singular values for the current time interval.

Implemented in CAROM::IncrementalSVD, and CAROM::StaticSVD.

virtual const Matrix* CAROM::SVD::getSnapshotMatrix ( )

pure virtual

Returns the singular values for the current time interval.

Returns: The singular values for the current time interval.

Implemented in CAROM::IncrementalSVD, and CAROM::StaticSVD.

virtual const Matrix* CAROM::SVD::getSpatialBasis ( )

pure virtual

Returns the basis vectors for the current time interval.

Returns: The basis vectors for the current time interval.

Implemented in CAROM::IncrementalSVD, and CAROM::StaticSVD.

virtual const Matrix* CAROM::SVD::getTemporalBasis ( )

pure virtual

Returns the temporal basis vectors for the current time interval.

Returns: The temporal basis vectors for the current time interval.

Implemented in CAROM::IncrementalSVD, and CAROM::StaticSVD.

bool CAROM::SVD::isNewTimeInterval ( ) const

inline

Returns true if the next sample will result in a new time interval.

Returns: True if the next sample results in the creation of a new time interval.

virtual bool CAROM::SVD::takeSample	(	double *	u_in,
		double	time,
		bool	add_without_increase
	)

pure virtual

Collect the new sample, u_in at supplied time.

Precondition: u_in != 0; time >= 0.0

Parameters

[in]	u_in	The new sample.
[in]	time	The simulation time of the new sample.
[in]	add_without_increase	If true, the addLinearlyDependent is invoked. This only applies to incremental SVD.

Returns: True if the sampling was successful.

Implemented in CAROM::IncrementalSVD, and CAROM::StaticSVD.

Member Data Documentation

Matrix* CAROM::SVD::d_basis

protected

The globalized basis vectors for the current time interval.

The basis vectors are large and each process owns all of the basis vectors.

Matrix* CAROM::SVD::d_basis_right

protected

The globalized right basis vectors for the current time interval.

Depending on the SVD algorithm, it may be distributed across all processors or each processor may own all of U.

Vector* CAROM::SVD::d_S

protected

The vector S which is small.

For all SVD algorithms, S is not distributed and the entire vector exists on each processor.

Matrix* CAROM::SVD::d_snapshots

protected

The globalized snapshot vectors for the current time interval.

The snapshot vectors are large and each process owns all of the snapshot vectors.

Matrix* CAROM::SVD::d_U

protected

The matrix U which is large.

Depending on the SVD algorithm, d_U may be distributed across all processors or each processor may own all of U.

Matrix* CAROM::SVD::d_W

protected

The matrix U which is large.

Depending on the SVD algorithm, d_W may be distributed across all processors or each processor may own all of U.

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

lib/linalg/svd/SVD.h

Public Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation