npstat::ConvolutionEngineND Class Reference

#include <ConvolutionEngineND.hh>

Public Member Functions
	ConvolutionEngineND (const unsigned *shape, unsigned rank, unsigned optimization=FFTW_ESTIMATE)
template<typename Real >
void	setFilter (const Real *data, const unsigned *dataShape, unsigned dataRank)
template<typename Real1 , typename Real2 >
void	convolveWithFilter (const Real1 *in, Real2 *out, const unsigned *dataShape, unsigned dataRank)
template<typename Real >
void	depositFilter (unsigned long id, const Real *data, const unsigned *dataShape, unsigned dataRank)
template<typename Real1 , typename Real2 >
void	convolveWithDeposit (unsigned long id, const Real1 *in, Real2 *out, const unsigned *dataShape, unsigned dataRank)
template<typename Real2 >
void	convolveFilterAndDeposit (unsigned long id, Real2 *out, const unsigned *dataShape, unsigned dataRank)
template<typename Real2 >
void	deconvolveFilterAndDeposit (unsigned long id, Real2 *out, const unsigned *dataShape, unsigned dataRank)
bool	discardFilter (unsigned long id)
void	clearFilterBank ()
bool	isShapeCompatible (const unsigned *shape, unsigned rank) const
unsigned	rank () const
const std::vector< unsigned > &	shape () const
unsigned long	dataLength () const
bool	isFilterSet () const
unsigned	filterBankSize () const

Detailed Description

Class for performing multivariate convolutions by FFT (using FFTW library). Internally, transforms are calculated in double precision.

Constructor & Destructor Documentation

ConvolutionEngineND()

npstat::ConvolutionEngineND::ConvolutionEngineND	(	const unsigned *	shape,
		unsigned	rank,
		unsigned	optimization = FFTW_ESTIMATE
	)

Constructor takes the shape of the data and FFTW optimization flag as parameters. Make sure the data dimensions are sufficient to avoid the convolution wrap-around.

Member Function Documentation

clearFilterBank()

void npstat::ConvolutionEngineND::clearFilterBank

(

)

Discard all filters deposited by depositFilter method

convolveFilterAndDeposit()

template<typename Real2 >

void npstat::ConvolutionEngineND::convolveFilterAndDeposit	(	unsigned long	id,
		Real2 *	out,
		const unsigned *	dataShape,
		unsigned	dataRank
	)

Convolve the filter established by the "setFilter" method with a previously deposited filter

convolveWithDeposit()

template<typename Real1 , typename Real2 >

void npstat::ConvolutionEngineND::convolveWithDeposit	(	unsigned long	id,
		const Real1 *	in,
		Real2 *	out,
		const unsigned *	dataShape,
		unsigned	dataRank
	)

Convolve provided data with one of the filters previously established by the depositFilter method. The "id" argument is used to identify a particular filter.

convolveWithFilter()

template<typename Real1 , typename Real2 >

void npstat::ConvolutionEngineND::convolveWithFilter	(	const Real1 *	in,
		Real2 *	out,
		const unsigned *	dataShape,
		unsigned	dataRank
	)

Convolve provided data with the filter previously established by setFilter

dataLength()

unsigned long npstat::ConvolutionEngineND::dataLength ( ) const

inline

Expected data length

deconvolveFilterAndDeposit()

template<typename Real2 >

void npstat::ConvolutionEngineND::deconvolveFilterAndDeposit	(	unsigned long	id,
		Real2 *	out,
		const unsigned *	dataShape,
		unsigned	dataRank
	)

Deconvolve the filter established by the "setFilter" method with a previously deposited filter

depositFilter()

template<typename Real >

void npstat::ConvolutionEngineND::depositFilter	(	unsigned long	id,
		const Real *	data,
		const unsigned *	dataShape,
		unsigned	dataRank
	)

Provide a filter for subsequent convolutions.

"id" is the user id for the filter which can be later reused with the method "convolveWithDeposit".

discardFilter()

bool npstat::ConvolutionEngineND::discardFilter

(

unsigned long

id

)

Discard a filter previously deposited by depositFilter method. This method returns "true" if the filter with the given id was indeed discarded and "false" if such a filter was not found.

isShapeCompatible()

bool npstat::ConvolutionEngineND::isShapeCompatible	(	const unsigned *	shape,
		unsigned	rank
	)		const

Check if an array shape is compatible with this filter

rank()

unsigned npstat::ConvolutionEngineND::rank ( ) const

inline

Expected data dimensionality

setFilter()

template<typename Real >

void npstat::ConvolutionEngineND::setFilter	(	const Real *	data,
		const unsigned *	dataShape,
		unsigned	dataRank
	)

Provide the filter for subsequent convolutions

shape()

const std::vector<unsigned>& npstat::ConvolutionEngineND::shape ( ) const

inline

Expected shape of the data

---

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

• nm/ConvolutionEngineND.hh