class LPEFilter : public FIRFilter

Linear Predictive Error Filter implementation.

Inheritance:

Public Methods

	LPEFilter (int filterLength = 0, int trainPeriod = 0, int trainLength = 0) Default constructor.
	LPEFilter (const LPEFilter& lpef) Copy constructor.
virtual	~LPEFilter (void) Destructor.
LPEFilter&	operator= (const LPEFilter& lpef) Assignment operator.
int	apply (const TSeries& in, TSeries& out) Filter TSeries
int	train (const TSeries& data) Determine filter coefficients
void	setTrainPeriod (int period) Set retraining period
void	setTrainLength (int length) Set training length
void	setHistory (const TSeries& in) Set History from TSeries
int	getTrainPeriod (void) const Get retraining period
int	getTrainLength (void) const Get training length
Time	getTrainTime (void) const Get time of most recent training
Interval	getTransientTime (void) const Get the filter length in seconds.

Inherited from FIRFilter:

Public Methods

virtual FIRFilter* clone(void) const

void dataCheck(const TSeries& ts) const

Time getCurrentTime(void) const

int getLength(void) const throw()

double getRate(void) const throw()

Time getStartTime(void) const

bool getStat(void) const throw()

Interval getTimeDelay(void) const

bool inUse(void) const

void dump(std::ostream& ostr) const throw()

void reset(void)

void setCoefs(int N, const double* Coefs) throw()

void setCoefs(const double* Coefs) throw()

void setLength(int N) throw()

void setRate(double F) throw()

virtual bool Xfer(fComplex& coeff, double f) const throw()

bool Xfer(fComplex* tf, const float* freqs, int points) const throw()

FSeries Xfer(float Fmin=0.0, float Fmax=0.0, float dF=1.0) const throw()

Protected Methods

bool xfer(fComplex& coeff, double f) const throw()

Inherited from Pipe:

Public Methods

virtual TSeries operator)(const TSeries& in)

FilterIO& operator)(const FilterIO& in)

virtual bool isDataValid(const TSeries& in) const

KeyChain& getInputKeys(void) const

KeyChain& getOutputKeys(void) const

Inherited from FilterBase:

Public Methods

virtual bool rootGetStartTime(Time& t) const throw()

virtual bool rootGetCurrentTime(Time& t) const throw()

LPEFilter(int filterLength = 0, int trainPeriod = 0, int trainLength = 0)

Construct a new filter with the given filter length, retraining period, and training length. The filter history and start times are cleared.

Parameters:

Filter - length
Retraining - period
Training - length

LPEFilter(const LPEFilter& lpef)

Creates new filter with the same filter length, retraining period, training length, sample rate, and filter coefficients as the given filter. The filter history and start times are cleared.

virtual ~LPEFilter(void)

Destroys the filter and releases any allocated memory.

LPEFilter& operator=(const LPEFilter& lpef)

Creates new filter with the same filter length, retraining period, training length, sample rate, and filter coefficients as the given filter. The filter history and start times are cleared.

int apply(const TSeries& in, TSeries& out)

Filters an input TSeries and places the result in the given output TSeries. Before filtering, training is performed if the filter has never been trained or if the retrain period has elapsed. A non-zero return value indicates a training was performed. The filter history is also updated.

Returns:

Non-zero if retrain occurred

Parameters:

TSeries - input
TSeries - output

int train(const TSeries& data)

Determine the optimal filter coefficients based on training data. The entire TSeries is used as the training data. The training data length must be greater than the filter length. A non-zero return value indicates that the training was unsuccessful.

Returns:

Non-zero on error

Parameters:

TSeries - training data

void setTrainPeriod(int period)

Sets the length of time (in seconds) between trainings. To ensure that training does not occur across the boundary between successive data blocks, the retraining period should be an integer multiple of the data block duration. A value of zero prevents automatic periodic retraining of the filter.

Parameters:

Retraining - period [seconds]

void setTrainLength(int length)

Sets the amount of data (in samples) which is used to train the filter. To determine the filter coefficients, the training algorithm requires a training length greater than the filter length.

Parameters:

Training - length [samples]

void setHistory(const TSeries& in)

Sets the filter history by applying the filter to a TSeries.

Parameters:

TSeries - data

int getTrainPeriod(void) const

Gets the length of time (in seconds) between trainings.

Returns:

Retraining period [seconds]

int getTrainLength(void) const

Gets the amount of data (in samples) which is used to train the filter.

Returns:

Training length [samples]

Time getTrainTime(void) const

Gets the time of the most recent filter training. The train time is set automatically whenever the filter is trained.

Returns:

Time of last training

Interval getTransientTime(void) const

Gets the length in seconds of the filter's impulse response. Determines the dead time after an impulsive event, during which the filter output is unreliable.

Returns:

Filter length in seconds.

alphabetic index hierarchy of classes

Please send questions and comments to zweizig_j@ligo.caltech.edu

generated by doc++