GSTLALWhiten

GSTLALWhiten — Whiten coloured Gaussian noise.

Functions

GstMessage * gstlal_whiten_message_psd_new ()
REAL8FrequencySeries * gstlal_whiten_message_psd_parse ()

Properties

guint average-samples Read / Write / Construct
gdouble delta-f Read
gboolean expand-gaps Read / Write / Construct
gdouble f-nyquist Read
gdouble fft-length Read / Write / Construct
GValueArray * mean-psd Read / Write
guint median-samples Read / Write / Construct
guint n-samples Read
GSTLAL_PSDMODE psd-mode Read / Write / Construct
gchar * psd-units Read
gdouble sigma-squared Read
GValueArray * spectral-correlation Read
gdouble zero-pad Read / Write / Construct

Types and Values

Object Hierarchy

    GObject
    ╰── GInitiallyUnowned
        ╰── GstObject
            ╰── GstElement
                ╰── GstBaseTransform
                    ╰── GSTLALWhiten

Description

This element constructs a whitening filter from a power spectral density function, which it uses to transform coloured Gaussian noise into white Gaussian noise. The power spectral density function can be provided externally or the element can measure it from the time series data using the LALPSDRegressor machinery. When measuring the spectrum internally, the element will track changes in the spectrum. The “average-samples” and “n-samples” parameters control the time scales of the averaging.

The power spectral density function measured by the element is available to external code via three communication channels. The PSD and its physical parameters can be retrieved through the “delta-f”, “f-nyquist”, and “psd” properites. Use GObject's notify mechanism to be informed of changes in these values. When a new PSD is measured, the element posts a GstMessage to the pipeline's message bus containing the PSD and its properites (see gstlal_whiten_message_psd_parse() for parsing the message). Finally, a request pad named "mean-psd" is available. If the pad is present, each PSD is packed into a GstBuffer and pushed out that GstPad.

To set a PSD, the “f-nyquist” property should be consulted. This property gives the Nyquist frequency (in Hertz) of the PSD data that must be provided and is determined by the sample rate of the time series. The “fft-length” property sets the length of the FFT blocks that will be used for filtering the data, and thus determines the value of the “delta-f” property.

Reviewed: cf799de0ed55bf454fcd7fd8bc88dcaa8c4bd97e 2014-08-10 K. Cannon, J. Creighton, B. Sathyaprakash.

Actions:

  • get plots into documentation

Functions

gstlal_whiten_message_psd_new ()

GstMessage *
gstlal_whiten_message_psd_new (GSTLALWhiten *element,
                               const gchar *instrument,
                               const REAL8FrequencySeries *psd);

gstlal_whiten_message_psd_parse ()

REAL8FrequencySeries *
gstlal_whiten_message_psd_parse (GstMessage *m);

Types and Values

enum gstlal_psdmode_t

Members

GSTLAL_PSDMODE_RUNNING_AVERAGE

   

GSTLAL_PSDMODE_FIXED

   

GSTLAL_PSDMODE_TYPE

#define             GSTLAL_PSDMODE_TYPE

GSTLAL_WHITEN_TYPE

#define             GSTLAL_WHITEN_TYPE

struct GSTLALWhiten

struct GSTLALWhiten;

struct GSTLALWhitenClass

struct GSTLALWhitenClass {
	GstBaseTransformClass parent_class;
};

Members

Property Details

The “average-samples” property

  “average-samples”          guint

Number of FFTs to be used in PSD average.

Flags: Read / Write / Construct

Allowed values: >= 1

Default value: 32


The “delta-f” property

  “delta-f”                  gdouble

PSD frequency resolution in Hertz.

Flags: Read

Allowed values: >= 0

Default value: 0


The “expand-gaps” property

  “expand-gaps”              gboolean

expand gaps to fill entire fft length.

Flags: Read / Write / Construct

Default value: FALSE


The “f-nyquist” property

  “f-nyquist”                gdouble

Nyquist frequency in Hertz.

Flags: Read

Allowed values: >= 0

Default value: 0


The “fft-length” property

  “fft-length”               gdouble

Total length of the FFT convolution (including zero padding) in seconds.

Flags: Read / Write / Construct

Allowed values: >= 0

Default value: 8


The “mean-psd” property

  “mean-psd”                 GValueArray *

Mean power spectral density being used to whiten the data. First bin is at 0 Hz, last bin is at f-nyquist, bin spacing is delta-f.

Flags: Read / Write


The “median-samples” property

  “median-samples”           guint

Number of FFTs to be used in PSD median history.

Flags: Read / Write / Construct

Allowed values: >= 1

Default value: 9


The “n-samples” property

  “n-samples”                guint

Number of FFTs used for PSD average. Stops increasing after the number of FFTs equals average-samples.

Flags: Read

Default value: 0


The “psd-mode” property

  “psd-mode”                 GSTLAL_PSDMODE

PSD estimation mode.

Flags: Read / Write / Construct

Default value: GSTLAL_PSDMODE_RUNNING_AVERAGE


The “psd-units” property

  “psd-units”                gchar *

LAL unit string giving units for PSD samples.

Flags: Read

Default value: "s"


The “sigma-squared” property

  “sigma-squared”            gdouble

FFT window mean square.

Flags: Read

Allowed values: >= 0

Default value: 0


The “spectral-correlation” property

  “spectral-correlation”     GValueArray *

Two-point spectral correlation function for output stream. Bin index is |k - k'|.

Flags: Read


The “zero-pad” property

  “zero-pad”                 gdouble

Length of the zero-padding to include on both sides of the FFT in seconds.

Flags: Read / Write / Construct

Allowed values: >= 0

Default value: 2