data_operations.find_whitened_outliers#

Back to Module page

Summary#

Zeros mask in place on either side of outliers in whitened data stream

Signature#

def find_whitened_outliers(strain_wt, qmask, valid_mask, dt, clipwidth, sigma_clipping_threshold = None, outlier_fraction = params.OUTLIER_FRAC, zero_mask = True, nfire = params.NPERFILE, renorm_wt = True, sc_n01 = 1.0, mask_save = None, verbose = True)

Input variables#
Name	Type	Default	Description
`strain_wt`			Whitened strain data
`qmask`			Boolean mask with zeros at holes in unwhitened data
`valid_mask`			Boolean mask with zeros where we cannot trust whitened data
`dt`			Time interval between successive elements of strain_wt (s)
`clipwidth`			Half-width of window around outliers to zero (in s)
`sigma_clipping_threshold`		None	Threshold for sigma clipping computed from waveform. If none, clipping depends on params.NPERFILE
`outlier_fraction`		params.OUTLIER_FRAC	If outlier is more than 1/outlier_fraction x sigma clipping threshold, adjust threshold to avoid overcorrecting due to ringing of the outlier against the whitening filter
`zero_mask`		True	Flag indicating whether to zero the mask at glitches. Turn off for vetoing
`nfire`		params.NPERFILE	Number of times glitch detector fires per perfect file
`renorm_wt`		True	Flag whether we scaled the whitened data to have unit variance after highpass
`sc_n01`		1.0	Factor to multiply the whitened data with to get each data point to be N(0,1), used only if renorm_wt is False (which is the default from now)
`mask_save`		None	If desired, boolean mask on strain_wt with zeros at data to avoid
`verbose`		True	If true, prints details about outlier check

Output variables#

Return annotation	Docstring type	Description
`None`		Zeros mask in place, returns 1. List of indices that jumped 2. Mask with zeros where we should zero the data as a result 3. Flag indicating whether we had an overly loud outlier 4. String with details of the test

Docstring#

Zeros mask in place on either side of outliers in whitened data stream
:param strain_wt: Whitened strain data
:param qmask: Boolean mask with zeros at holes in unwhitened data
:param valid_mask:
    Boolean mask with zeros where we cannot trust whitened data
:param dt: Time interval between successive elements of strain_wt (s)
:param clipwidth: Half-width of window around outliers to zero (in s)
:param sigma_clipping_threshold:
    Threshold for sigma clipping computed from waveform. If none, clipping
    depends on params.NPERFILE
:param outlier_fraction:
    If outlier is more than 1/outlier_fraction x sigma clipping threshold,
    adjust threshold to avoid overcorrecting due to ringing of the outlier
    against the whitening filter
:param zero_mask:
    Flag indicating whether to zero the mask at glitches. Turn off for
    vetoing
:param nfire: Number of times glitch detector fires per perfect file
:param renorm_wt:
    Flag whether we scaled the whitened data to have unit variance after
    highpass
:param sc_n01:
    Factor to multiply the whitened data with to get each data point to be
    N(0,1), used only if renorm_wt is False (which is the default from now)
:param mask_save:
    If desired, boolean mask on strain_wt with zeros at data to avoid
:param verbose: If true, prints details about outlier check
:return: Zeros mask in place, returns
         1. List of indices that jumped
         2. Mask with zeros where we should zero the data as a result
         3. Flag indicating whether we had an overly loud outlier
         4. String with details of the test