NuRadioReco.modules.channelBandPassFilter module
- class NuRadioReco.modules.channelBandPassFilter.channelBandPassFilter(caching=True)[source]
Bases:
object
Band pass filters the channels using different band-pass filters.
Methods
get_filter
(frequencies, station_id, ...[, ...])helper function to return the filter that the module applies.
run
(evt, station, det[, passband, ...])Run the filter
begin
end
get_filter_arguments
- Parameters:
- caching: bool, default True
If True (default), internally caches the filter. This speeds up the (common) case where the same filter is applied to multiple channels.
Methods
get_filter
(frequencies, station_id, ...[, ...])helper function to return the filter that the module applies.
run
(evt, station, det[, passband, ...])Run the filter
begin
end
get_filter_arguments
- static get_filter_arguments(channel_id, passband, filter_type, order=2, rp=None, roll_width=None, half_hann_percent=None)[source]
- run(evt, station, det, passband=None, filter_type='rectangular', order=2, rp=None, roll_width=0.0025, half_hann_percent=0.1)[source]
Run the filter
- Parameters:
- evt, station, det
Event, Station, Detector
- passband: list or dict of lists, (default: [55 * units.MHz, 1000 * units.MHz])
passband[0]: lower boundary of filter, passband[1]: upper boundary of filter
a dict can be used to specify a different bandwidth per channel, the key is the channel_id
- filter_type: string or dict
‘rectangular’: perfect straight line filter
‘butter’: butterworth filter from scipy
‘butterabs’: absolute of butterworth filter from scipy
‘gaussian_tapered’ : a rectangular bandpass filter convolved with a Gaussian
‘hann_tapered’ : a rectangular bandpass filter with the ends replaced by a half-Hann window to applied in the time domain. In this case the passband parameter is ignored
‘FIR <type> <parameter>’ - see below for FIR filter options
or any filter that is implemented in
NuRadioReco.detector.filterresponse
. In this case the passband parameter is ignoreda dict can be used to specify a different bandwidth per channel, the key is the channel_id
- order: int (optional, default 2) or dict
for a butterworth filter: specifies the order of the filter
- rp: float
The maximum ripple allowed below unity gain in the passband. Specified in decibels, as a positive number. (for chebyshev filter)
a dict can be used to specify a different bandwidth per channel, the key is the channel_id
- roll_widthfloat, default=2.5MHz
Determines the sigma of the Gaussian to be used in the convolution of the rectangular filter. (Relevant for the Gaussian tapered filter)
- half_hann_percentfloat, default=0.1
The size of the half-Hann window expressed as a percentage of the length of the trace. (Relevant for the Hann tapered filter)
Notes
Added Jan-07-2018 by robert.lahmann@fau.de: FIR filter: see https://docs.scipy.org/doc/scipy/reference/generated/scipy.signal.get_window.html for window types; some windows need additional parameters; Default window is hamming (does not need parameters). Examples:
filter_type=’FIR’ : use hamming window
filter_type=’FIR hamming’: same as above
filter_type=’FIR kaiser 10’ : Use Kaiser window with beta parameter 10
filter_type=’FIR kaiser’ : Use Kaiser window with default beta parameter 6
In principle, window names are just passed on to signal.firwin(), but if parameters are required, then these cases must be explicitly implemented in the code below.
The four main filter types can be implemented:
LP: passband[0]=None, passband[1] = f_cut
HP: passband[0]=f_cut, passband[1] = None
BP: passband[0]=f_cut_low, passband[1] = f_cut_high
BS: passband[0]=f_cut_high, passband[1] = f_cut_low (i.e. passband[0] > passband[1])
- get_filter(frequencies, station_id, channel_id, det, passband, filter_type, order=2, rp=None, roll_width=0.0025, half_hann_percent=None)[source]
helper function to return the filter that the module applies.
The unused parameters station_id, channel_id, det are needed to have the same signature as the
get_filter
functions of other modules, e.g. the hardwareResponseIncorporator.Note that this function returns the filter response in the frequency domain. Filters which are applied in the time domain (e.g. ‘FIR’, ‘hann_tapered’) will raise a
NotImplementedError
- Parameters:
- frequencies: array of floats
the frequency array for which the filter should be returned
- station_id: int
the station id
- channel_id: int
the channel id
- det: detector instance
the detector
- passband: list or dict of lists, (default: [55 * units.MHz, 1000 * units.MHz])
passband[0]: lower boundary of filter, passband[1]: upper boundary of filter a dict can be used to specify a different bandwidth per channel, the key is the channel_id
- filter_type: string or dict
‘rectangular’: perfect straight line filter
‘butter’: butterworth filter from scipy
‘butterabs’: absolute of butterworth filter from scipy
‘gaussian_tapered’ : Gaussian tapered version of the rectangular filter
or any filter that is implemented in
NuRadioReco.detector.filterresponse
. In this case the passband parameter is ignoreda dict can be used to specify a different bandwidth per channel, the key is the channel_id
- order: int (optional, default 2) or dict
for a butterworth filter: specifies the order of the filter
- rp: float
The maximum ripple allowed below unity gain in the passband. Specified in decibels, as a positive number. (for chebyshev filter)
a dict can be used to specify a different bandwidth per channel, the key is the channel_id
- roll_widthfloat, default=2.5MHz
Determines the sigma of the Gaussian to be used in the convolution of the rectangular filter. (Relevant for the Gaussian tapered filter)
- Returns:
- array of complex floats
the complex filter amplitudes
- Other Parameters:
- half_hann_percent: None
This parameter is included to have the same signature as the
run()
method. As it is only relevant for the hann_tapered filter it is not actually used here.
- NuRadioReco.modules.channelBandPassFilter.test_equality(a, b)[source]
Test if two things are equal.
Generalizes a==b to support lists, lists of lists and arrays etc. Will return True if and only if a and b are equal.
- Parameters:
- a: object | list | array
- b: object | list | array
- Returns:
- is_equal: bool
True if and only if a == b