Transient gravitational wave data I/O

This document describes how bilby handles interferometer data and how you can load data.

What is used by the likelihood?

The primary likelihhod for use in transient gravitational wave data analysis is the GravitationalWaveTransient . This takes an argument interferometers which is a list of objects. These objects know about the geometry of the detector, the noise properties of the detector, and the segment of data which is to be analysed. In the following, we’ll describe different ways to initilalse a likelihood for gravitational wave data analysis.

Making an Interferometer

To make an empty interferometer, for example the Hanford detector:

>>> H1 ='H1')

By default, these will have power spectral densities based on the expected design sensitivity of the detector. The strain data (i.e. the data about the segment of interferomer data which we want to analyse) is in an attribute H1.strain_data. The following is a list of ways to set this strain data.

Setting the strain data

Setting the strain data using gwpy

The gwpy module is the recommended way to read in and manipulate gravitational wave strain data. For example, here is a snippet taken from the documentation to obtain the Hanford open date for GW150914:

>>> from gwpy.timeseries import TimeSeries
>>> time_series = TimeSeries.fetch_open_data('H1', 1126259446, 1126259478)

Gwpy provides a complete interface for reading any type of data related to gravitational wave strain data. Once you have created your time series, you can pass it into your bilby interferometer as follows:

>>> H1.set_strain_data_from_gwpy_time_series(time_series=time_series)

Setting the strain data directly

If you have an array of the frequency-domain strain data, you can set it directly like this:

>>> H1.set_strain_data_from_frequency_domain_strain(frequency_domain_strain,

Where the given arguments are things you have already defined in your python script. If you’d prefer to give the frequency_array to which the data corresponds instead of the sampling_frequency and duration this can also be done:

>>> H1.set_strain_data_from_frequency_domain_strain(frequency_domain_strain,

Setting the strain data to be Gaussian noise

Often, for testing, you may want to just generate a realization of coloured Gaussian noise from the power spectral density. This can be done using this method:

>>> H1.set_strain_data_from_power_spectral_density

Setting the strain data to be zero noise

You can also set the strain data without any noise at all

>>> H1.set_strain_data_from_zero_noise

Injecting a signal

If you wish to inject a signal into the data, you can use this function:


To inject a signal a is required. This is where you must specify the injection’s duration, start-time, waveform. Additionally, a parameter_conversion function can be passed. The default is parameter_conversion = This expects injection parameters to be provided in the detector-frame, e.g.

>>> parameters = dict(
    mass_1=36., mass_2=29.,
    a_1=0.4, a_2=0.3, tilt_1=0.5, tilt_2=1.0, phi_12=1.7, phi_jl=0.3,
    luminosity_distance=2000., theta_jn=0.4, psi=2.659,
    phase=1.3, geocent_time=1126259642.413, ra=1.375, dec=-1.2108