Configuration files for cogwheelpipe¶

In order to allow cogwheel to be configured in a way which asimov understands we have introduced simple configuration files for cogwheel pipe which allow the inference code to be configured in a way which is easy to reproduce.

This page provides a guide as to the format of this configuration file.

The configuration file is written in YAML format, and we will sometimes use “inline YAML” to describe it here. That means that the following hierarchical dictionary in YAML format

waveform:
  approximant: IMRPhenomXPHM

Is written as waveform: approximant: IMRPhenomXPHM in this document, in order to make things a little more concise.

Event-specific Settings¶

These settings are specific to a given gravitational wave signal.

event: name

The name of the gravitational wave event, for example

event:
  name: GW150914

This value can be set to automatically obtain data for a given event.

event: event time

The GPS time of the gravitational wave event. This is optional when downloading data from GWOSC (as the event time can be looked up automatically), but is required when using frame files.

event:
  name: GW150914
  event time: 1126259462.4

event: fiducial parameters

The fiducial parameters, or “best guess” parameters for the event. For example

event:
  fiducial parameters:
    chirp mass: 30

Data Settings¶

These settings control how strain data is acquired for the analysis.

data

The data section contains configuration for frame files and channel names. This follows asimov’s blueprint structure for data configuration.

data: frame files

Specify GWF (Gravitational Wave Frame) files containing strain data for each interferometer. This is useful when you want to use data from local files or from previous analyses (e.g., from asimov-gwdata pipeline) instead of downloading from GWOSC.

When using frame files, the event: event time parameter is required.

data:
  frame files:
    H1: /path/to/H-H1_GWOSC_16KHZ_R1-1126257415-4096.gwf
    L1: /path/to/L-L1_GWOSC_16KHZ_R1-1126257415-4096.gwf
    V1: /path/to/V-V1_GWOSC_16KHZ_R1-1126257415-4096.gwf

If frame files are not specified, data will be downloaded from GWOSC.

data: channels

Optionally specify custom channel names for each detector when reading frame files. If not specified, the default channel naming convention {IFO}:GWOSC-16KHZ_R1_STRAIN is used.

data:
  frame files:
    H1: /path/to/H1_frame.gwf
    L1: /path/to/L1_frame.gwf
  channels:
    H1: H1:GWOSC-16KHZ_R1_STRAIN
    L1: L1:GWOSC-16KHZ_R1_STRAIN

This approach makes data handling compatible with bilby, allowing results from the two pipelines to be compared directly.

data: segment length

Optionally specify the total duration of data to read in seconds. Default is 32.0 seconds. The time before the event is calculated as: segment length - post trigger time.

data:
  segment length: 32.0

likelihood: post trigger time

Optionally specify the time after the event (in seconds) to include in the analysis. Default is 2.0 seconds. This is used to calculate the data segment boundaries and padding for filtering/whitening operations.

likelihood:
  post trigger time: 2.0

quality: maximum frequency

Optionally specify the maximum frequency for the analysis in Hz. Default is 1024.0 Hz. Can be specified per-detector (as a dictionary) or as a single value. If specified per-detector, the highest value will be used for all detectors.

quality:
  maximum frequency:
    H1: 1024.0
    L1: 1024.0
  # Or as a single value:
  # maximum frequency: 1024.0

psds

Specify Power Spectral Density (PSD) files for each interferometer. This is useful when you want to use PSDs generated by other tools (e.g., BayesWave) for on-source analysis.

PSD files should be in two-column text format: frequency (Hz), PSD value.

psds:
  H1: /path/to/H1_psd.txt
  L1: /path/to/L1_psd.txt

Frame files and PSDs can be used together to provide complete control over input data and noise characterization.

psds: whitening_filter

Configure the whitening filter parameters when using custom PSDs.

psds:
  H1: /path/to/H1_psd.txt
  L1: /path/to/L1_psd.txt
  whitening_filter:
    fmin: 15.0       # Minimum frequency (Hz), default: 15.0
    df_taper: 1.0    # Taper width (Hz), default: 1.0

Waveform settings¶

These settings control the waveform approximant used in the analysis.

waveform: approximant

The name of the waveform approximant to be used in the analysis, for example

waveform:
  approximant: IMRPhenomXPHM

Sampler settings¶

These settings control the behaviour of the sampler.

sampler: live points: The number of live points to use when sampling. If not specified directly a default of 1000 is used.

Likelihood settings¶

These settings control the behaviour of the likelihood function.

likelihood: marginalisation

Enable marginalisation over extrinsic parameters.

You may need to alter your prior in order to use a marginalisation.

For example, for distance marginalisation:

::

likelihood:

marginalisation:

distance

Prior settings¶

prior: class

Use a specified prior class for the analysis.

For example

::

prior:: class: CartesianintrinsicIASPrior

prior: distributions

Set specific values for prior distributions.

For example:

::

prior:

distributions:

chirp mass:: minimum: 10 maximum: 40