Gravelamps INI Structure

This page details the options available in INI files that may be supplied to Gravelamps. For all programs using the INI as an input, the INI are compatible. Specifically, the gravelamps_inference INI is a superset of all other INIs.

Note

The section headers in the Gravelamps INI are mandatory.

The following sections will have the title of the INI section as their title.

Output

Keys here deal with the output of Gravelamps. This includes both files and the logger.

  • outdir

    Name of the output directory into which all subdirectores are placed. The default is the current working directory.

  • label

    Label applied to varied files produced for the run. The default value is lens.

  • logging_level

    String level denoting the lowest level of logger message that will be noted. The default value is INFO, and in all programs may be overriden by use of the -v option.

Run

Keys here deal with settings for the running of the program.

  • local

    Determines whether the job will run locally or be built for the HTCondor scheduler. The default value is False and in all programs may be overriden by use of the -l option.

  • no_copy_generation_files_to_data_dict

    Determines whether the generation program should move the files to the data directory. This is necessary to use the HTCondor transfer files system and will thus default to False.

  • number_generation_jobs

    The number of jobs the task of generating interpolator data files should be split into, which can speed up the construction of the data files when using the HTCondor scheduler. Default is 1.

  • no_analysis_generation

    Specifies that the analysis waveform for inference runs does not require generation.

  • no_injection_generation

    Specifies that the injection waveform for inference runs does not require generation.

  • includes_injection

    Specifies that the inference run requires handling of injection generation.

Scheduler

Keys in this section determine properties of the scheduler chosen.

  • scheduler Determines which scheduler is used. Supported values are condor and slurm.

Subsections are then given which provide arguments directly to the scheduler. To see the available options, for HTCondor, see the HTCondor manual and for slurm, see the slurm documation.

Note

For the HTCondor scheduler, if the accounting_group key is not present, Gravelamps programs will issue a warning that the job may fail to submit.

Lens

Keys that are given here determine the information regarding which lens model and properties to use for that lens.

  • module

    Python path to the implementation of the given model.

  • minimum_dimensionless_frequency

    Minimum dimensionless frequency for the generated grid.

  • maximum_dimensionless_frequncy

    Maximum dimensionless frequency for the generated grid.

  • length_dimensionless_frequency

    Number of points for the dimensionless frequency axis of the generated grid.

  • function_dimensionless_frequency

    Function type to use for generating the grid axis. Must be an attribute of numpy. Default is linspace.

  • minimum_source_position

    Minimum source position for the generated grid.

  • maximum_source_position

    Maximum source position for the generated grid.

  • length_source_position

    Number of points for the source position axis of the generated grid.

  • function_source_position

    Function type to use for generating the grid axis. Mut be an attribute of numpy. Default is linspace.

  • dimensionless_frequency_file

    File path to retrieve dimensionless frequency grid information rather than generating using the above settings.

  • source_position_file

    File path to retrieve source position grid information rather than generating using the above settings.

  • amplification_factor_real_file

    File path to retrieve real component of amplification factor data rather than generating from the above grid.

  • amplification_factor_imag_file

    File path to retrieve imaginary component of amplification factor data rather than generating from the above grid.

  • data_generation_kwargs

    Any additional keyword arguments to pass to the data generation function. Should be in the form of a python dictionary.

  • extra_lens_keys

    Additional parameters that are required by the amplification factor function.

  • use_redshifted_lens_mass

    Instructs the waveform generator to use the redshifted lens mass instead of the lens mass and lens fractional distance when computing the amplification factor.

  • interpolator_kwargs

    Additional parameters to pass to the interpolator when generating the waveform.

  • no_conversion_to_dimensionless_frequency

    Instructs the waveform generator to pass the frequency array directly to the amplification function rather than first converting to the dimensionless frequency.

An additional section termed lens.injection may be given in gravelamps_inference INI files that will override any of these settings for the injected lens. Will otherwise default to these settings for both injection and recovery. Additional keys in this section may be added, which will be passed into the waveform generator’s waveform_arguments attribute in the lens_settings dictionary. Gravelamps’ waveform generators have direct access to this dictionary through its lens_settings property. Requirements and supported options for these keys are detailed in the Model Specific Options page.

Prior

Keys and values here are fed directly to the bilby_pipe prior-dict argument.

Injection.Parameters

Keys and values here are fed directly to the bilby_pipe injection-dict argument, i.e. they specify the parameters of the injection waveform.