Routine options

spectral_separation_routine requires that you create 3 custom objects from the module storage_classes that holds options and extra parameters for the parent and two child routines.

These are:

• a RoutineOptions object for the parent routine.
• a SeparateComponentsOptions object for the component separation subroutine.
• a RadialVelocityOptions object for the radial velocity subroutine.

While these objects can be initialized directly in your python script by supplying the inputs shown below, they can also be initialized through configuration files with the function storage_classes.load_configuration_files(). It requires input arguments:

• loc_routine_file: string, path of configuration file for RoutineOptions object.
• loc_separation_file: string, path of configuration file for SeparateComponentsOptions object.
• loc_rv_file: string, path of configuration file for RadialVelocityOptions object.

The example script with KIC8430105 shows how to initialize these objects either by using configuration files or directly in the script.

Below, ALL parameters in the objects are described. Some parameters are accessible from multiple objects, making it possible to give different values for each. This is discouraged. Many of the parameters can be kept at their default values without loss of quality in the output.

RoutineOptions:

• convergence_limit (float), Default = 1E-5. Convergence criteria for the radial velocity RMS needed to exit the parent routine.
• iteration_limit (int), Default = 10. Maximum allowed iterations of the parent routine before automatic exit.
• plot (bool), Default = True. If True, diagnostic plots are shown and updated between iterations.
• verbose (bool), Default = True. Indicates if parent routine should print extra updates during iteration.
• return_unbuffered (bool), Default = True. If received spectra are padded around the edges and the buffer_mask parameter is supplied, setting this to True will remove said padding before returning separated spectra of both components.
• save_plot_path (string), Default = None. If not None, the final diagnostic plots will be saved as a multipage PDF file in the specified folder.
• save_all_results (bool), Default = True. If True, raw data is saved to disk before returning from the function.
• save_path (string), Default = ‘/.’. Defines the folder that raw data is saved to when save_all_results is set to True.
• filename_bulk (string), Default = None. If save_all_results is True, output filenames will be prepended with this string. If None, they will be prepended with the wavelength interval used (for example ‘4500_5825’).

SeparateComponentsOptions:

• delta_v (float), Default = 1.0. This is a required parameter, and must be set to be equal to the delta_v used to resample the spectra.
• convergence_limit (float), Default = 1E-2. Convergence criteria for the changes induced in the component spectra between iterations.
• max_iterations (int), Default = 10. Maximum allowed iterations for the separate_component_spectra subroutine.
• rv_proximity_limit (float), Default = 0.0. If 0.0, this parameter is ignored. Otherwise, it is used to exclude spectra for the component calculation based on the calculated RV difference between each component. It is superceded by either rv_lower_limit or use_for_spectral_separation_X if any of them are provided for the component. See note after.
• rv_lower_limit (float), Default = 0.0. If 0.0, this parameter is ignored. Otherwise, it is used to exclude spectra for the component calculation based on the RV of the primary (A) component. This requires that the system RV has been approximately removed from the spectra beforehand. It is superceded by use_for_spectral_separation_X if provided for the component. See note after.
• ignore_component_B (bool), Default = False. If True, the secondary component (B) will be ignored for ALL provided spectra. This should NEVER be used unless you know with certainty that all the provided spectra does not include light from the secondary. In such cases, I don’t know why you would use this program instead of something simpler, but it is here if you need it.
• verbose (bool), Default = False. If True, provides additional prints to the console.
• use_for_spectral_separation_A (array_like shape (n_used_spectra, )), Default = None. If supplied, should be an array of integers indicating which of the provided spectra should be used for the calculation of the separated spectrum of component A. If 5 out of 21 spectra are used, this array should have length 5. Example: [0, 2, 6, 11, 15]
• use_for_spectral_separation_B. Same as previous, but for component B instead.
• weights (numpy array shape (n_spectra,)), Default = None. If provided, supplies a weight to each spectrum in the calculation of both separated component spectra.

  Note on spectra used for calculating separated spectra

  5 of the above parameters can affect which spectra are used in the separated component spectra, depending on which are provided. These are

• rv_proximity_limit
• rv_lower_limit
• use_for_spectral_separation_A and use_for_spectral_separation_B
• weights

Best results are usually obtained by manually specifying which spectra to use with the two use_for_spectral_separation_X parameters. If these are supplied, rv_lower_limit and rv_proximity_limit will be ignored for this purpose. The weights parameter can serve a similar purpose if values 0 and 1 are used to indicate which spectra to use.

If the use_for_spectral_separation_X parameter is not supplied for a component, and both rv_lower_limit and rv_proximity_limit are given, the latter parameter will be ignored.

RadialVelocityOptions:

• vsini_guess_A (float), Default = 1.0. Guess for the fit parameter of vsini for component A, for the rotational broadening function profile fit. If fitting for vsini, this will be modified by the parent routine after each iteration and set to the mean of all the spectra.
• vsini_guess_B (float), Default = 1.0.
• vary_vsini_A (bool), Default = True. If False, vsini will not be a free parameter during RV fit for component A.
• vary_vsini_B (bool), Default = True.
• vsini_vary_limit_A (float), Default = None. If provided, vsini_A can be changed only by +- vsini_vary_limit_A*vsini_A during each iteration. If not provided, parent routine will set it to 0.7 for stability.
• vsini_vary_limit_B (float), Default = None. Same as above
• delta_v (float), Default = 1.0. Must be same as for SeparateComponentsOptions.
• spectral_resolution (float), Default = 60000. Provides the spectral resolution such that the BF fit can be appropriately broadened. If you are fitting for vsini and not interested in its value, this does not need to be the exact spectral resolution.
• velocity_fit_width_A (float), Default = 50. How far away to include data during fitting of component A RV (in km/s). The rotational broadening function profile also separately masks the data using (velocity - rv)/vsini for stability, which is worth keeping in mind.
• velocity_fit_width_B (float), Default = 20.
• refit_width_A (float), Default = None. If either this or refit_width_B is provided, will repeat fitting procedure afterwards with a smaller fitting width. This slows computation but can improve final precision on found RVs and might improve convergence.
• refit_width_B (float), Default = None.
• limbd_coef_A (float), Default = 0.68. Linear limb darkening parameter for fitting profile. Not of great importance if RV is the only fitted parameter of interest.
• limbd_coef_B (float), Default = 0.68.
• vary_limbd_coef_A (bool), Default = False. If True, linear limb darkening coefficient will be a free parameter for component A fit.
• vary_limbd_coef_B (bool), Default = False.
• RV_A (numpy array shape (n_spectra, ), Default = None. Routine will set this automatically, so there is no reason to supply this manually. Fitting routine will mask data using this and vsini with (velocity - rv)/vsini.
• RV_B (numpy array shape (n_spectra, ), Default = None.
• smooth_sigma_A (float), Default = 4.0. Smoothing value in km/s to use when performing gaussian smoothing on the component A broadening function before fitting.
• smooth_sigma_B (float), Default = 4.0
• bf_velocity_span (float), Default = 200. The width of the broadening function in velocity space (km/s). This is a fudge parameter that does not have a clear correct value. It must not be too small, and not too large, and should be experimented with.
• ignore_at_phase_A (tuple(float, float)), Default = None. If supplied, separated component A spectrum will not be subtracted from spectra used for component B RV if they are within specific phase interval (lower, upper). This interval can wrap around, e.g. (0.98, 0.02) is a valid interval. Make sure that the provided time_values has the same phase-shift when calculating phase with np.mod(time_values, period)/period.
• ignore_at_phase_B (tuple(float, float)), Default = None. Same as previous, but opposite.
• verbose (bool), Default = False. If True, extra prints are done.
• iteration_limit (int), Default = 6. Maximum allowed iterations per spectrum for RV caculation subroutine. Amount is effectively doubled if refit_width_X is provided.
• convergence_limit (float), Default = 5E-3. Convergence criterium for RV in absolute units.
• rv_lower_limit (float), Default = 0.0. Used only for plotting the limits in broadening function plots if RoutineOptions.plot = True.

Next Page: Accessing outputs