Ctrl+K
Ctrl+K

The peakbagging module

The peakbagging module#

Fit functions#

apollinaire.peakbagging.stellar_framework(freq, psd, r=1, m=1, teff=5770, nwalkers=64, dnu=None, back=None, wdw=None, numax=None, Hmax=None, Wenv=None, epsilon=None, width=None, alpha=None, d02=None, d01=None, d13=None, n_harvey=2, low_cut=10.0, guess_back=None, low_bounds_back=None, up_bounds_back=None, frozen_param_back=None, power_law=False, high_cut_plaw=20.0, spectro=False, filename_back='background.png', back_mcmc=True, back_mle=False, filemcmc_back='mcmc_sampler_background.h5', nsteps_mcmc_back=1000, guess_pattern=None, low_bounds_pattern=None, up_bounds_pattern=None, pattern_mcmc=True, pattern_mle=False, n_order=3, n_order_peakbagging=3, h0_screening=False, format_cornerplot='pdf', filename_pattern='pattern.png', filemcmc_pattern='mcmc_sampler_pattern.h5', nsteps_mcmc_pattern=1000, parallelise=False, fit_l1=True, fit_l3=False, use_numax_back=False, progress=False, a2z_file='modes_param.a2z', a2z_in=None, nsteps_mcmc_peakbagging=1000, mcmcDir='.', instr='geometric', amp_l=[1.0, 1.5, 0.7, 0.2], filename_peakbagging='summary_peakbagging.png', nopeakbagging=False, discard_back=200, discard_pattern=200, discard_pkb=200, thin_back=1, thin_pattern=1, thin_pkb=1, quickfit=False, num=500, num_numax=500, reboxing_behaviour='advanced_reboxing', reboxing_strategy='median', save_only_after_sampling=False, apodisation=False, fit_angle=False, guess_angle=90, fit_splittings=False, fit_splittings_per_order_in_global=False, guess_split=0, fit_amplitude_ratio=False, frozen_harvey_exponent=False, fit_amp=False, extended=False, author=None, pkb_fmt=None, projected_splittings=False, bins=100, existing_chains_back='read', existing_chains_pattern='read', existing_chains_pkb='read', strategy='order', common_width=False, save_background_vector=True, dpi=300, show_corner=True, estimate_autocorrelation=False, plot_datapoints=True, show_gaussian=True)#

Framework for stellar peakbagging considering only a few input parameters. Background, global pattern and individual mode parameters are successively fitted.

Parameters:

  • freq (ndarray) – frequency vector

  • psd (ndarray) – real power or power spectral density vector

  • r (float) – stellar radius. Given in solar radius. Optional, default 1.

  • m (float) – stellar mass. Given in solar masses. Optional, default 1.

  • teff (float) – stellar effective temperature. Given in Kelvins. Optional, default 5770.

  • nwalkers (int) – number of walkers for the MCMC process. Optional, default 64.

  • dnu (float) – large frequency separation. Must be given in µHz. If None, will be computed through scaling laws. Optional, default None.

  • back (ndarray) – array of fitted background, of same length as freq and psd. If given, no fit will be performed on background and the pattern step will directly take place. Hmax, Wenv and numax must in this case also be given. Optional, default None.

  • wdw (ndarray) – observational window of the light curve to take into account the side lobes pattern in the modes. Must be 1 when an actuak observation is acquired and zero otherwise. Optional, default None.

  • numax (float) – maximum of power in p-mode envelope. Must be given in µHz. Optional, default None.

  • Hmax (float) – amplitude of p-mode envelope. Optional, default None.

  • Wenv (float) – width of p-mode envelope. Optional, default None.

  • epsilon (float) – epsilon guess value of mode patterns. If given, will override the value computed by the automatic guess. Optional, default None.

  • width (float) – width guess value of mode patterns. If given, will override the value computed by the automatic guess. Optional, default None.

  • alpha (float) – alpha value for global pattern guess. If specified, this value will override guess value computed by the automatic guess. Optional, default None.

  • d02 (float) – d02 value for global pattern guess. If specified, this value will override guess value computed by the automatic guess. Optional, default None.

  • d01 (float) – d01 value for global pattern guess. If specified, this value will override guess value computed by the automatic guess. Optional, default None.

  • d13 (float) – d13 value for global pattern guess. If specified, this value will override guess value computed by the automatic guess. Optional, default None.

  • n_harvey (int) – number of Harvey laws to use to build the background model. Optional, default 2. With more than two Harvey laws, it is strongly recommended to manually feed the ‘guess’ parameter.

  • guess_back (array-like.) – first guess directly passed by the user. If guess is None, the function will automatically infer a first guess. Optional, default None. Backgrounds parameters given in the following order: param_harvey (3 n_harvey ), param_plaw (2), param_gaussian (3), white noise (1). You can use create_background_guess_arrays to create guess and bounds arrays with proper structure.

  • low_bounds_back (ndarray) – lower bounds to consider in the background parameters space exploration. Must have the same dimension than guess_back.

  • up_bounds_back (ndarray) – upper bounds to consider in the background parameters space exploration. Must have the same dimension than guess_back.

  • frozen_param_back (boolean array) – boolean array of the same size as guess_back. Components set to True will not be fitted.

  • power_law (bool) – set to True to fit a power law on the background. Optional, default False.

  • high_cut_plaw (float) – if power_law=True, the function will consider the PSD segment between low_cut and high_cut_plaw to compute the power law parameters initial guess.

  • spectro (bool) – if set to True, make the plot with unit consistent with radial velocity, else with photometry. Optional, default False.

  • filename_back (str) – name of the of the dat file of the fitted background, the background parameters pdf file where the plot will be stored. Optional, default background.

  • back_mcmc (bool) – If set to False, no MCMC sampling will be performed for the background. Optional, default True.

  • back_mle (bool) – If set to False, no MLE optimisation will be performed for the background. Optional, default True.

  • filemcmc_back (str) – Name of the hdf5 file where the MCMC background chain will be stored. Optional, default mcmc_sampler_background.h5.

  • nsteps_mcmc_back (int) – number of MCMC steps for the background parameters exploration. Optional, default 1000.

  • guess_pattern (array-like.) – first guess directly passed by the user. If guess is None, the function will automatically infer a first guess. Optional, default None. Pattern parameters have to be given in the following order: eps, alpha, Dnu, numax, Hmax, Wenv, w, d02, b02, d01, b01, d13, b03. A first version of the guess and bounds array can be created through the create_pattern_guess_arrays.

  • low_bounds_pattern (ndarray) – lower bounds to consider in the background parameters space exploration. Must have the same dimension than guess_pattern.

  • up_bounds_pattern (ndarray) – upper bounds to consider in the background parameters space exploration. Must have the same dimension than guess_pattern.

  • pattern_mle (bool) – If set to False, no MLE optimisation will be performed for the pattern. Optional, default True.

  • pattern_mcmc (bool) – If set to False, no MCMC sampling will be performed for the pattern. Optional, default True.

  • n_order (int) – number of orders to consider for the global pattern fit. Optional, default 3.

  • n_order_peakbagging (int) – number of orders to fit at the individual mode parameters step. Optional, default 3.

  • h0_screening (bool) – if set to True, a fast frequency screening of the p-mode region will be performed in order to determine the mode that it possible to fit. n_order_peakbagging value will be ignored in this case. Optional, default False.

  • format_cornerplot (str) – set cornerplot file format. Optional, default pdf.

  • filename_pattern (str) – name of the pdf file where the fitted global pattern parameters and the plot of the fitted global pattern will be stored. Optional, default pattern.

  • filemcmc_pattern (str) – Name of the hdf5 file where the MCMC pattern chain will be stored. Optional, default mcmc_sampler_pattern.h5.

  • nsteps_mcmc_pattern (int) – number of MCMC steps for the pattern parameters exploration. Optional, default 1000.

  • parallelise (bool) – if set to True, multiprocessing will be used to sample the MCMC. Optional, default False.

  • fit_l1 (bool) – set to True to fit the d01 and b03 param and create guess for l=1 modes. Optional, default True.

  • fit_l3 (bool) – set to True to fit the d03 and b03 param and create guess for l=3 modes. Optional, default False. fit_l1 must be set to True.

  • use_numax_back (bool) – if set to True, the numax values computed with the background fit will be used as input guess for the pattern fit. Otherwise, the initial numax for the pattern fit will be taken from the numax argument or computed with the scaling laws.

  • progress (bool) – show progress bar in terminal. Optional, default False.

  • a2z_file (str) – name of the a2z file where the individual parameters will be stored. This will also be used to name the output pkb file. Optional, default modes_param.a2z

  • a2z_in (str) – a2z file with guess input for individual parameters. If provided, the global pattern step will be ignored and the function will directly sample individual modes parameters (after sampling the background if it has not been provided too). Optional, default None.

  • nsteps_mcmc_peakbagging (int) – number of MCMC steps for the peakbagging parameters exploration. Optional, default 1000.

  • mcmcDir (str) – Name of the directory where the MCMC of the individual parameters should be stored. Optional, default ..

  • instr (str) – name of the instrument for which m-amplitude ratios will be computed. geometric, golf and virgo are available.

  • amp_l (array-like) – amplitude ratio between harmonic degrees (relative to l=0 mode). Optional, default [1., 1.5, 0.7, 0.2].

  • filename_peakbagging (str) – name of the file where the plot summary of the individual mode parameters peakbagging will be stored. Optional, default summary_peakbagging.pdf.

  • nopeakbagging (bool) – if set to True, individual modes parameters will not be fitted. Optional, default False.

  • discard_back (int) – number of step to discard for the background sampling. Optional, default 200.

  • discard_pattern (int) – number of step to discard for the pattern sampling. Optional, default 200.

  • discard_pkb (int) – number of step to discard for the peakbagging sampling. Optional, default 200.

  • thin_back (int) – take only every thin steps from the chain sampled for backgrounds parameters. Optional, default 10.

  • thin_pattern (int) – take only every thin steps from the chain sampled for pattern parameters. Optional, default 10.

  • quickfit (bool) – if set to True, the fit will be performed over a smoothed and logarithmically resampled background. Optional, default False.

  • num (int) – Only considered if quickfit=True. Number of point in the resampled PSD, or target number of point for region away from numax if advanced_reboxing is selected. Optional, default 5000.

  • num_numax (int) – Only considered if quickfit=True and reboxing_behaviour=advanced_reboxing. Number of resampling points in the numax region. Optional, default 1000.

  • reboxing_behaviour (str) – behaviour for quickfit new power vector computation. It can be smoothing, reboxing or advanced_reboxing. Optional, default advanced_reboxing.

  • reboxing_strategy (str) – reboxing strategy to take box values when using quickfit and reboxing_behaviour=reboxing. Can be median or mean. Optional, default median.

  • save_only_after_sampling – if set to True, hdf5 file with chains information will only be saved at the end of the sampling process. If set to False, the file will be saved step by step (see emcee documentation).

  • apodisation (bool) – if set to True, distort the model to take the apodisation into account. Optional, default False.

  • fit_angle (bool) – if set to True, the peakbagging process will include inclination angle of the star in the parameters to fit. Optional, default False.

  • guess_angle (float) – initial guess value for angle. Will be taken into account only if fit_angle is True. Optional, default 90.

  • fit_splittings (bool) – if set to True, the peakbagging process will include mode splittings in the parameters to fit. Optional, default False.

  • fit_splittings_per_order_in_global (bool) – if set to True and fit_splittings=True with strategy='global', the peakbagging process will include one mode splittings per order in the parameters to fit. Optional, default False.

  • guess_split (float) – initial guess value for splittings. Will be taken into account only if fit_splittings is True. Optional, default 0.

  • fit_amplitude_ratio (bool) – if the selected strategy is global, the global amp_l amplitude ratio between degrees parameters will be fitted only if this parameter is set to True. Optional, default False.

  • frozen_harvey_exponent (bool) – set to True to have the Harvey models exponent fixed to 4. Optional, default False.

  • fit_amp (bool) – if set to True, amplitude of the modes will be fitted instead of heights (input heights will be transformed inside the code and retransformed as outputs).

  • extended (bool) – if set to True, the returned pkb array will be extended to contain asymmetric uncertainties over each parameter.

  • author (str) – identifier of the peakbagger, if given it will be specified in the header of the output pkb file. Optional, default None.

  • pkb_fmt – float formatting of the pkb file that will be created by the function. Optional, default None.

  • projected_splittings (bool) – if set to True, the function will consider that the split parameters of the input are projected splittings and will build the model consequently. Optional, default False.

  • bins (int) – number of bins for each cornerplot panel. Optional, default 100.

  • existing_chains_back – controls the behaviour of the function concerning a background parameters existing hdf5 file. If, read, existing files will be read without sampling and function output will be updated consequently, if reset, the backend will be cleared and the chain will be sampled from scratch, if sample the function will sample the chain from where the previous exploration was stopped, if ignore, the chain is not read. Optional, default read. The option ignore will apply only to individual mode chains, it will be changed to read for background and pattern steps.

  • existing_chains_pattern (str) – Same as existing_chains_back but for a global pattern parameters existing hdf5 file.

  • existing_chains_pkb (str) – Same as existing_chains_back but for individual mode parameters existing hdf5 files.

  • strategy (str) – fitting strategy for individual modes. Can be order or global. Optional, default order. Note that the global behaviour is experimental.

  • common_width (bool) – if set to True, only a global width parameter will be set in the a2z DataFrame. Optional, default False.

  • save_background_vector (bool) – the full background vector will be saved only if this option is set to True. Optional, default True.

  • dpi (int) – dot per inch value for figures.

  • show_corner (bool) – if set to True, corner plots will be shown and saved. Optional, default True.

  • estimate_autocorrelation (bool) – if set to True, the autocorrelation time of the sampled chains will be estimated by emcee. Optional, default False.

  • plot_datapoints (bool) – data points outside contours will be drawn if set to True. Optional, default True.

  • show_gaussian (bool) – If set to True, show p-mode acoustic hump modelled by a Gaussian function in the background summary plot. Optional, default True.

Returns:

None

apollinaire.peakbagging.peakbagging(a2z_file, freq, psd, back=None, wdw=None, dnu=None, spectro=True, nsteps_mcmc=1000, discard=200, show_corner=False, store_chains=False, mcmcDir='.', order_to_fit=None, parallelise=False, progress=False, strategy='order', fit_02=True, fit_13=True, nwalkers=64, normalise=False, instr='geometric', show_summary=False, filename_summary=None, show_prior=False, use_sinc=False, asym_profile='nigam-kosovichev', thin=1, restrict=False, remove_asymmetry_outside_window=True, do_not_use_dnu=False, save_only_after_sampling=False, size_window=None, fit_amp=False, extended=False, projected_splittings=False, bins=100, existing_chains='read', fit_splittings=True, fit_angle=True, fit_amplitude_ratio=False, dpi=300, format_cornerplot='pdf', estimate_autocorrelation=False, plot_datapoints=True)#

Perform the peakbagging process for a given set of modes.

Parameters:

  • a2z_file (string) – name of the file to read the parameters.

  • freq (ndarray) – frequency vector

  • psd (ndarray) – real power vector

  • back (ndarray.) – activity background vector that will be used to complete the model to fit. Optional default None. Must have the same length than freq and psd.

  • wdw (ndarray.) – observation window (0 and 1 array of the same lenght as the original timeseries) to analyse in order to predict the sidelob pattern. Optional, default None.

  • dnu (float.) – large separation of the modes. In this function, it will only be used to adapt the fitting window for fitted modes above 1200 muHz. If not given and if at least two l=0 modes are specified in the a2z file, dnu will be automatically computed.

  • nsteps_mcmc (int) – number of steps to process into each MCMC exploration.

  • discard (int) – number of steps to discard into each MCMC exploration.

  • show_corner (bool) – if set to True, show and save the corner plot for the posterior distribution sampling.

  • store_chains (bool) – if set to True, each MCMC sampler will be stored as an hdf5 files. Filename will be autogenerated with modes parameters. Optional, default False

  • mcmcDir (string) – directory where to save the MCMC sampler. Optional, default ..

  • order_to_fit (array-like) – list of order to fit if the input a2z file contains order that are supposed not to be fitted. Optional, default None.

  • parallelise (bool) – If set to True, use Python multiprocessing tool to parallelise process. Optional, default False.

  • strategy (str) – strategy to use for the fit, pair, order, global. Optional, default order. If ‘pair’ is used, a2z input must contain individual heights, widths and splittings for each degree. The global behaviour is experimental.

  • fit_02 (bool) – if strategy is set to pair, l=0 and 2 modes will only be fitted if this parameter is set to True. Optional, default True.

  • fit_13 (bool) – if strategy is set to pair, l=1 and 3 modes will only be fitted if this parameter is set to True. Optional, default True.

  • nwalkers (int) – number of walkers to use for the sampling.

  • instr (str) – instrument to consider (amplitude ratio inside degrees depend on geometry AND instrument and should be adaptated). Possible argument : geometric, kepler, golf, virgo. Optional, default geometric.

  • show_summary (bool) – show summary plot at the end of the fit. Optional, default False.

  • use_sinc (bool) – if set to True, mode profiles will be computed using cardinal sinus and not Lorentzians. No asymmetry term will be used if it is the case. Optional, default False.

  • asym_profile (str) – depending on the chosen argument, asymmetric profiles will be computed following Korzennik 2005 (korzennik) or Nigam & Kosovichev 1998 (nigam-kosovichev). Default nigam-kosovichev.

  • restrict (bool) – if set to True, will only use the modes with unfrozen parameters in the model.

  • remove_asymmetry_outside_window (bool) – if set to True, asymmetry of modes outside of the fitting window will be set to 0. Optional, default True.

  • do_not_use_dnu (bool) – if set to True, fitting window will be computed without using dnu value.

  • save_only_after_sampling – if set to True, hdf5 file with chains information will only be saved at the end of the sampling process. If set to False, the file will be saved step by step (see emcee documentation).

  • size_window (float) – size of the fitting window, in muHz. If not given, the size of the fitting window will be automatically set, using dnu or the input frequencies of the parameter to fit. Optional, default None.

  • fit_amp (bool) – if set to True, amplitude of the modes will be fitted instead of heights (input heights will be transformed inside the code and retransformed as outputs).

  • extended (bool) – if set to True, the returned pkb array will be extended to contain asymmetric uncertainties over each parameter.

  • projected_splittings (bool) – if set to True, the function will consider that the split parameters of the input are projected splittings and will build the model consequently. Optional, default False.

  • bins (int) – number of bins for each cornerplot panel. Optional, default 100.

  • existing_chains (str) – controls the behaviour of the function concerning existing hdf5 files. If, read, existing files will be read without sampling and function output will be updated consequently, if reset, the backend will be cleared and the chain will be sampled from scratch, if sample the function will sample the chain from where the previous exploration was stopped, if ignore, the chain is not read. Optional, default read.

  • fit_splittings (bool) – if the selected strategy is global, the global split parameter will be fitted only if this parameter is set to True. Optional, default True.

  • fit_angle (bool) – if the selected strategy is global, the global angle parameter will be fitted only if this parameter is set to True. Optional, default True.

  • fit_amplitude_ratio (bool) – if the selected strategy is global, the global amp_l amplitude ratio between degrees parameters will be fitted only if this parameter is set to True. Optional, default True.

  • dpi (int) – dot per inch value for figures.

  • estimate_autocorrelation (bool) – if set to True, the autocorrelation time of the sampled chains will be estimated by emcee. Optional, default False.

  • plot_datapoints (bool) – data points outside contours will be drawn if set to True. Optional, default True.

Returns:

a2z fitted modes parameters as a DataFrame, and corresponding pkb array (extended pkb array if extended is set to True)

Return type:

tuple of objects

apollinaire.peakbagging.perform_mle_background(freq, psd, n_harvey=2, r=1.0, m=1.0, teff=5770.0, dnu=None, numax=None, guess=None, frozen_param=None, power_law=False, gaussian=True, frozen_harvey_exponent=False, low_cut=100.0, method=<function _minimize_powell>, fit_log=False, low_bounds=None, up_bounds=None, no_bounds=False, show=True, show_guess=False, filename=None, spectro=True, quickfit=False, num=5000, num_numax=500, two_gaussian=False, reboxing_behaviour='advanced_reboxing', reboxing_strategy='median', apodisation=False, high_cut_plaw=20.0, show_gaussian=True, **kwargs)#

Perform MLE over background model.

Parameters:

  • freq (ndarray) – frequency vector in µHz.

  • psd (ndarray) – power density vector in ppm^2/µHz or (m/s)^2/muHz.

  • n_harvey (int) – number of Harvey laws to use to build the background model. Optional, default 2. With more than two Harvey laws, it is strongly recommended to manually feed the ‘guest’ parameter.

  • r (float) – stellar radius. Given in solar radius. Optional, default 1.

  • m (float) – stellar mass. Given in solar mass. Optional, default 1.

  • teff (float) – stellar effective temperature. Given in K. Optional, default 5770.

  • dnu (float) – large separation of the p-mode. If given, it will superseed the scaling law guess using r, m and teff.

  • numax (float) – maximum p-mode bump power frequency. If given, it will superseed the scaling law guess using r, m and teff.

  • guess (array-like.) – first guess directly passed by the users. If guess is None, the function will automatically infer a first guess. Optional, default None. Backgrounds parameters given in the following order: param_harvey (3 n_harvey ), param_plaw (2), param_gaussian (3), white noise (1)

  • frozen_param (boolean array) – boolean array of the same size as guess. Components set to True will not be fitted.

  • power_law (bool) – set to True to fit a power law on the background. Optional, default False.

  • gaussian (bool) – set to True to fit the p-mode gaussian on the background. Optional, default True.

  • frozen_harvey_exponent (bool) – set to True to freeze and not fit Harvey laws exponent to 4. Optional, default False.

  • low_cut (float) – Spectrum below this frequency will be ignored for the fit. The frequency value must be given in µHz. Optional, default 100.

  • method – minimization method used by the scipy minimize function. Optional, default _minimize_powell (modified version allowing to use bounds)

  • fit_log (bool) – if set to True, fit natural logarithm of the parameters. Optional, default False.

  • low_bounds (ndarray) – lower bounds to consider in the parameter space exploration. Must have the same structure than guess.

  • up_bounds (ndarray) – upper bounds to consider in the parameter space exploration. Must have the same structure than guess.

  • no_bounds (bool) – If set to True, no bounds will be considered for the parameter space exploration. Optional, default False.

  • show (bool) – if set to True, will show at the end a plot summarising the fit. Optional, default True.

  • show_guess (bool) – if set to True, will show at the beginning a plot summarising the guess. Optional, default False.

  • filename (str) – if given, the summary plot will be saved under this filename. Optional, default None. show argument must have been set to True.

  • spectro (bool) – if set to True, make the plot with unit consistent with radial velocity, else with photometry. Automated guess will also be computed consistently with spectroscopic measurements. Optional, default True.

  • quickfit (bool) – if set to True, the fit will be performed over a smoothed and logarithmically resampled background. Optional, default False.

  • num (int) – Only considered if quickfit=True. Number of point in the resampled PSD, or target number of point for region away from numax if advanced_reboxing is selected. Optional, default 5000.

  • num_numax (int) – Only considered if quickfit=True and reboxing_behaviour=advanced_reboxing. Number of resampling points in the numax region. Optional, default 500.

  • reboxing_behaviour (str) – behaviour for quickfit new power vector computation. It can be smoothing, reboxing or advanced_reboxing. Optional, default advanced_reboxing.

  • reboxing_strategy (str) – reboxing strategy to take box values when using quickfit and reboxing_behaviour=reboxing. Can be median or mean. Optional, default median.

  • apodisation (bool) – if set to True, distort the model to take the apodisation into account. Optional, default False.

  • high_cut_plaw (float) – high cut in frequency to compute the power law guess. Optional, default 20.

  • show_gaussian (bool) – If set to True, show p-mode acoustic hump modelled by a Gaussian function in the summary plot. Optional, default True.

Returns:

fitted background and fitted background parameters.

Return type:

tuple of array

apollinaire.peakbagging.explore_distribution_background(freq, psd, n_harvey=2, r=1.0, m=1.0, teff=5770.0, dnu=None, numax=None, guess=None, frozen_param=None, power_law=False, gaussian=True, frozen_harvey_exponent=False, low_cut=100.0, fit_log=False, low_bounds=None, up_bounds=None, spectro=True, show=True, show_guess=False, show_corner=True, nsteps=1000, discard=200, filename=None, parallelise=False, progress=False, nwalkers=64, filemcmc=None, thin=1, quickfit=False, num=5000, num_numax=500, estimate_autocorrelation=False, reboxing_behaviour='advanced_reboxing', reboxing_strategy='median', format_cornerplot='pdf', save_only_after_sampling=False, apodisation=False, high_cut_plaw=20.0, bins=100, two_gaussian=False, existing_chains='read', norm=None, plot_datapoints=True, xlim=None, ylim=None, show_gaussian=True, **kwargs)#

Use a MCMC framework to fit the background.

Parameters:

  • freq (ndarray) – frequency vector

  • psd – real power vector

  • n_harvey (int) – number of Harvey laws to use to build the background model. Optional, default 2. With more than two Harvey laws, it is strongly recommended to manually feed the guess parameter.

  • r (float) – stellar radius. Given in solar radius. Optional, default 1.

  • m (float) – stellar mass. Given in solar mass. Optional, default 1.

  • teff (float) – stellar effective temperature. Given in K. Optional, default 5770.

  • dnu (float) – large separation of the p-mode. If given, it will superseed the scaling law guess using r, m and teff.

  • numax (float) – maximum p-mode bump power frequency. If given, it will superseed the scaling law guess using r, m and teff.

  • guess (array-like.) – first guess directly passed by the user. If guess is None, the function will automatically infer a first guess. Optional, default None. Backgrounds parameters given in the following order: param_harvey (3 n_harvey ), param_plaw (2), param_gaussian (3), white noise (1)

  • frozen_param (boolean array) – boolean array of the same size as guess. Components set to True will not be fitted.

  • power_law (bool) – set to True to fit a power law on the background. Optional, default False.

  • gaussian (bool) – set to True to fit the p-mode gaussian on the background. Optional, default True.

  • frozen_harvey_exponent (bool) – set to True to freeze and not fit Harvey laws exponent to 4. Optional, default False.

  • low_cut (float) – Spectrum below this frequency will be ignored for the fit. The frequency value must be given in µHz. Optional, default 100.

  • fit_log (bool) – if set to True, fit natural logarithm of the parameters. Optional, default False.

  • low_bounds (ndarray) – lower bounds to consider in the parameter space exploration. Must have the same structure than guess.

  • up_bounds (ndarray) – upper bounds to consider in the parameter space exploration. Must have the same structure than guess.

  • spectro (bool) – if set to True, make the plot with unit consistent with radial velocity, else with photometry. Automated guess will also be computed consistently with spectroscopic measurements. Optional, default True.

  • filename (string) – Name of the file to save the summary plot. If filename is None, the name will not be stored. Optional, default None.

  • filemcmc – name of the hdf5 where to store the chain. If filename is None, the name will not be stored. Optional, default None.

  • parallelise (bool) – If set to True, use Python multiprocessing tool to parallelise process. Optional, default False.

  • show (bool) – if set to True, will show at the end a plot summarising the fit. Optional, default True.

  • show_corner – if set to True, will show the corner plot summarising the MCMC process. Plot will be saved as a pdf is filemcmc is also specified. Optional, default True.

  • nsteps (int) – number of MCMC steps to perform.

  • discard (int) – number of step to discard in the sampling. Optional, default 200.

  • show_guess (bool) – if set to True, will show at the beginning a plot summarising the guess. Optional, default False.

  • thin (int) – take only every thin steps from the chain. Optional, default 1.

  • quickfit (bool) – if set to True, the fit will be performed over a smoothed and logarithmically resampled background. Optional, default False.

  • num (int) – Only considered if quickfit=True. Number of point in the resampled array, or target number of point for region away from numax if advanced_reboxing is selected. Optional, default 5000.

  • num_numax (int) – Only considered if quickfit=True and reboxing_behaviour=advanced_reboxing. Number of resampling points in the numax region. Optional, default 1000.

  • reboxing_behaviour (str) – behaviour for quickfit new power vector computation. It can be smoothing, reboxing or advanced_reboxing. Optional, default advanced_reboxing.

  • reboxing_strategy (str) – reboxing strategy to take box values when using quickfit and reboxing_behaviour=reboxing. Can be median or mean. Optional, default median.

  • save_only_after_sampling – if set to True, hdf5 file with chains information will only be saved at the end of the sampling process. If set to False, the file will be saved step by step (see emcee documentation).

  • apodisation (bool) – if set to True, distort the model to take the apodisation into account. Optional, default False.

  • high_cut_plaw (float) – high cut in frequency to compute the power law guess. Optional, default 20.

  • bins (int) – number of bins for each cornerplot panel. Optional, default 100.

  • existing_chains (str) – controls the behaviour of the function concerning existing hdf5 files. If, read, existing files will be read without sampling and function output will be updated consequently, if reset, the backend will be cleared and the chain will be sampled from scratch, if sample the function will sample the chain from where the previous exploration was stopped. Optional, default read.

  • estimate_autocorrelation (bool) – if set to True, the autocorrelation time of the sampled chains will be estimated by emcee. Optional, default False.

  • plot_datapoints (bool) – data points outside contours will be drawn if set to True. Optional, default True.

  • show_gaussian (bool) – If set to True, show p-mode acoustic hump modelled by a Gaussian function in the summary plot. Optional, default True.

Returns:

the fitted background, its param and their sigma values as determined by the MCMC walk.

Return type:

tuple of array

Chain management functions#

apollinaire.peakbagging.read_chain(filename, thin=1, discard=0, read_order=True, chain_type='peakbagging', fit_amp=False, projected_splittings=False)#

Read a chain created by apollinaire and return flattened sampled chain with auxiliary informations.

Parameters:

  • filename (str) – name of the hdf5 file with the stored chain.

  • thin (int) – the returned chain will be thinned according to this factor. Optional, default 1.

  • discard (int) – number of iteration step to discard. Optional, default 0.

  • read_order (bool) – set to False when reading a peakbagging chain created with the strategy=global. The order output will be -9999 in this case.

  • chain_type (str) – can be peakbagging, pattern or background. Optional, default peakbagging.

  • fit_amp (bool) – set to True if amplitudes were fitted instead of heights.

  • projected_splittings (bool) – set to True if projected splittings were fitted instead of splittings.

Returns:

tuple with flatchain, param labels, degrees of param and order of modes if chain_type=peakbagging, flatchain and labels otherwise.

Return type:

tuple of ndarrays

apollinaire.peakbagging.make_cornerplot(flatchain, labels=None, filename=None, bins=10, figsize=(24, 24), labelsize=20, use_formatter=False, reverse=False, title_fmt='.4f', **kwargs)#

Make a cornerplot from a flattened chain. This is simply a wrapper for corner.corner.

Parameters:

  • flatchain (ndarray) – the flattened chain.

  • labels (array-like) – parameters labels

  • filename (str) – name of the file where the cornerplot will be saved.

  • bins (int) – number of bins in the histograms. Optional, default 10.

apollinaire.peakbagging.hdf5_to_a2z(workDir='.', a2zname='summary_fit.a2z', discard=0, thin=10, instr='geometric', centile=False, verbose=False)#

Create a a2z file with the hdf5 files stored in a given folder.

Parameters:

  • workDir (str) – the directory where the hdf5 file will be read.

  • a2zname – the name of the output file. Set to None if you do not want to save any file.

  • discard (int) – the number of elements to ignore at the beginning of the chain.

  • thin (int) – one element of the chain every thin elements will be considered.

  • centile (bool) – if set to True, change the structure of the output to write a centile frame. Optional, default False.

  • instr (str) – instrument to consider. Possible argument : geometric, kepler, golf, virgo. Optional, default geometric.

  • verbose (bool) – if set to True, print the name of the considered files. Optional, default False.

Returns:

a2z DataFrame

Return type:

pandas DataFrame

apollinaire.peakbagging.hdf5_to_pkb(workDir='.', pkbname='summary_fit.pkb', discard=0, thin=10, instr='geometric', extended=False, verbose=False)#

Create a pkb file with the hdf5 files stored in a given folder.

Parameters:

  • workDir (str) – the directory where the hdf5 file will be read.

  • pkbname (str) – the name of the output pkb. Set to None if you do not want to save a file.

  • discard (int) – the number of elements to ignore at the beginning of the chain.

  • thin (int) – one element of the chain every thin elements will be considered.

  • extended (bool) – if set to True, change the structure of the output to write a pkb extended. Optional, default False.

  • instr (str) – instrument to consider. Possible argument : geometric, kepler, golf, virgo. Optional, default geometric.

  • verbose (bool) – if set to True, print the name of the considered files. Optional, default False.

Returns:

pkb array

Return type:

ndarray

apollinaire.peakbagging.chain_to_a2z(filename, thin=1, discard=0, add_ampl=False, instr='geometric')#

Function to create an a2z dataframe from a sampled chain.

Parameters:

  • filename (str) – name of the chain. The name format is the following mcmc_sampler_order_[n_order]_degrees_[n_degrees].h5 with n_degrees being 13 or 02.

  • instr (str) – instrument to consider. Possible argument : geometric, kepler, golf, virgo. Optional, default geometric.

Returns:

a2z style dataframe.

Return type:

pandas DataFrame

Input/output files management functions#

apollinaire.peakbagging.read_a2z(a2z_file)#

Read a file with a a2z standard syntax (doc to be written) and return a2z-style parameters (sorted by orders and degrees).

Parameters:

a2z_file (string) – name of the file to read the parameters.

Returns:

input parameters as a pandas DataFrame with the a2z syntax.

Return type:

pandas DataFrame

apollinaire.peakbagging.save_a2z(filename, df)#

Write with a a2z standard syntax (doc to be written).

Parameters:

filename (string) – name of the file where to write the parameters.

apollinaire.peakbagging.save_pkb(filename, pkb, author=None, spectro=False, extended=False, fmt=None, projected_splittings=False, nwalkers=None, nsteps=None, discard=None, fit_amp=False)#

Save pkb file with dedicated header.

apollinaire.peakbagging.a2z_to_pkb(df_a2z, nopandas=True)#

Take a2z dataframe and return pkb_style parameters. Frequency units are given in µHz and not Hz. pkb format is the following:

parameters

n

l

nu

nu_error

height

height_error

width

width_error

angle

angle_error

split

split_error

asym

asym_error

units

.

.

µHz

µHz

power/µHz

power/µHz

µHz

µHz

degree

degree

µHz

µHz

.

.
Parameters:

df_a2z (pandas DataFrame) – input parameters as a pandas DataFrame with the a2z syntax.

Returns:

array under pkb format.

Return type:

ndarray

apollinaire.peakbagging.test_a2z(a2z_file, verbose=True)#

Test a a2z file to check that it is valid. The function checks the bounds set for the parameters, convert the a2z DataFrame to pkb

Parameters:

a2z_file (str) – path of the a2z file to test.

Returns:

a2z DataFrame and pkb array.

Return type:

tuple

Quality assurance functions#

apollinaire.peakbagging.bayes_factor(freq, psd, back, df_a2z, n, strategy='pair', l02=True, dnu=None, do_not_use_dnu=True, size_window=None, thin=10, discard=0, instr='geometric', use_sinc=False, asym_profile='nigam-kosovichev', hdf5Dir='./', wdw=None, add_ampl=False, parallelise=False, fit_amp=False, projected_splittings=False, nodes=4)#
Parameters:

  • freq (ndarray) – frequency vector

  • psd (ndarray) – real power vector

  • back (ndarray) – model background power vector.

  • df_a2z (pandas DataFrame) – a2z input DataFrame that was used for the fit. The DataFrame is only used to compute the window inside which the fit was performed.

  • n (int) – mode order for which the statistical test will evaluate the model.

  • strategy (str) – strategy that was used for the fit, pair or order. Optional, default pair.

  • l02 (bool) – set to True if strategy is pair and if the pair to consider is even. Set to False if the fitted mode was odd.

  • dnu (float.) – large separation of the modes. In this function, it will only be used to adapt the fitting window for fitted modes above 1200 muHz. If not given, it will be computed (if possible) with the input df_a2z. Optional, default None.

  • do_not_use_dnu (bool) – if set to True, fitting window will be computed without using dnu value. Optional, default True.

  • size_window (float) – size of the window that was used for the fit, it it was given manually to the peakbagging function. Optional, default None.

  • thin (int) – one element of the chain every thin elements will be considered.

  • discard (int) – the number of elements to ignore at the beginning of the chain.

  • use_sinc (bool) – if set to True, mode profiles will be computed using cardinal sinus and not Lorentzians. No asymmetry term will be used if it is the case. Optional, default False.

  • asym_profile (str) – depending on the chosen argument, asymmetric profiles will be computed following Korzennik 2005 (korzennik) or Nigam & Kosovichev 1998 (nigam-kosovichev). Default nigam-kosovichev.

  • hdf5Dir (str) – directory where the hdf5 file storing the MCMC should be read. Optional, default ‘./’.

  • wdw (ndarray) – Observation window of the considered timeseries. Optional, default None.

  • add_ampl (bool) – if set to True, standard values will be used to compute the mode height ratios (if heights have not been specified individually for each degree). Optional, default False.

  • parallelise (bool) – set to True to parallelise process. Optional, default False.

Returns:

tuple of float, in the following order: probability of model with two modes, probability of model with only the strong mode, probaility of the H0 hypothesis, number of models on which the probability has been estimated.

Return type:

ndarray

Plot functions#

apollinaire.peakbagging.plot_from_param(param_pkb, freq, psd, back=None, wdw=None, smoothing=50, spectro=True, correct_width=1.0, show=False, filename=None, instr='geometric', use_sinc=False, asym_profile='korzennik', projected_splittings=False, **kwargs)#

Plot the results of a fit according to an input given with a pkb format.

Parameters:

  • param_pkb (ndarray) – parameters contained in the pkb files.

  • freq (ndarray) – frequency vector, must be given in muHz.

  • psd (ndarray) – real power vector of the observed data.

  • back (ndarray) – real power vector of the fitted background.

  • wdw (bool) – set to True if the mode have been fitted using the sidelobes fitting method, default False.

  • smoothing (int) – size of the rolling window used to smooth the psd in the plot.

  • spectro (bool) – set to True if the instruments uses spectroscopy, set the units in m/s instead of ppm, default True.

  • correct_width (float) – param to adjust the width of the Lorentzian if it has been manually modified during the fitting

  • show (bool) – automatically show the plot, default False.

  • instr (str) – instrument to consider (amplitude ratio inside degrees depend on geometry AND instrument and should be adaptated). Possible argument : geometric, kepler, golf, virgo. Optional, default geometric.

  • use_sinc (bool) – if set to True, mode profiles will be computed using cardinal sinus and not Lorentzians. No asymmetry term will be used if it is the case. Optional, default False.

  • asym_profile (str) – depending on the chosen argument, asymmetric profiles will be computed following Korzennik 2005 (korzennik) or Nigam & Kosovichev 1998 (nigam-kosovichev).

Returns:

None

apollinaire.peakbagging.banana_diagram(freq, psd, back, pkb, grid=None, n=30, k=30, angle_min=0, angle_max=90, split_min=0, split_max=2, figsize=(9, 6), shading='auto', marker_color='black', cmap='plasma', contour_color='black', marker='o', annotate_max=True, show_contour=True, param_wdw=None, instr='geometric', normalise_likelihood=True, use_sinc=False, asym_profile='nigam-kosovichev', fit_amp=False, projected_splittings=False, add_colorbar=False, levels=None, centile_max=95, centile_min=75, vmin=None, vmax=None, pcolormesh_options={}, contour_options={})#

Compute banana diagram (see Ballot et al 2006) for a given set of fitted parameters. Parameters are fixed, except for angles and splittings which are set to vary between 0 and 90 degrees and 0 and 2 muHz, respectively.

Parameters:

  • freq (ndarray) – frequency vector.

  • psd (ndarray) – psd vector.

  • back (ndarray) – background vector.

  • pkb (ndarray) – pkb array that will be used to compute the model and the likelihood.

  • grid (ndarray) – if grid is not None, no computation will be performed and the plot will directly be made with the array passed with this argument.

  • n (int) – number of elements along angle axis. Optional, default 30.

  • k (int) – number of elements along splittings axis. Optional, default 30.

  • angle_min – minimal value for angle on the diagram. Optional, default 0.

  • angle_max – maximal value for angle on the diagram. Optional, default 90.

  • split_min – minimal value for splittings on the diagram. Optional, default 0.

  • figsize (tuple) – size of the matplotlib Figure. Optional, default (9,6)

  • split_max – maximal value for splittings on the diagram. Optional, default 2.

  • shading (str) – shading parameter. Optional, default auto.

  • marker_color (str) – color of markers. Optional, default black.

  • cmap (str or colormap instance.) – colormap. Optional, default plasma.

  • contour_color (str) – color of contours. Optional, default black.

  • marker (str) – marker type. Optional, default o.

  • annotate_max (bool) – if True, coordinates of the maximum in the computed grid will be given as a legend. Optional, default True.

  • show_contour (bool) – whether to show contour or not. Optional, default True.

  • param_wdw (ndarray.) – window parameters to apply on the mode model.

  • instr (str) – instrument used to collect the time series. Optional, default geometric.

  • normalise_likelihood (bool) – if set to True, log likelihood will be divided by the maximum of its absolute value. Optional, default True.

  • pcolormesh_options (dict) – ditionary that will be passed as the kwargs for matplotlib.axes.Axes.pcolormesh.

  • contour_options (dict) – ditionary that will be passed as the kwargs for matplotlib.axes.Axes.contour.

Returns:

tuple with likelihood-grid and figure

Return type:

tuple

Model building functions#

apollinaire.peakbagging.compute_model(freq, param_pkb, param_wdw=None, correct_width=1.0, instr='kepler', use_sinc=False, asym_profile='nigam-kosovichev', fit_amp=False, projected_splittings=False)#

Compute a p-mode model from a given set of parameters.

Parameters:

  • freq (ndarray) – frequency vector.

  • param_pkb (ndarray) – parameters contained in the pkb files.

  • param_wdw (ndarray) – parameters given by the analysis of the window.

  • correct_width (float) – param to adjust the width of the Lorentzian if it has been manually modified during the fitting

  • instr (str) – instrument to consider (amplitude ratio inside degrees depend on geometry AND instrument and should be adaptated). Possible argument : geometric, kepler, golf, virgo.

  • use_sinc (bool) – if set to True, mode profiles will be computed using cardinal sinus and not Lorentzians. No asymmetry term will be used if it is the case. Optional, default False.

  • asym_profile (str) – depending on the chosen argument, asymmetric profiles will be computed following Korzennik 2005 (korzennik) or Nigam & Kosovichev 1998 (nigam-kosovichev).

  • fit_amp (bool) – if set to True, the function consider that it got amplitudes and not heights as input parameters. Optional, default False.

  • projected_splittings (bool) – if set to True, the function will consider that the split parameters of the input are projected splittings and will build the model consequently. Optional, default False.

Returns:

computed model

Return type:

ndarray

apollinaire.peakbagging.compute_model_from_pattern(freq, pattern, back, n_order=3, amp_l=[1.0, 1.5, 0.7, 2])#

Compute a p-mode model from global pattern parameters.

Parameters:

  • freq (ndarray) – frequency vector.

  • pattern (ndarray) – pattern parameters. Pattern parameters have to be given in the following order: eps, alpha, Dnu, numax, Hmax, Wenv, w, d02, b02, d01, b01, d13, b03. A first version of the guess and bounds array can be created through the create_pattern_guess_arrays.

  • back (ndarray) – background vector to use for the model.

  • n_order (int) – number of orders to consider around numax to build the model. Optional, default 3.

Returns:

computed model

Return type:

ndarray

apollinaire.peakbagging.build_background(freq, param, n_harvey=2, apodisation=False, remove_gaussian=True, two_gaussian=False)#

Rebuild background vector from complete set of parameters.

Parameters:

  • freq (ndarray) – frequency vector (in muHz).

  • param (array like) – 1D vector with backgrounds parameters given in the following order: param_harvey (3x``n_harvey``), param_plaw (2), param_gaussian (3), white noise (1)

  • n_harvey (int) – number of Harvey laws to use to build the background model. Optional, default 2.

  • apodisation (bool) – if set to True, distort the model to take the apodisation into account. Optional, default False.

  • remove_gaussian – if set to True, the p-mode gaussian bump will not be taken into account when building the vector. Optional, default True.

  • remove_gaussian – bool

Returns:

background array

Return type:

ndarray

Convenience functions#

apollinaire.peakbagging.create_background_guess_arrays(freq, psd, r=1.0, m=1.0, teff=5770.0, dnu=None, numax=None, n_harvey=2, spectro=False, power_law=False, high_cut_plaw=20.0, return_labels=False, two_gaussian=False)#

Create guess and bound arrays for background fit.

Parameters:

  • freq (ndarray) – frequency vector

  • psd – real power vector

  • n_harvey (int) – number of Harvey laws to use to build the background model. Optional, default 2. With more than two Harvey laws, it is strongly recommended to manually feed the guess parameter.

  • r (float) – stellar radius. Given in solar radius. Optional, default 1.

  • m (float) – stellar mass. Given in solar mass. Optional, default 1.

  • teff (float) – stellar effective temperature. Given in K. Optional, default 5770.

  • dnu (float) – large separation of the p-mode. If given, it will superseed the scaling law guess using r, m and teff.

  • numax (float) – maximum p-mode bump power frequency. If given, it will superseed the scaling law guess using r, m and teff.

  • power_law (bool) – set to True to fit a power law on the background. Optional, default False.

  • spectro (bool) – if set to True, make the plot with unit consistent with radial velocity, else with photometry. Automated guess will also be computed consistently with spectroscopic measurements. Optional, default True.

  • high_cut_plaw (float) – high cut in frequency to compute the power law guess. Optional, default 20.

  • return_labels (bool) – if set to True, the label array will be returned among the returned tuple of arrays.

Returns:

guess, low bounds, up bounds (and labels if return_labels=True) arrays.

Return type:

tuple of arrays

apollinaire.peakbagging.create_pattern_guess_arrays(Hmax, Wenv, r=1.0, m=1.0, teff=5770.0, dnu=None, numax=None, mcmc=True)#

Create guess and bound arrays for pattern fit.

Parameters:

  • Hmax (float) – height at numax.

  • Wenv (float) – width of the p-mode gaussian.

  • r (float) – stellar radius. Given in solar radius. Optional, default 1.

  • m (float) – stellar mass. Given in solar mass. Optional, default 1.

  • teff (float) – stellar effective temperature. Given in K. Optional, default 5770.

  • dnu (float) – large separation of the p-mode. If given, it will superseed the scaling law guess using r, m and teff.

  • numax (float) – maximum p-mode bump power frequency. If given, it will superseed the scaling law guess using r, m and teff.

  • mcmc (bool) – if set to True, will compute the guess values used for the MCMC exploration. Otherwise, will compute the guess values for the MLE optimisation. Optional, default True.

Returns:

guess, low bounds and up bounds array

Return type:

tuple of arrays

apollinaire.peakbagging.hide_modes(freq, psd, pkb, back, l=1, n_width=4, remove_left=True, remove_zero_bin=True)#

Hide PSD region around modes of given l.

Parameters:

  • freq (ndarray) – frequency vector

  • psd (ndarray) – power vector

  • back (ndarray) – background power vector

  • pkb (ndarray) – pkb array with modes parameters.

  • l (int) – mode degrees to remove. Optional, default 1.

  • n_width (float) – frequency bins will be removed n_width from the central frequency of the remaining modes. Optional, default 4.

  • remove_left (bool) – if set to True, remove a region Dnu/2 wide before the first mode given in the pkb. Optional, default True.

  • remove_zero_bin (bool) – if set to True, bins with zero values will be removed from returned vector. Optional, default True.

apollinaire.peakbagging.remove_pattern(freq, psd, back, pkb, l=[0, 2])#

Remove a given pattern from the PSD by dividing the PSD by the computed model.

Parameters:

  • freq (ndarray) – frequency vector

  • psd (ndarray) – power vector

  • back (ndarray) – background power vector

  • pkb (ndarray) – pkb array with modes parameters.

  • l (int or array-like) – mode degrees to remove. Optional, default [0,2].

Contents