Code Reference

afino_start

afino.afino_start.analyse_series(times, flux, description=None, low_frequency_cutoff=None, savedir=None, overwrite_gauss_bounds=None, overwrite_extra_gauss_bounds=None, use_json=True, model_ids=[0, 1, 2])

Analyse a single, generic timeseries using the AFINO model comparison code.

Parameters:
  • times (ndarray) – an array of times
  • flux (ndarray) – an array of data points
  • description (string, optional) – a string descriptor of the analysis run, that is incorporated into output filenames
  • low_frequency_cutoff (float, optional) – specifies a frequency above which the input Fourier spectrum is not analysed
  • savedir (string, optional) – specifies a directory for output save files
  • use_json (bool) – If set to True, saves analysis output in JSON format. If False, pickle format is used. Default is True.
afino.afino_start.analyse_series_twobump(times, flux, description=None, low_frequency_cutoff=None, savedir=None, overwrite_gauss_bounds=None, overwrite_extra_gauss_bounds=None)

Analyse a single, generic timeseries using the AFINO model comparison code., with extra models

afino.afino_start.create_generic_summary_plot(ts, analysis_summary, description, low_frequency_cutoff=None, savedir=None)

Create a summary plot showing the result of an AFINO analysis run.

afino.afino_start.create_generic_summary_plot_twobump(ts, analysis_summary, description, low_frequency_cutoff=None, savedir=None)

Create a summary plot showing the result of an AFINO analysis run.

afino_model_comparison

afino.afino_model_comparison.model_comparison(ts, description=None, low_frequency_cutoff=None, overwrite_gauss_bounds=None, overwrite_extra_gauss_bounds=None, use_json=True, model_ids=[0, 1, 2])

Initiate the comparison of different models to the Fourier power spectrum of a timeseries.

afino_main_analysis

This routine initiates the AFINO analysis method for a particular chosen model. The fitting is done via SciPy, and the results, including best fit and associated BIC value, are stored in a dictionary and returned.

afino_model_fitting

afino.afino_model_fitting.lnlike(variables, x, y, model_function)

Log likelihood of the data given a model. Assumes that the data is exponentially distributed. Can be used to fit Fourier power spectra. :param variables: array like, variables used by model_function :param x: the independent variable (most often normalized frequency) :param y: the dependent variable (observed power spectrum) :param model_function: the model that we are using to fit the power spectrum :return: the log likelihood of the data given the model.

afino.afino_model_fitting.prob_this_rchi2_or_larger(rchi2, m, nu)
Parameters:
  • rchi2 – reduced chi-squared value
  • m – number of spectra considered
  • nu – degrees of freedom
Returns:
afino.afino_model_fitting.rchi2(m, nu, rhoj)

Goodness-of-fit estimator (Eq. 16)

Parameters:
  • m – number of spectra considered
  • nu – degrees of freedom
  • rhoj – sample to model ratio estimator
Returns:

A chi-square like goodness of fit estimator
Return type:

float
afino.afino_model_fitting.rchi2distrib(m, nu)

The distribution of rchi2 may be approximated by the analytical expression below. Comparing Eq. (2) with the implementation in scipy stats we find the following equivalencies: k (Nita parameter) = a (scipy stats parameter) theta (Nita parameter) = 1 / lambda (scipy stats value) :param m: number of spectra considered :param nu: degrees of freedom :return: a frozen scipy stats function that represents the distribution of the data

afino.afino_model_fitting.rhoj(Sj, shatj)

Sample to Model Ratio (SMR) estimator (Eq. 5)

Parameters:
  • Sj – random variables (i.e. data)
  • shatj – best estimate of the model. Should be same length as Sj
Returns:

The Sample-to-Model ratio
Return type:

ndarray

afino_spectral_models

Power Spectrum Models

afino.afino_spectral_models.bpow(a, f)

Broken power law. This model assumes that there is a break in the power spectrum at some given frequency.

Parameters:
  • a (ndarray[3]) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the power law index at frequencies lower than the break frequency
    • a[2] : break frequency
    • a[3] : the power law index at frequencies higher than the break frequency
  • f (ndarray) – frequencies
afino.afino_spectral_models.bpow_const(a, f)

Broken power law with constant. This model assumes that there is a break in the power spectrum at some given frequency. At high frequencies the power spectrum is dominated by the constant background.

Parameters:
  • a (ndarray(5)) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the power law index at frequencies lower than the break frequency
    • a[2] : break frequency
    • a[3] : the power law index at frequencies higher than the break frequency
    • a[4] : the natural logarithm of the constant background
  • f (ndarray) – frequencies
afino.afino_spectral_models.broken_power_law_with_constant_with_lognormal(a, f)

Broken power law with constant with lognormal. This model assumes that there is a break in the power spectrum at some given frequency. At high frequencies the power spectrum is dominated by the constant background. At some particular frequency there is a lognormal (narrowband distribution)

Parameters:
  • a (ndarray(5)) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the power law index at frequencies lower than the break frequency
    • a[2] : break frequency
    • a[3] : the power law index at frequencies higher than the break frequency
    • a[4] : the natural logarithm of the constant background
  • f (ndarray) – frequencies
afino.afino_spectral_models.constant(a)

The power spectrum is a constant across all frequencies

Parameters:a (float) – the natural logarithm of the power
afino.afino_spectral_models.fnorm(f, normalization)

Normalize the frequency spectrum.

afino.afino_spectral_models.lognormal(a, f)

A lognormal distribution

Parameters:
  • a (ndarray(3)) –
    • a[0] : the natural logarithm of the Gaussian amplitude
    • a[1] : the natural logarithm of the center of the Gaussian
    • a[2] : the width of the Gaussian in units of natural logarithm of the frequency
  • f (ndarray) – frequencies
afino.afino_spectral_models.pow(a, f)

Simple power law. This model assumes that the power spectrum is made up of a power law at all frequencies.

Parameters:
  • a (ndarray[2]) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the power law index
  • f (ndarray) – frequencies
afino.afino_spectral_models.pow_const(a, f)

Power law with a constant. This model assumes that the power spectrum is made up of a power law and a constant background. At high frequencies the power spectrum is dominated by the constant background.

Parameters:
  • a (ndarray[2]) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the power law index
    • a[2] : the natural logarithm of the constant background
  • f (ndarray) – frequencies
afino.afino_spectral_models.pow_const_2gauss(a, f)

Simple power law with a constant, plus a Gaussian bump, and a second Gaussian bump. The second bump is implemented to account for a persistent signal in the data, such as a spin period.

Parameters:
  • a (ndarray[8]) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the power law index
    • a[2] : the natural logarithm of the constant background
    • a[3] : The amplitude of the first bump
    • a[4] : the frequency location of the first bump
    • a[5] : the width of the first bump
    • a[6] : The amplitude of the second bump
    • a[7] : the frequency location of the second bump
    • a[8] : the width of the second bump
  • f (ndarray) – frequencies
afino.afino_spectral_models.pow_const_gauss(a, f)

Simple power law with a constant, plus a Gaussian shaped bump. This model assumes that the powe spectrum is made up of a power law and a constant background. At high frequencies the power spectrum is dominated by the constant background.

Parameters:
  • a (ndarray[2]) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the power law index
    • a[2] : the natural logarithm of the constant background
  • f (ndarray) – frequencies
afino.afino_spectral_models.power_law_with_constant_with_lognormal(a, f)

Power law with constant and a lognormal.

Parameters:
  • a (ndarray[6]) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the power law index
    • a[2] : the natural logarithm of the constant background
    • a[3] : the natural logarithm of the Gaussian amplitude
    • a[4] : the natural logarithm of the center of the Gaussian
    • a[5] : the width of the Gaussian in units of natural logarithm of the frequency
  • f (ndarray) – frequencies
afino.afino_spectral_models.sum_of_pulses(a, f)

Sum of pulses. This model is based on Aschwanden “Self Organized Criticality in Astrophysics”, Eq. 4.8.23. Simulations implementing this equation come up with a shape that is modeled below.

Parameters:
  • a (ndarray[3]) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the scale frequency
    • a[2] : the power law index
  • f (ndarray) – frequencies
afino.afino_spectral_models.sum_of_pulses_with_constant(a, f)

Sum of pulses plus constant. This model is based on Aschwanden “Self Organized Criticality in Astrophysics”, Eq. 4.8.23, with a constant background to model detector noise.

Parameters:
  • a (ndarray[3]) –
    • a[0] : the natural logarithm of the normalization constant
    • a[1] : the scale frequency
    • a[2] : the power law index
    • a[3] : natural logarithm of the background constant
  • f (ndarray) – frequencies

afino_utils

This module provides a number of convenience functions for use throughout the AFINO codebase.

class afino.afino_utils.NumpyEncoder(*, skipkeys=False, ensure_ascii=True, check_circular=True, allow_nan=True, sort_keys=False, indent=None, separators=None, default=None)

This encoder converts numpy arrays to lists so that they can be saved in JSON format

default(obj)

Implement this method in a subclass such that it returns a serializable object for o, or calls the base implementation (to raise a TypeError).

For example, to support arbitrary iterators, you could implement default like this:

def default(self, o):
    try:
        iterable = iter(o)
    except TypeError:
        pass
    else:
        return list(iterable)
    # Let the base class default method raise the TypeError
    return JSONEncoder.default(self, o)
afino.afino_utils.model_string_from_id(id)

This function converts an AFINO model ID integer to a string descriptor

afino.afino_utils.relative_bics(saveresult)

This convenience function calculates all the relative BIC comparisons from an AFINO save result.

Parameters:saveresult (dict) – An AFINO save result dictionary. Can be restored from a previous save file
Returns:dbic_values – A dictionary of relative BIC value comparisons, i.e. BICa - BICb for every a,b pair
Return type:dict
afino.afino_utils.restore_json_save_file(fname)

This function restores an AFINO JSON save file and converts certain dictionary entries back into their original ndarray form. The output should be identical to that from the pickle save files

afino.afino_utils.save_afino_results(results, use_json=False, description=None)

This function saves the results of an AFINO analysis run to either a JSON file or a Pickle file.

Parameters:
  • results (list) – An AFINO results list, obtained via the model_comparison function
  • use_json (bool, optional) – If True, use JSON format to save the analysis results. If False, use pickle format
  • description (string, optional) – An optional descriptor for the result that is incorporated into the filename

afino_series

afino.afino_series

alias of afino.afino_series

A simple time series object

afino.afino_series.prep_series(ts)

put the data series into the form (I - <I>) / <I>. multiply the series by a Hanning window to aid the FFT process