Gaussian Process

The Gaussian Process module is used to fit the SN light curves and mangle the SED template (PISCOLA makes use of tinygp). This module is documented bellow:

piscola.gaussian_process.fit_gp_model(times, wavelengths, fluxes, flux_errors, k1='Matern52', fit_mean=False, fit_type='flux', wave_log=True, time_scale=None, wave_scale=None, add_noise=True)

Fits a Gaussian Process model to a SN multi-colour light curve.

All input arrays MUST have the same length.

Parameters:

  • times (ndarray) – Light-curve epochs.

  • wavelengths (ndarray) – Light-curve effective wavelength.

  • fluxes (ndarray) – Light-curve fluxes.

  • flux_errors (ndarray) – Light-curve flux errors.

  • k1 (str) – Kernel to be used for the time axis. Either Matern52, Matern32 or ExpSquared.

  • fit_type (str) – Transformation used for the light-curve fits: flux, log, arcsinh.

  • wave_log (bool, default True.) – Whether to use logarithmic (base 10) scale for the wavelength axis.

  • time_scale (float, default None) – If given, the time scale is fixed using this value, in units of days.

  • wave_scale (float, default None) – If given, the wavelength scale is fixed using this value, in units of angstroms. Note that if ‘wave_log=True’, the logarithm base 10 of this value is used.

  • add_noise (bool, default True) – Whether to add a white-noise component to the GP model. This “inflates” the errors.

Returns:

gp_model – Gaussian Process multi-colour light-curve model.

Return type:

GaussianProcess

piscola.gaussian_process.fit_single_lightcurve(times, fluxes, flux_errors, k1='Matern52')

Fits a Gaussian Process model to a single SN light curve.

All input arrays MUST have the same length.

Parameters:

  • times (ndarray) – Light-curve epochs.

  • fluxes (ndarray) – Light-curve fluxes.

  • flux_errors (ndarray) – Light-curve flux errors.

  • k1 (str) – Kernel to be used for the time axis. Either Matern52, Matern32 or ExpSquared.

Returns:

gp_model – Gaussian Process light-curve model.

Return type:

GaussianProcess

piscola.gaussian_process.prepare_gp_inputs(times, wavelengths, fluxes, flux_errors, fit_type, wave_log)

Prepares the inputs for the Gaussian Process model fitting.

Parameters:

  • times (ndarray) – Light-curve epochs.

  • wavelengths (ndarray) – Light-curve effective wavelength.

  • fluxes (ndarray) – Light-curve fluxes.

  • flux_errors (ndarray) – Light-curve flux errors.

  • fit_type (str) – Transformation used for the light-curve fits: flux, log, arcsinh.

  • wave_log (bool) – Whether to use logarithmic (base 10) scale for the wavelength axis.

Returns:

  • X (ndarray) – X-axis array for the Gaussian Process model.

  • y (ndarray) – Y-axis array for the Gaussian Process model.

  • yerr (ndarray) – Y-axis errors for the Gaussian Process model.

  • y_norm (float) – Normalisation used for the fluxes and errors. The maximum of the fluxes is used.