Statistical tests for jackknives

The hera_stats.stats module contains various statistical convenience functions to compare jackknifed power spectra and other data.

hera_stats.stats.redgrp_pspec_covariance(uvp, red_grp, dly_idx, spw, polpair, mode='cov', verbose=False)[source]

Calculate the covariance or correlation matrix for all pairs of delay spectra in a redundant group, for a single delay bin. The matrix is estimated by averaging over all LST samples.

Parameters:
  • uvp (UVPSpec) – Input UVPSpec object.
  • red_grp (list) – List of redundant baseline pairs within a group.
  • dly_idx (int) – Index of the delay bin to calculate the covariance matrix for.
  • spw (int) – Index of spectral window to use.
  • polpair (int or str or tuple) – Polarization pair.
  • mode (str, optional) – Whether to calculate the covariance matrix (‘cov’) or correlation matrix (‘corr’). Default: ‘cov’.
  • verbose (bool, optional) – Whether to print status messages. Default: false.
Returns:

cov_real, cov_imag – Real and imaginary covariance or correlation matrices, of shape (Nblps, Nblps).
Return type:

ndarrays
hera_stats.stats.uvp_zscore(uvp, error_field='bs_std', inplace=False)[source]

Calculate a zscore of a UVPSpec object using entry ‘error_field’ in its stats_array. This assumes that the UVPSpec object has been already mean subtracted using hera_pspec.uvpspec_utils.subtract_uvp().

The resultant zscore is stored in the stats_array as error_field + “_zscore”.

Parameters:
  • uvp (UVPSpec object)
  • error_field (str, optional) – Key of stats_array to use as z-score normalization.
  • inplace (bool, optional) – If True, add zscores into input uvp, else make a copy of uvp and return with zscores.
Returns:

uvp : UVPSpec object
Return type:

if inplace