Advanced power spectrum averaging functions

The hera_stats.average module contains several advanced averaging functions for power spectra. Currently, this only consists of cumulative averaging in time or baseline-pair (average_spectra_cumul) and differencing redundant groups with respect to their average (redundant_diff).

hera_stats.average.average_spectra_cumul(uvp, blps, spw, polpair, mode='time', min_samples=1, shuffle=False, time_avg=True, verbose=False)[source]

Cumulatively average a set of delay spectra as a function of time or blpair. The cumulative average can be performed sequentially, e.g. in time order, or in a random (shuffled) order.

Parameters:
  • uvp (UVPSpec) – Set of power spectra to be cumulatively averaged.
  • blps (list of ints or tuples) – List of blpair ints or tuples. Only one redundant set of blpairs should be passed at once.
  • spw, polpair (int, str) – Spectral window ID (integer) and polarization-pair (integer or str) of the power spectra to average.
  • mode (str, optional) – Whether to cumulatively average in time or blpair. The other dimension will be averaged over non-cumulatively. (See the ‘time_avg’ kwarg below for other behaviors.) Possible options are ‘time’ or ‘blpair’. Default: ‘time’.
  • min_samples (int, optional) – Minimum number of samples to allow in the cumulative average. Default: 1.
  • shuffle (bool, optional) – Whether to randomly shuffle the order of the cumulative averaging, or to keep it in order. Default: False.
  • time_avg (bool, optional) – Whether to average over times. This option is only used if mode=’blpair’; otherwise it will be ignored. Default: True.
  • verbose (bool, optional) – Whether to print status messages as the cumulative averaging progresses. Default: False.
Returns:

  • ps (array_like) – Cumulative averages of delay spectra, in a 2D array of shape (Nsamp, Ndelay), where Nsamp = Ntimes or Nblpairs.
  • dly (array_like) – Power spectrum delay modes (in s).
  • n_samples (array_like) – Number of times or blpairs that went into each average in the ps array.
hera_stats.average.redundant_diff(uvd, bls, pol, return_mean=False)[source]

Calculate the difference between all baselines in a redundant group and the mean of the redundant group (as a fn. of frequency and time).

Empty (fully flagged) baselines are excluded from the average.

N.B. The averaging does not currently take into account data weights or number of samples.

Parameters:
  • uvd (UVData) – UVData object containing the data that will be averaged and differenced.
  • bls (list of tuple or int) – List of baseline tuples or integers to be treated as a group. The mean will be calculated over all
  • pol (str) – Which polarization to extract from the UVData file.
  • return_mean (bool, optional) – If True, return the mean over the redundant group too. Default: False
Returns:

  • bls (list of baseline tuple or int) – List of baselines that were kept in the average.
  • diffs (list of array_like) – List of arrays of differences between each baseline and the group mean.
  • mean (array_like, optional) – Mean over data from all non-flagged baselines, as a function of freq. and time. Only retuend if return_mean is True.
hera_stats.average.redundant_diff_summary(uvd, red_bls, pol, op_upper=<Mock name='mock.max' id='140500795349200'>, op_lower=None, verbose=False)[source]

Calculate summary stats on the differences of baselines with the mean of their redundant group. Returns an antenna-antenna matrix where each element is the value of the summary statistic for a given antenna pair.

See hera_stats.plot.antenna_matrix for an accompanying plotting function.

Parameters:

  • uvd (UVData) – Visibility data object.

  • red_bls (list of lists) – List of redundant baseline group lists.

  • pol (str or int) – String or integer specifying the polarization to use.

  • op_upper (func) – Function that operates on the difference waterfall (visibility for baseline minus the mean over its redundant group) to return a single number, to be used as the summary statistic for that baseline.

    Functions must have the following signature:

    scalar = func(d, grp_mean)

    where d is the difference data (a 2D complex array) and grp_mean is the mean over the redundant group (also a 2D complex array).

    The values of the summary statistic are placed on the upper triangle of the output matrix.

  • op_lower (func, optional) – Same as op_upper, but places values on the lower triangle of the output matrix.

  • verbose (bool, optional) – Whether to print progress messages. Default: False.
Returns:

  • unique_ants (array_like) – Ordered array of unique antennas that were found in the redundant groups.

  • mat (array_like) – 2D array of summary statistic values for every pair of antennas.

    NB. np.nan values are returned for antenna pairs that were not present in the data/redundant baseline list.