Version 0.4.2

This is a patch release implementing support for newer versions of pyuvdata and its future_array_shapes.

What's changed:

• All data files imported into a PSPecData object are now read with future_array_shapes, that is the spectral window axis has been removed.
• All UVBeam objects imported to form a PSPecBeam are also read with future_array_shapes by default.
• Following the new pyuvdata syntax, the channel_width attribute of UVPSpec objects is now an array.
• Backwards compatibility is allowed by the use_future_array_shapes() method of pyuvdata, which changes the array shapes of an object to match future array shapes (see here). Therefore, no (test) data file has been modified. If a conversion was required, a warning message is printed.
• Tutorials and example notebooks have been modified accordingly.
• Please pay extra care to the missing array dimension when you convert your data arrays from Jy/beam to mK by hand.

Version 0.4.1

This is a patch release including various bug fixes and version dependencies updates, some related to the new array shapes enforced by pyuvdata.

What's changed

• Enhancement to the UVWindow module: The beam is fed to UVWindow.from_uvpspec with the ftbeam key (previously ftfile). It can the path to a beam simulation (Path or str) or a FTBeam object directly. This facilitates using Gaussian beam models.
• np.nan_to_num method applied to weights in grouping.spherical_average to remove persisting NaN terms in the imaginary part.
• Checks on spws within pspec_run to exit early if not correct
• Improved documentation
• Updates to required version (scipy, pyuvdata, h5py, hera_cal)
• Removed usage of deprecated np.float, np.complex, etc.
• Sped up the environment setup in tests and added python 3.10.

Full changelog: v0.4.0...v0.4.1.

Version 0.4.0

This releases version 0.4.0 which was used to derive upper limits using the full Phase I dataset (arXiv:2210.04912).

Major upgrades include the Gorce+2022 (arXiv:2210.03721) window function derivation method, use of coverage comparison and checks with codecov and other minor features.

Version 0.3.0

This releases version 0.3.0, which was used to run the HERA+2021 limits for the IDR2.2 (Phase I) data set.

Major upgrades includes the Tan+2021 errorbar methods, as well as spherical averaging, and other minor features.

Version 0.2.0

v0.2.0 (2019-08-14)

• Python 3 compatibility.
• Allow cross-polarization spectra to be calculated as long as they aren't beam-normalized; polarizations now specified as polpairs.
• Time-dependent noise power spectra capability in generate_noise.
• Methods to fetch redundant baselines/blpairs from UVPSpec objects.
• Exact normalization mode (Saurabh Singh) with optimization (Ronan Legin).
• Updated covariance handling, incl. averaging and analytic variance (Jianrong Tan).
• New 'lazy covariance' weighting mode (Aaron Ewall-Wice).
• Fix bug where little_h was passed incorrectly (Zachary Martinot).
• Store additional statistical info in stats_array (Duncan Rocha).
• Add delay wedge plotting function (Paul Chichura).
• Various minor interface changes and improved tests.

v0.1.0

This is an initial "feature-complete" release of hera_pspec, ready for the IDR2.1 analysis. It should include all of the OQE algorithm features, data structures, and pipeline and plotting features needed for an initial analysis.