Changelog

v3.2.0

Added

  • New CompositeModel class that combines modular linear models and behaves like a simple Model.

Changed

  • Fully reworked NoiseWaves class to use the CompositeModel class.

v3.1.1

Changed

  • New model_filter function that xrfi_model calls and can be called by other more general functions.

Added

  • New tests of ModelFilterInfo and ModelFilterInfoContainer

v3.1.0

Added

  • New semi-rigid S-parameters file from 2017.

v3.0.0

Fixed

  • Tests were slow because they were using the wrong default for getting the standard deviation in xrfi_model. Now uses the fast kind.

Changed

  • modelling module got an overhaul. It now consists of smaller, self-consistent classes that are based on attrs and are read-only (but clonable). There is an explicit split between a Model and a FixedLinearModel which is defined at some coordinates x, allowing the latter to provide speedups for fitting lots of data at the same coordinates. They are also YAML-read/writable.

v2.1.0

Added

  • FourierDay linear model

Fixed

  • models were being mutated in ModelFit. No more.

v2.0.0

Added

  • Ability to pass through LNA S11 options

  • Ability to specify t_load and t_load_ns in spectra

  • Easier conversion to Calibration object from CalibrationObservation

  • New NoiseWaves linear model

  • Modular InternalSwitch class

Fixed

  • Ensure fitted models are always copied

v1.0.0

Added

  • Visualization of xrfi_model output.

  • Allow init_flags as input to xrfi_model

Changed

  • Performance boost for xrfi_model_sweep

  • Faster modelling in detail

  • Compatibility with edges-io v1.

  • More consistency in API between different XRFI models.

  • Allow not passing spectrum to xrfi_explicit ### Fixed

  • Improvements to typing.

  • Improvements to xrfi tests.

v0.7.0

Added

  • Automatic notebook reports for calibration

  • xrfi_model_sweep that actually works and is tested.

  • Faster model evaluations

Fixed

  • xRFI doesn’t assume that input spectrum is all positive (could be residuals, and therefore have negatives).

Version 0.4.0

Changed

  • Much faster modeling via re-use of basis function evaluations

Version 0.3.0

Changed

  • load_name now always an alias (hot_load, ambient, short, open)

  • Load.temp_ave now always the correct one (even for hot load)

Version 0.2.1

Added

  • Basic tests

  • Travis/tox/codecov setup

Version 0.2.0

Added

  • Many many many new features, and complete modularisation of code

  • Now based on edges-io package to do the hard work.

  • Refined most modules to remove redundant code

  • Added better package structure

Version 0.1.0

  • First working version on github.