v2.0.0-beta.32

What's Changed

• pandas2.0 take III: Revenge of the Sith by @AdamOrmondroyd in #292

Full Changelog: v2.0.0-beta.31...v2.0.0-beta.32

v2.0.0-beta.31

What's Changed

• Order contour levels from outermost to innermost by @lukashergt in #173
• Fix for plotting with integer labels by @williamjameshandley in #163
• fix axlines and axspans for axes with upper=False by @lukashergt in #175
• patch for getdistreader by @lukashergt in #172
• Copy tex and optionally set label when merging samples by @Stefan-Heimersheim in #177
• adapt to breaking chanes in matplotlib API version 3.5.0 by @lukashergt in #184
• Fixing unexpected behaviour associated with reset_index by @williamjameshandley in #185
• Fix for unsorted logL construction bug by @williamjameshandley in #187
• Fix #189, merge_samples_weighted breaking triangular_sample_compression_2d by @Stefan-Heimersheim in #191
• Allow NestedSamples to take lists as well as numpy arrays as constructors by @williamjameshandley in #190
• Fix to (near) empty phys_live-birth.txt file reading by @williamjameshandley in #186
• Updated for most recent PEP compliant assert statements by @williamjameshandley in #195
• Check for NaNs in logL by @AdamOrmondroyd in #192
• Manual setting of the bandwidth by @lukashergt in #193
• Quick fix c/=c[-1] to c=c/c[-1] by @lukashergt in #206
• WeightedDataFrame plotting by @williamjameshandley in #194
• MCMC plotting by @williamjameshandley in #207
• create parent class Samples for child classes MCMCSamples and NestedSamples by @lukashergt in #209
• remove limits attribute and xmin and xmax kwargs by @lukashergt in #212
• Issue with plot labels by @williamjameshandley in #218
• WeightedPandas index improvements by @williamjameshandley in #198
• Change to reading system by @williamjameshandley in #217
• 100% coverage by @williamjameshandley in #221
• logZ(beta) functionality and logX function by @williamjameshandley in #210
• Have separate ncompress and nplot by @lukashergt in #216
• tex labels in MultiIndex columns by @williamjameshandley in #215
• update gui script by @yallup in #228
• Fix for filenames which contain regular expressions by @williamjameshandley in #231
• Fixes for dropping weights and labels by @williamjameshandley in #230
• fixes for pandas 1.5.0 and matplotlib 3.6.0 updates by @lukashergt in #238
• Fix warnings shown with pytest by @AdamOrmondroyd in #234
• Change figure and axes output handling by @lukashergt in #224
• Updated .readthedocs.yml to be python 3.8 by @williamjameshandley in #247
• reduce CI workflow matrices by @lukashergt in #249
• pass labels argument through when reading chains by @AdamOrmondroyd in #232
• Fix to non-deterministic behaviour by removing long-range side effects by @williamjameshandley in #248
• API hints for existing users by @williamjameshandley in #246
• Perfect nested sampling functionality by @williamjameshandley in #205
• Close figures on teardown by @AdamOrmondroyd in #254
• .loc to avoid warnings by @AdamOrmondroyd in #255
• replace .boxplot()with .plot.box() by @AdamOrmondroyd in #256
• MCMCSamples methods for burn-in removal and Gelman--Rubin statistic by @lukashergt in #258
• Packaging improvements by @williamjameshandley in #259
• matplotlib 3.7.0 moved SubplotBase by @AdamOrmondroyd in #263
• Improved packaging mark 2 by @williamjameshandley in #262
• Improved packaging mark 3 by @williamjameshandley in #265
• Improved packaging mark 4 by @williamjameshandley in #266
• Update documentation for 2.0 release by @lukashergt in #243
• Improved packaging mark 5 by @williamjameshandley in #267
• Improved packaging mark 6 by @williamjameshandley in #269
• Updated matplotlib version requirements by @tobyLovick in #268
• add flake8 pre-commit by @AdamOrmondroyd in #261
• Replace overridden _make_plot() with _plot() by @AdamOrmondroyd in #235
• Correct script in pyproject.toml + Improved packaging mark 7 by @williamjameshandley in #271
• WeightedGroupBy by @AdamOrmondroyd in #272
• handley-lab migration by @williamjameshandley in #279
• rename [extras] optional dependency to [all] by @AdamOrmondroyd in #281
• Use the scipy version of linalg.cholesky which consistently raises LinAngError with linearly dependent columns by @AdamOrmondroyd in #287
• Make name of Colormaps returned by basic_cmap a string by @ThomasGesseyJones in #289
• Release workflow by @ThomasGesseyJones in #291

New Contributors

• @yallup made their first contribution in #228
• @tobyLovick made their first contribution in #268
• @ThomasGesseyJones made their first contribution in #289

Full Changelog: 2.0.0-beta.9...v2.0.0-beta.31

2.0.0-beta.11: Fixing unexpected behaviour associated with reset_index (#185)

* Fixed unexpected behavior associated with reset_index

* version bump

2.0.0-beta.9: Inner ticks (#153)

* add inner ticks, i.e. also in subplots

* add test to check that inner ticks are updated correctly when modifying in single subplot

* add corner case test for 00 subplot in triangle to appease codecov

* fix too long line to appease lint

* rename  to just  and change options to ['outer', 'inner', None] with None referring to no ticks at all

* add test for ValueError when arg  is not one of 'inner', 'outer', or None

* fix renaming from NotImplementedError to ValueError, the latter is more appropriate here

Co-authored-by: Will Handley <[email protected]>

2.0.0-beta.7: 1d Confidence Regions (#145)

* 1D confidence regions function added to utils and optional plot added to fastkde_plot_1d

* generalisation of iso_probability_1d function to handle different contour values and changes to related plot functions

* testing of iso_probability_1d via plot functions

* double checking and editing docs

* flake8 compliant and increased coverage

* added explicit tests to for utils 1d propability contour function

* flake8 on test_utils.py

* changed color kwargs to use pre-existing, fixed docs and moved fill_between plotting into "helper" function

* flake8 fix

* docstring fix in fill_plot_1d

* docstring fix in fill_plot_1d

* adapting 1d contour method to properly plot intervals using existing utils functions

* removed redundant iso_probability_1d function

* updated test_utils.py to meet changes with the code

* minor change to fill colors

* changed color scheme to match up with 2D posterior plotting

* updating tests to fit with change in code

* added back in cmap kwarg, seems to crash tests without it

* flake8 fix

* added filledkde and filledfaskde plot types

* fixed tests

* added new samples tests for the additional code

* removing filledfastkde test in samples as fastkde doesnt appear to be in requirements

* trying to get codecov/path to pass

* fix for fastkde 1d filled plots

* adding in weights for iso_probability_contours_from_samples call

* fixed issue when changing colormaps

* fixing flake issue

* triggering fill with facecolor for 1d plots and removing filled1d kwarg

* reorganised kwargs for filled 1d plots/color scheme

* edits to color scheme

* few edits to plot.py and additional tests added

* bumping beta version number and fixing docstring error

2.0.0-beta.5: fix `MCMCSamples.importance_sample` (#139)

* improve test for MCMC importance_sample, ensure that weights are changed corresponding to change in logL

* fix missing adjustment of weights in MCMCSamples.importance_sample, the current version left the posteriors invariant

Importance sampling

#122 #126

• Added an importance sample function which creates valid importance sampled nested sampling run

nlive bug fixes for plateau regions

#119

• Fix to nlive computation for a bug that was occurring for plateau regions at the start of the run
• Added a wedding cake example likelihood to demonstrate anesthetic's effectiveness in dealing with plateaus

WeightedDataFrame upgrade

#115

• Moved weights in WeightedDataFrame into MultiIndex of nested sampling iteration and weight. This simplifies the implementation, makes weights visible as part of the index and is compatible with past and current version of pandas
• rename weight, w -> weights throughout for compatibility with matplotlib

Last stable version of anesthetic 1

Following the update of pandas to 1.1 and consequent breaking changes outlined in #115, we are now upgrading the API to deal with this and other issues to be resolved in the release of anesthetic 2.0.0. We therefore freeze anesthetic 1, and its dependency on pandas <=1.0.5 in this release