Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased

  • differentiable NCFA

  • interventions

[2.0.0] - 202X-XX-XX

Removed

  • dcor dependency

[1.1.0] - 2024-08-19

Added

  • source code for xicorrelation, updated for NumPy 2.0.0+

Removed

  • hypothesis_test() in favor of estimate_UDG()

[1.0.0] - 2024-05-29

Added

  • Chaterjee’s xi correlation for independence testing

  • sampling from (given or randomly generated) linear Gaussian minMCM

  • random generation of UDGs satisfying 1-pure-child condition

  • BIC-optimal UDG estimation

  • examples, documentation, and tests

Changed

  • functional_mcm to use VAE framework instead of GAN-inspired

  • making api more like sklearn

  • improved script to reproduce PGM demo

  • back to AGPLv3+ license

  • simplified extra dependencies to just “dgm” and “vis”

  • refactoring and reorganizing to improve PEP 8 compliance

Removed

  • gues, grues, and abstraction submodules

[0.7.0] - 2022-10-18

Changed

  • updated licence to CNPLv7+

  • updated description in setup.py and readme to reflect expanded focus of the package

Added

  • grues submodule implementing an MCMC-based method for UEC learning

[0.6.0] - 2021-06-07

Changed

  • licences from AGPLv3+ to CNPLv6+

Fixed

  • name in setup.py

  • numpy depricated boolean subtraction graph.py l269

[0.5.0] - 2020-09-22

Added

  • can now learn functional MCM instead of just structure (using GAN architecture)

  • data simulation (using same GAN architecture)

  • module for visualizing graphs and making various other plots

  • vastly improved documentation site

  • examples submodule, with some sample structures and data sets

  • PGM conference demo script

  • more metadata to setup.py

Changed

  • independence_testing.dependencies() subsumed by independence_testing.hypothesis_test()

  • license from GPLv3 to AGPLv3+

  • using Black code formatter

Fixed

  • various small typo and bug fixes

[0.4.0] - 2020-07-23

Added

  • package now available on pip

Fixed

  • bug in max_intersection_num calculation

  • exception handling for edgeless graph

  • bug with graph.cover_edges() use in ecc.find_cm.branch()

  • bug with using graph copy outside of ecc.find_cm.branch() rather than within

  • bug where graph.choose_nbrhood() picked wrong edge (and gave edgeless nbrhood)

Removed

  • hack to stop while loop in find_clique_min_cover()

Changed

  • default verbosity of output

  • reducible_graph var name to branch_graph

  • package requirements

  • updated README

[0.3.0] - 2020-04-13

Added

  • installation via pip and git

  • rudimentary output to update on progress

  • doc generating script and initial (incomplete) documentation website

  • tests/ directory and various files for more detailed integration and unit testing

Fixed

  • bug in reduction rule 2 of find_cm ecc alg that shows up on fully connected graphs

  • bug in reduction rule 3 of the find_cm ecc alg

  • incorrect dates (2018 instead of 2019) in changelog

Removed

  • utils.py—distributed functions to relevant modules

Changed

  • indpendence_test.py –> independence_testing.py

  • split up ecc_algs.reducee() internals for better testing/readability

[0.2.0] - 2019-11-04

Added

  • references and better example to README

  • graph.py and ecc_algorithms.py modules—makes package more maintainable and extensible

  • test script

Changed

  • organization more in line with python package guidelines

  • changed from using distance covariance to distance correlation

  • many functions renamed according to new modules and organization

Removed

  • need for importing networkx to find minMCM

  • junk code in misc dir

[0.1.0] - 2019-10-07

Added

  • code in its raw, disorganized, poorly documented, inefficient state