v2.10.0

New capabilities to handle missing data problems

• Sample data through mvpc_gen_data module
• Causal discovery using the mvpc module

v2.9.0

Three major new features:

• The ability to incorporate background knowledge in terms of edge Constraints.
• The ability to specify multiple benchmark setups in the same config file.
• A new module gcastle_dag, for generating DAGs using the gCastle package

v2.8.0

• This version introduces the new graph_estimation module which will replace the graph_plots module. It enables conversions to other graph types and the possibility to specify which features.

v2.7.1

• New algorithm modules from the huge package for estimating Gaussian graphical models: huge_glasso, huge_mb, and huge_tiger.
• Updated trilearn to version 2.0.5
• Updated docs

v2.6.1

• Mainly bug fixes, e.g. in the trilearn_pgibbs module

v2.6.0

• Added the GrUES algorithm (Alex Markham).
• Changed all occurrences of replicate to seed in the code. This will unfortunately trigger a rerun of existing jobs as the scripts are changed.
• Added F1 score based on the pattern graph (v-structure kept while the other edges are undirected) of a DAG.

v2.5.0

• Added the feature of taking the output of one algorithm as input for another
• Added the chordal graph MCMC learners module athomas_jtsamplers
• Fixed bugs

v2.3.0

• Updated causal-cmd to version 1.10.0.
• Added the BOSS algorithm.

v2.2.0

• Bechpress is now working through Docker on Apple MX chips.
• Several packages have been updated and the versions are stated explicitly in the Dockerfiles.

v2.1.0

* Using shell scripts for tetrad_ (causal-cmd) algorithms.

* Updating docs.

* Skipping hidden files (starting with .) when iterating over modules.

* Adding GRaSP from causal-learn.

* Adapting fges to causal-cmd 1.9.0

* Adding Tetrad GRaSP.

* Fixed bug for checking if CPDAG in the benchmarks module.

* Using Tetrad (causal-cmd) algorithms use version 1.9.0.

* Removing some algs in tetrad.json from the comparison.