This package contains algorithms for solving General Graphical Lasso (GGLasso) problems, including single, multiple, as well as latent Graphical Lasso problems.

Docs | Examples

The package is available on pip and conda and can be installed with

pip install gglasso

or

conda install -c conda-forge gglasso

Alternatively, you can install the package from source using the following commands:

git clone https://github.com/fabian-sp/GGLasso.git
pip install -r requirements.txt
python setup.py

Test your installation with

pytest gglasso/ -v

If you want to create a conda environment with full development dependencies (for building docs, testing etc), run:

conda env create -f environment.yml

If you wish to install gglasso in developer mode, i.e. not having to reinstall gglasso everytime the source code changes (either by remote or local changes), run

python setup.py clean --all develop clean --all

GGLasso can solve multiple problem forumulations, e.g. single and multiple Graphical Lasso problems as well as with and without latent factors. Therefore, the main entry point for the user is the glasso_problem class which chooses automatically the correct solver and model selection functionality. See our documentation for all the details.

GGLasso contains algorithms for solving a multitude of Graphical Lasso problem formulations. For all the details, we refer to the solver overview in our documentation.

The package includes solvers for the following problems:

• Single Graphical Lasso
  

• Group and Fused Graphical Lasso
  We implemented the ADMM (see [2] and [3]) and a proximal point algorithm (see [4]).

• Non-conforming Group Graphical Lasso
  A Group Graphical Lasso problem where not all variables exist in all instances/datasets.

• Functional Graphical Lasso
  A variant of Graphical Lasso where each variables has a functional representation (e.g. by Fourier coefficients).

Moreover, for all problem formulation the package allows to model latent variables (Latent variable Graphical Lasso) in order to estimate a precision matrix of type sparse - low rank.

If you use GGLasso, please consider the following citation

@article{Schaipp2021,
  doi = {10.21105/joss.03865},
  url = {https://doi.org/10.21105/joss.03865},
  year = {2021},
  publisher = {The Open Journal},
  volume = {6},
  number = {68},
  pages = {3865},
  author = {Fabian Schaipp and Oleg Vlasovets and Christian L. Müller},
  title = {GGLasso - a Python package for General Graphical Lasso computation},
  journal = {Journal of Open Source Software}
}

1. Contributions and suggestions to the software are always welcome. Please, consult our contribution guidelines prior to submitting a pull request.
2. Report issues or problems with the software using github’s issue tracker.
3. Contributors must adhere to the Code of Conduct.

• [1] Friedman, J., Hastie, T., and Tibshirani, R. (2007). Sparse inverse covariance estimation with the Graphical Lasso. Biostatistics, 9(3):432–441.
• [2] Danaher, P., Wang, P., and Witten, D. M. (2013). The joint graphical lasso for inverse covariance estimation across multiple classes. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 76(2):373–397.
• [3] Tomasi, F., Tozzo, V., Salzo, S., and Verri, A. (2018). Latent Variable Time-varying Network Inference. InProceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. ACM.
• [4] Zhang, Y., Zhang, N., Sun, D., and Toh, K.-C. (2020). A proximal point dual Newton algorithm for solving group graphical Lasso problems. SIAM J. Optim., 30(3):2197–2220.