joaomarcelino1/Structure-learning-of-homogeneous-decomposable-models: Structure Learning of homogeneous decomposable models v1.0.0

This repository contains the implementation of the functions and algorithms described in Sections 4 and 5 of the paper Structure Learning of Homogeneous Decomposable Models. In particular, the code includes: Integer Linear Programming formulations for learning the structure of homogeneous decomposable models; LP relaxations of the homogeneous chordal graph polytope based on local constraints; Extensions incorporating a maximum clique size constraint; Procedures for synthetic data generation; Numerical experiments and performance evaluation using the Structural Hamming Distance (SHD). The main goal of this repository is to study and evaluate exact and relaxed optimization-based approaches for structure learning in undirected graphical models with homogeneity constraints.