Random Graphical Model Estimation under L0 Penalty // Provides functions for estimating sparse precision matrices using a random graphical model framework under an L0-style penalty. The method evaluates candidate theta values and returns both continuous and binary precision matrices representing inferred network structures.