Trek separation for Gaussian graphical models

Abstract

Gaussian graphical models are semi-algebraic subsets of the cone of positive definite covariance matrices. Submatrices with low rank correspond to generalizations of conditional independence constraints on collections of random variables. We give a precise graph-theoretic characterization of when submatrices of the covariance matrix have small rank for a general class of mixed graphs that includes directed acyclic and undirected graphs as special cases. Our new trek separation criterion generalizes the familiar $d$-separation criterion. Proofs are based on the trek rule, the resulting matrix factorizations and classical theorems of algebraic combinatorics on the expansions of determinants of path polynomials.

Published in at http://dx.doi.org/10.1214/09-AOS760 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

References in corpus (1)

• Moments of minors of Wishart matrices

Cited by in corpus (11)

• Geometry of the faithfulness assumption in causal inference
• Half-trek criterion for generic identifiability of linear structural equation models
• Wald tests of singular hypotheses
• Semidefinite tests for latent causal structures
• Matrix Schubert varieties and Gaussian conditional independence models
• Computing Maximum Likelihood Estimates for Gaussian Graphical Models with Macaulay2
• Learning Linear Gaussian Polytree Models with Interventions
• Directed Gaussian graphical models with toric vanishing ideals
• Confounder selection via iterative graph expansion
• Colored Gaussian directed acyclic graphical models
• A Trek Rule for the Lyapunov Equation