Faithlessness in Gaussian graphical models
- 1 February 2026
- journal article
- Published by Bernoulli Society for Mathematical Statistics and Probability in Bernoulli
- Vol. 32 (1) , 638-663
Abstract
The implication problem for conditional independence (CI) asks whether the fact that a probability distribution obeys a given finite set of CI relations implies that a further CI statement also holds in this distribution. This problem has a long and fascinating history, cumulating in positive results about implications now known as the semigraphoid axioms as well as impossibility results about a general finite characterization of CI implications. Motivated by violation of faithfulness assumptions in causal discovery, we study the implication problem in the special setting where the CI relations are obtained from a directed acyclic graphical (DAG) model along with one additional CI statement. Focusing on the Gaussian case, we show that this implication problem reduces to a principal ideal membership problem. Using this we derive necessary and sufficient combinatorial criteria for when the set of faithless distributions decomposes into graphical components. Moreover, prompted by the relevance of strong faithfulness in statistical guarantees for causal discovery algorithms, we give a graphical solution for an approximate CI implication problem, in which we ask whether small values of one additional partial correlation entail small values for yet a further partial correlation.Keywords
All Related Versions
This publication has 22 references indexed in Scilit:
- Separators and adjustment sets in causal graphs: Complete criteria and an algorithmic frameworkArtificial Intelligence, 2019
- The Geometry of GaussoidsFoundations of Computational Mathematics, 2018
- Ideals, Varieties, and Algorithms2015
- Positivity for Gaussian graphical modelsAdvances in Applied Mathematics, 2013
- Gaussian conditional independence relations have no finite complete characterizationJournal of Pure and Applied Algebra, 2009
- Lectures on Algebraic Statistics2009
- Algorithms in Real Algebraic Geometry2006
- Semigraphoids and structures of probabilistic conditional independenceAnnals of Mathematics and Artificial Intelligence, 1997
- Identifying independence in bayesian networksNetworks, 1990
- On the Logic of Causal Models* *This work was partially supported by the National Science Foundation Grants #IRI-8610155, “Graphoids: A Computer Representation for Dependencies and Relevance in Automated Reasoning,” and #IRI-8821444, “Probabilistic Networks for Automated Reasoning.”1990