Half-trek criterion for generic identifiability of linear structural equation models

DOI10.1214/12-AOS1012zbMATH Open1257.62059arXiv1107.5552OpenAlexW3104704650WikidataQ57566399 ScholiaQ57566399MaRDI QIDQ116505FDOQ116505

Authors: Rina Foygel, Jan Draisma, Mathias Drton, Jan Draisma, Mathias Drton, Rina Foygel Barber 

Publication date: 1 June 2012

Published in: The Annals of Statistics (Search for Journal in Brave)

Abstract: A linear structural equation model relates random variables of interest and corresponding Gaussian noise terms via a linear equation system. Each such model can be represented by a mixed graph in which directed edges encode the linear equations and bidirected edges indicate possible correlations among noise terms. We study parameter identifiability in these models, that is, we ask for conditions that ensure that the edge coefficients and correlations appearing in a linear structural equation model can be uniquely recovered from the covariance matrix of the associated distribution. We treat the case of generic identifiability, where unique recovery is possible for almost every choice of parameters. We give a new graphical condition that is sufficient for generic identifiability and can be verified in time that is polynomial in the size of the graph. It improves criteria from prior work and does not require the directed part of the graph to be acyclic. We also develop a related necessary condition and examine the "gap" between sufficient and necessary conditions through simulations on graphs with 25 or 50 nodes, as well as exhaustive algebraic computations for graphs with up to five nodes.

Full work available at URL: https://arxiv.org/abs/1107.5552

Recommendations

• Generic identifiability of linear structural equation models by ancestor decomposition
• Global identifiability of linear structural equation models
• Identifiability of Gaussian structural equation models with equal error variances
• Identifiability of Gaussian linear structural equation models with homogeneous and heterogeneous error variances
• Algebraic problems in structural equation modeling

zbMATH Keywords

multivariate normal distributiongraphical modelscovariance matrixGaussian distributionparameter identification

Mathematics Subject Classification ID

Linear regression; mixed models (62J05) Applications of graph theory (05C90) Characterization and structure theory for multivariate probability distributions; copulas (62H05) Measures of association (correlation, canonical correlation, etc.) (62H20)

Cites Work

• SINGULAR: A computer algebra system for singularity theory, algebraic geometry and commutative algebra
• Causation, prediction, and search
• Title not available (Why is that?)
• Introduction to algorithms
• Title not available (Why is that?)
• Global identifiability of linear structural equation models
• Half-trek criterion for generic identifiability of linear structural equation models
• Ancestral graph Markov models.
• Combinatorial Optimization. Polyhedra and efficiency. CD-ROM
• Distinctness of the eigenvalues of a quadratic form in a multivariate sample
• Title not available (Why is that?)
• The Method of Path Coefficients
• Probability distributions with summary graph structure
• Assumptions of IV methods for observational epidemiology
• Trek separation for Gaussian graphical models

Cited In (22)

• Positivity for Gaussian graphical models
• TRYGVE HAAVELMO AND THE EMERGENCE OF CAUSAL CALCULUS
• Computation of maximum likelihood estimates in cyclic structural equation models
• Half-trek criterion for identifiability of latent variable models
• Graphs for margins of Bayesian networks
• Generic identifiability of linear structural equation models by ancestor decomposition
• A survey of discrete methods in (algebraic) statistics for networks
• A note on the connection between trek rules and separable nonlinear least squares in linear structural equation models
• Foundations of structural causal models with cycles and latent variables
• Learning instrumental variables with structural and non-Gaussianity assumptions
• Sequences of regressions and their independences
• Algebraic properties of HTC-identifiable graphs
• Identifiability of directed Gaussian graphical models with one latent source
• Identifiability of homoscedastic linear structural equation models using algebraic matroids
• Global identifiability of linear structural equation models
• Half-trek criterion for generic identifiability of linear structural equation models
• SEMID
• Identifiability in Continuous Lyapunov Models
• Empirical likelihood for linear structural equation models with dependent errors
• Nested covariance determinants and restricted trek separation in Gaussian graphical models
• Algebraic problems in structural equation modeling
• Causal inference in partially linear structural equation models

Uses Software

• Matlab
• R
• SINGULAR
• TETRAD

This page was built for publication: Half-trek criterion for generic identifiability of linear structural equation models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q116505)