Maximum likelihood geometry in the presence of data zeros

DOI10.1145/2608628.2608659MaRDI QIDQ3452399zbMATH OpenOpenAlexFDO

Authors Elizabeth Gross, Jose Israel Rodriguez

Publication date 11 November 2015

Published in Proceedings of the 39th International Symposium on Symbolic and Algebraic Computation (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1310.4197

zbMATH Keywords

algebraic statisticscritical pointsmaximum likelihood estimationML degreemodel zerossampling zeros

Mathematics Subject Classification ID

Applications of commutative algebra (e.g., to statistics, control theory, optimization, etc.) (13P25) Varieties and morphisms (14A10) Point estimation (62F10) Symbolic computation and algebraic computation (68W30)

Abstract: Given a statistical model, the maximum likelihood degree is the number of complex solutions to the likelihood equations for generic data. We consider discrete algebraic statistical models and study the solutions to the likelihood equations when the data contain zeros and are no longer generic. Focusing on sampling and model zeros, we show that, in these cases, the solutions to the likelihood equations are contained in a previously studied variety, the likelihood correspondence. The number of these solutions give a lower bound on the ML degree, and the problem of finding critical points to the likelihood function can be partitioned into smaller and computationally easier problems involving sampling and model zeros. We use this technique to compute a lower bound on the ML degree for $2imes2imes2imes2$ tensors of border rank $leq2$ and $3imesn$ tables of rank $leq2$ for $n=11,12,13,14$, the first four values of $n$ for which the ML degree was previously unknown.

Recommendations

• Geometry of maximum likelihood estimation in Gaussian graphical models
• Likelihood geometry
• Geometry of F-likelihood estimators and F-max-ent theorem
• Maximum likelihood estimation in convex hull models
• A Geometric Approach to Maximum Likelihood Estimation for Infinite-Dimensional Location. III
• Geometry of the log-likelihood ratio statistic in misspecified models
• Maximum likelihood estimation in Gaussian models under total positivity
• The empty set and zero likelihood problems in maximum empirical likelihood estimation
• Maximum likelihood estimation in linear infinite dimensional models

Cites work

• An effective implementation of symbolic-numeric cylindrical algebraic decomposition for quantifier elimination
• Cylindrical algebraic decomposition using validated numerics
• Definability and decision problems in arithmetic
• scientific article; zbMATH DE number 1952952 (Why is no real title available?)
• scientific article; zbMATH DE number 3068536 (Why is no real title available?)
• On Computable Numbers, with an Application to the Entscheidungsproblem
• The Jordan Curve Theorem, Formally and Informally

Cited in

(20)

• Maximum likelihood estimation of symmetric group-based models via numerical algebraic geometry
• The empty set and zero likelihood problems in maximum empirical likelihood estimation
• Likelihood equations and scattering amplitudes
• Discrete statistical models with rational maximum likelihood estimator
• Fixed points of the EM algorithm and nonnegative rank boundaries
• Solving the likelihood equations
• A probabilistic algorithm for computing data-discriminants of likelihood equations
• Likelihood degenerations
• Mini-workshop: Scattering amplitudes, cluster algebras, and positive geometries. Abstracts from the mini-workshop held December 5--11, 2021 (hybrid meeting)
• Algebraic boundary of matrices of nonnegative rank at most three
• The maximum likelihood data singular locus
• Likelihood geometry
• scientific article; zbMATH DE number 7476198 (Why is no real title available?)
• Critical points via monodromy and local methods
• Identifying the number of components in Gaussian mixture models using numerical algebraic geometry
• Bounding the maximum likelihood degree
• The maximum likelihood degree of mixtures of independence models
• scientific article; zbMATH DE number 2221430 (Why is no real title available?)
• Algebraic degree of optimization over a variety with an application to P-norm distance degree
• New directions in algebraic statistics: three challenges from 2023

Describes a project that uses

Uses Software

• Jordan

This page was built for publication: Maximum likelihood geometry in the presence of data zeros

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3452399)