Real birational implicitization for statistical models

TL;DR

This paper presents a method to derive implicit equations for the images of semialgebraic sets under birational maps, aiding statistical model analysis and testing, with broad applicability to various graphical and algebraic models.

Contribution

It introduces a novel approach to implicitization for statistical models using birational maps, enabling model membership testing and equivalence analysis.

Findings

Implicit equations recover classical Markov properties.

Method applies to generalized graphical models and Lyapunov models.

Generated ideals match the model's vanishing ideal up to saturation.

Abstract

We derive an implicit description of the image of a semialgebraic set under a birational map, provided that the denominators of the map are positive on the set. For statistical models which are globally rationally identifiable, this yields model-defining constraints which facilitate model membership testing, representation learning, and model equivalence tests. Many examples illustrate the applicability of our results. The implicit equations recover well-known Markov properties of classical graphical models, as well as other well-studied equations such as the Verma constraint. They also provide Markov properties for generalizations of these frameworks, such as colored or interventional graphical models, staged trees, and the recently introduced Lyapunov models. Under a further mild assumption, we show that our implicit equations generate the vanishing ideal of the model up to a saturation, generalizing previous results of Geiger, Meek and Sturmfels, Duarte and G\"orgen, Sullivant, and others.