Sequential Importance Sampling for Two-dimensional Ising Models

TL;DR

This paper introduces a sequential importance sampling method tailored for 2D Ising models with 0-1 tables, efficiently handling linear and quadratic constraints, and demonstrating strong performance on sparse data.

Contribution

The paper adapts SIS to 2D Ising models with complex constraints, providing a practical approach with computational bounds and high acceptance rates.

Findings

Method performs well on sparse tables

Computational times are short and acceptance rates high

Results are theoretically reasonable with proper testing

Abstract

In recent years, sequential importance sampling (SIS) has been well developed for sampling contingency tables with linear constraints. In this paper, we apply SIS procedure to 2-dimensional Ising models, which give observations of 0-1 tables and include both linear and quadratic constraints. We show how to compute bounds for specific cells by solving linear programming (LP) problems over cut polytopes to reduce rejections. The computational results, which includes both simulations and real data analysis, suggest that our method performs very well for sparse tables and when the 1's are spread out: the computational times are short, the acceptance rates are high, and if proper tests are used then in most cases our conclusions are theoretically reasonable.