π€ AI Summary
This work investigates the discrepancy between the permanent of non-negative block-structured matrices and its Bethe approximation, which can theoretically differ by an exponential factor yet exhibits a surprisingly concentrated ratio in practiceβan observation whose origin has remained unclear. For the first time, the authors introduce the double-cover graph method to analyze the Bethe permanent of block-structured matrices. By integrating graph cover theory, the sum-product algorithm on factor graphs, and numerical experiments, they demonstrate that this concentration arises from intrinsic structural properties of the matrices. Specifically, the ratio concentrates around values determined by only a few key parameters, a phenomenon the authors successfully explain and quantify within a graph-cover framework, thereby providing a theoretical foundation for understanding the empirical effectiveness of the Bethe approximation in structured matrices.
π Abstract
We consider the permanent of a square matrix with non-negative entries. A tractable approximation is given by the so-called Bethe permanent that can be efficiently computed by running the sum-product algorithm on a suitable factor graph. While the ratio of the permanent of a matrix to its Bethe permanent is, in the worst case, upper and lower bounded by expressions that are exponentially far apart in the matrix size, in practice it is observed for many ensembles of matrices of interest that this ratio is strongly concentrated around some value that depends only on the matrix size. In this paper, for an ensemble of block-structured matrices where entries in a block take the same value, we numerically study the ratio of the permanent of a matrix to its Bethe permanent. It is observed that also for this ensemble the ratio is strongly concentrated around some value depending only on a few key parameters of the ensemble. We use graph-cover-based approaches to explain the reasons for this behavior and to quantify the observed value.