English

Arithmetic Circuits and Neural Networks for Regular Matroids

Combinatorics 2025-11-05 v1 Computational Complexity Discrete Mathematics Machine Learning Optimization and Control

Abstract

We prove that there exist uniform (+,×,/)(+,\times,/)-circuits of size O(n3)O(n^3) to compute the basis generating polynomial of regular matroids on nn elements. By tropicalization, this implies that there exist uniform (max,+,)(\max,+,-)-circuits and ReLU neural networks of the same size for weighted basis maximization of regular matroids. As a consequence in linear programming theory, we obtain a first example where taking the difference of two extended formulations can be more efficient than the best known individual extended formulation of size O(n6)O(n^6) by Aprile and Fiorini. Such differences have recently been introduced as virtual extended formulations. The proof of our main result relies on a fine-tuned version of Seymour's decomposition of regular matroids which allows us to identify and maintain graphic substructures to which we can apply a local version of the star-mesh transformation.

Keywords

Cite

@article{arxiv.2511.02406,
  title  = {Arithmetic Circuits and Neural Networks for Regular Matroids},
  author = {Christoph Hertrich and Stefan Kober and Georg Loho},
  journal= {arXiv preprint arXiv:2511.02406},
  year   = {2025}
}
R2 v1 2026-07-01T07:20:54.087Z