Discovering the Gell-Mann-Okubo Formula with Kolmogorov-Arnold Networks

TL;DR

This paper demonstrates how Kolmogorov-Arnold Networks can autonomously discover the Gell-Mann-Okubo mass relations in baryon spectra, providing an interpretable, data-driven approach to uncover fundamental physical laws without prior assumptions.

Contribution

It introduces the application of spline-based Kolmogorov-Arnold Networks for symbolic regression in hadron physics, enabling discovery of classical mass relations and symmetry-breaking parameters without theoretical priors.

Findings

KAN recovers Gell-Mann-Okubo relations from data

Achieves comparable accuracy to traditional methods

Provides enhanced interpretability and transparency

Abstract

Uncovering physical laws from experimental data is a fundamental goal of theoretical physics. In this work, we apply the spline-based, interpretable Kolmogorov-Arnold Network (KAN) to explore the algebraic structure underlying the baryon octet and decuplet mass spectra. Within a symbolic regression framework and without imposing theoretical priors, KAN autonomously recovers the classical Gell-Mann-Okubo mass relations and accurately extracts the associated SU(3) symmetry-breaking parameters. Compared to conventional fitting approaches, this method achieves comparable predictive accuracy while offering substantially improved interpretability and analytic transparency. Our results demonstrate the potential of KAN as a powerful tool for symbolic discovery in hadron physics and for bridging data-driven modeling with fundamental physical laws.