2609.05207

Total: 1

#1 FluxDisco: Symbolic Regression for Stoichiometric Dynamical Systems via Monte Carlo Graph Search [PDF] [Copy] [Kimi] [REL]

Authors: Cassandra Durr, Alvaro Köhn-Luque, Chris Jewell, Lloyd A. C. Chapman

Dynamical symbolic regression methods identify governing differential equations from noisy data, balancing interpretability and predictive accuracy. However, standard methods often produce expressions that violate known physical laws. To address this, we propose FluxDisco, a physics-informed framework tailored for flux-based, stoichiometric ODE systems. By leveraging a known stoichiometry, we reduce the expression search space and ensure physical adherence. Our framework adapts the Monte Carlo Graph Search algorithm for the unique challenges associated with joint flux discovery of stoichiometric systems. We evaluate our method across a range of physical and biological systems, demonstrating its ability to accurately recover governing dynamics through interpretable equations.

Subjects: Machine Learning , Machine Learning , Data Analysis, Statistics and Probability

Publish: 2026-09-04 14:38:59 UTC