AutomataScales: Computationally Efficient Multiphysics Simulation for Early-Stage System Design

Abstract This article introduces AutomataScales, a new simulation method for early-stage system design employing a discrete deterministic formulation grounded in physical and probabilistic theories. By incorporating statistical and quantum-based transitions with a cellular automation framework, AutomataScales enables energy-driven multi-cell propagation, providing a more realistic representation of high-energy particle behavior and complex multiphysics interactions. The design of electric propulsion systems for deep space missions, which require precise and time-efficient simulations, is used to validate the AutomataScales method. Results demonstrate intricate and accurate behaviors in various particle trajectories, and plasma with an approximate error of 5.74% across four scenarios, while achieving up to 36.7× faster runtimes and up to 1.86 times lower memory than the comsol multiphysics simulation tool. A scaling analysis from a 2D to 2.5D model shows near-linear growth with cell count. To demonstrate generality, the FHP-I lattice gas model is implemented as transitional rules for a 2D microchannel flow model, obtaining a mean exit-error at 8.6% from ansys fluent. These results are validated and demonstrate that AutomataScales is more efficient for early design of 2D and 2.5D models. The key contributions of this work are the development of a fast, lightweight, and versatile simulation framework; a novel particle dynamics model that supports multiple energy stages and multi-cell propagation capable of overcoming limitations of previous methods; and the integration of a multiscale, multiphysics simulation capability for early-stage system design.