AI-Accelerated Operator Learning Framework for Rarefied Microflows

TL;DR

This paper introduces a neural operator framework that significantly accelerates rarefied microflow simulations by combining physics-guided neural networks with surrogate modeling, enabling real-time analysis without sacrificing accuracy.

Contribution

It presents a novel unified deep learning framework that integrates neural operators and physics-guided architectures to efficiently model complex rarefied gas flows.

Findings

Achieves significant speedups in flow simulations.

Maintains high accuracy across multiple flow regimes.

Enhances robustness with uncertainty quantification methods.

Abstract

The high computational cost of kinetic solvers such as DSMC remains a major challenge in rarefied flow simulations. This work presents a unified framework combining deep neural networks and neural operators to accelerate kinetic and hybrid solvers while preserving physical fidelity. GPU-native DNN surrogates eliminate costly moment-closure operations in Fokker Planck methods, achieving significant speedups without accuracy loss, while physics-guided and shock-aware DeepONet architectures enable accurate, data efficient modeling of multi regime micro nozzle, micro-step, and hypersonic flows. Extensions including ensemble uncertainty quantification and family-of-experts strategies further enhance robustness across wide Mach and Knudsen number ranges. Together, these results demonstrate a scalable and physics-consistent pathway toward real-time surrogate modeling in rarefied gas dynamics.