Neural operators struggle to learn complex PDEs in pedestrian mobility: Hughes model case study

2025-04-25

AI Summary

Mainstream neural operators (FNO/WNO/MWNO) exhibit fundamental limitations in solving the Hughes model for pedestrian dynamics—a nonlinear hyperbolic conservation law system coupling the Fokker–Planck and Eikonal equations—particularly under multi-discontinuous initial conditions and dynamic boundary conditions. Method: We conduct a systematic empirical analysis of solution accuracy, quantifying shock localization error and total variation (TV) decay, and identify the implicit regularization mechanism underlying these operators. Contribution/Results: We demonstrate that neural operators induce excessive smoothing, yielding a 62% shock position deviation and an average 37% TV reduction—comparable to Daganzo’s artificial diffusion. This implicit regularization inherently degrades fidelity to discontinuous transport structures. Our work is the first to rigorously establish this deficiency, challenging the physical plausibility of neural operators in shock-sensitive applications such as traffic flow modeling, and providing critical guidance for designing hyperbolic-PDE-aware operator architectures.

Technology Category

• Search and Optimization: Non-convex Optimization
• Machine Learning: Mixture of Experts (MoE)
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Social Networks and Social Media: Media and governance, opinion dynamics, filter bubbles, polarization

Abstract

This paper investigates the limitations of neural operators in learning solutions for a Hughes model, a first-order hyperbolic conservation law system for crowd dynamics. The model couples a Fokker-Planck equation representing pedestrian density with a Hamilton-Jacobi-type (eikonal) equation. This Hughes model belongs to the class of nonlinear hyperbolic systems that often exhibit complex solution structures, including shocks and discontinuities. In this study, we assess the performance of three state-of-the-art neural operators (Fourier Neural Operator, Wavelet Neural Operator, and Multiwavelet Neural Operator) in various challenging scenarios. Specifically, we consider (1) discontinuous and Gaussian initial conditions and (2) diverse boundary conditions, while also examining the impact of different numerical schemes. Our results show that these neural operators perform well in easy scenarios with fewer discontinuities in the initial condition, yet they struggle in complex scenarios with multiple initial discontinuities and dynamic boundary conditions, even when trained specifically on such complex samples. The predicted solutions often appear smoother, resulting in a reduction in total variation and a loss of important physical features. This smoothing behavior is similar to issues discussed by Daganzo (1995), where models that introduce artificial diffusion were shown to miss essential features such as shock waves in hyperbolic systems. These results suggest that current neural operator architectures may introduce unintended regularization effects that limit their ability to capture transport dynamics governed by discontinuities. They also raise concerns about generalizing these methods to traffic applications where shock preservation is essential.

Problem

• Neural operators struggle with Hughes model's discontinuities and shocks
• Performance drops in complex scenarios with dynamic boundary conditions
• Current architectures introduce unintended smoothing, losing key physical features

Innovation

• Neural operators learn Hughes model dynamics
• Tested with discontinuous initial conditions
• Evaluated shock preservation in solutions