On the feasibility of foundational models for the simulation of physical phenomena

TL;DR

This paper investigates whether foundational models can reliably simulate physical phenomena like solid and fluid mechanics, especially under significant domain and condition changes, assessing their robustness and accuracy.

Contribution

It provides an exhaustive analysis of the robustness of learned simulators to drastic domain, boundary, and law changes in physical simulations.

Findings

Models maintain robustness under domain shifts

Simulators show limited hallucinations in extreme scenarios

Performance varies with the extent of domain change

Abstract

We explore the feasibility of foundation models for the simulation of physical phenomena, with emphasis on continuum (solid and fluid) mechanics. Although so-called learned simulators have shown some success when applied to specific tasks, it remains to be studied to what extent they are able to undergo severe changes in domain shape, boundary conditions and/or constitutive laws and still provide robust (i.e., hallucination-free) and accurate results. In this paper we perform an exhaustive study of these features, put ourselves in the worst-case scenario and study their resistance to such strong changes in their domain of application.