Interpreting CFD Surrogates through Sparse Autoencoders

TL;DR

This paper presents a posthoc interpretability framework for graph-based CFD surrogate models using sparse autoencoders to extract interpretable features aligned with physical phenomena, enhancing trustworthiness.

Contribution

It introduces a novel method leveraging sparse autoencoders to interpret latent representations in CFD surrogates, improving explainability in safety-critical applications.

Findings

Extracted interpretable features aligned with physical concepts

Enabled identification of flow structures and vorticity

Enhanced trust in surrogate models for CFD

Abstract

Learning-based surrogate models have become a practical alternative to high-fidelity CFD solvers, but their latent representations remain opaque and hinder adoption in safety-critical or regulation-bound settings. This work introduces a posthoc interpretability framework for graph-based surrogate models used in computational fluid dynamics (CFD) by leveraging sparse autoencoders (SAEs). By obtaining an overcomplete basis in the node embedding space of a pretrained surrogate, the method extracts a dictionary of interpretable latent features. The approach enables the identification of monosemantic concepts aligned with physical phenomena such as vorticity or flow structures, offering a model-agnostic pathway to enhance explainability and trustworthiness in CFD applications.