Interpreting CFD Surrogates through Sparse Autoencoders

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Hu, Yeping, Liu, Shusen
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913952328318976
author Hu, Yeping
Liu, Shusen
author_facet Hu, Yeping
Liu, Shusen
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2507_16069
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Interpreting CFD Surrogates through Sparse Autoencoders
Hu, Yeping
Liu, Shusen
Computational Engineering, Finance, and Science
Machine Learning
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.
title Interpreting CFD Surrogates through Sparse Autoencoders
topic Computational Engineering, Finance, and Science
Machine Learning
url https://arxiv.org/abs/2507.16069