Courant: a State-Adaptive Perceiver-Based Neural Surrogate with Local Support and Interpretable Field Decomposition

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Kumar, Anuj, Bjorgaard, Josiah, Bouklas, Nikolaos, Salvador, Matteo, Lavin, Alexander
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916043296866304
author Kumar, Anuj
Bjorgaard, Josiah
Bouklas, Nikolaos
Salvador, Matteo
Lavin, Alexander
author_facet Kumar, Anuj
Bjorgaard, Josiah
Bouklas, Nikolaos
Salvador, Matteo
Lavin, Alexander
contents We introduce "Courant", a Perceiver-based encoder-processor-decoder surrogate model that has latent features exhibiting adaptive specialization and local support in the physical space, enabling functionality akin to an adaptive hp-refinement scheme, an attribute that is highly desirable in traditional numerical solvers and scientific machine learning broadly. The proposed architecture combines a shared random Fourier feature coordinate embedding, state-adapted latent queries, and a light-weight decoder. Courant is trained end-to-end with steady or transient simulation data and only a standard L_2 prediction loss in the physical space, achieving competitive accuracy on benchmarks. We demonstrate that Courant's inductive biases yield latents that are interpretable by design: they develop multiscale geometric specialization in the simulation domain and track coherent structures in the time-dependent case, acting analogously to time-evolving spatial basis functions and allowing for decoding a compact, geometry-anchored, partition-of-unity-like decomposition of the simulated field.
format Preprint
id arxiv_https___arxiv_org_abs_2605_25115
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Courant: a State-Adaptive Perceiver-Based Neural Surrogate with Local Support and Interpretable Field Decomposition
Kumar, Anuj
Bjorgaard, Josiah
Bouklas, Nikolaos
Salvador, Matteo
Lavin, Alexander
Machine Learning
Artificial Intelligence
Computational Engineering, Finance, and Science
Applied Physics
We introduce "Courant", a Perceiver-based encoder-processor-decoder surrogate model that has latent features exhibiting adaptive specialization and local support in the physical space, enabling functionality akin to an adaptive hp-refinement scheme, an attribute that is highly desirable in traditional numerical solvers and scientific machine learning broadly. The proposed architecture combines a shared random Fourier feature coordinate embedding, state-adapted latent queries, and a light-weight decoder. Courant is trained end-to-end with steady or transient simulation data and only a standard L_2 prediction loss in the physical space, achieving competitive accuracy on benchmarks. We demonstrate that Courant's inductive biases yield latents that are interpretable by design: they develop multiscale geometric specialization in the simulation domain and track coherent structures in the time-dependent case, acting analogously to time-evolving spatial basis functions and allowing for decoding a compact, geometry-anchored, partition-of-unity-like decomposition of the simulated field.
title Courant: a State-Adaptive Perceiver-Based Neural Surrogate with Local Support and Interpretable Field Decomposition
topic Machine Learning
Artificial Intelligence
Computational Engineering, Finance, and Science
Applied Physics
url https://arxiv.org/abs/2605.25115