Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

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

About

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.

Anuj Kumar, Josiah Bjorgaard, Nikolaos Bouklas, Matteo Salvador, Alexander Lavin• 2026

Related benchmarks

TaskDatasetResultRank
Neural Surrogate ModelingAirfoil
NMAE1.4
5
Neural Surrogate ModelingPipe
NMAE1.4
5
Neural Surrogate ModelingDiff-Sorp
NMAE3.8
5
Neural Surrogate ModelingNS-3D
NMAE1.7
5
Unsteady flow simulation on unstructured meshes2D Cylinder-Obstructed Flow
NMAE1.7
5
Neural Surrogate ModelingDarcy
NMAE5.7
5
Steady-state flow simulation on unstructured meshes3D Branched Pipe Flow (Br.Pipe)
NMAE9.6
4
Steady-state flow simulation on unstructured meshes3D Centrifugal Pump (Ctr.Pump)
NMAE1.7
3
Steady-state physical simulation on unstructured meshesElasticity (Elast.)
NMAE0.0021
3
Showing 9 of 9 rows

Other info

Follow for update