HyCOP: Hybrid Composition Operators for Interpretable Learning of PDEs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Jinpai, Panda, Nishant, Lin, Yen Ting, Valseth, Eirik, Oyen, Diane, Dawson, Clint
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914524938895360
author Zhao, Jinpai
Panda, Nishant
Lin, Yen Ting
Valseth, Eirik
Oyen, Diane
Dawson, Clint
author_facet Zhao, Jinpai
Panda, Nishant
Lin, Yen Ting
Valseth, Eirik
Oyen, Diane
Dawson, Clint
contents We introduce HyCOP, a modular framework that learns parametric PDE solution operators by composing simple modules (advection, diffusion, learned closures, boundary handling) in a query-conditioned way. Rather than learning a monolithic map, HyCOP learns a policy over short programs - which module to apply and for how long - conditioned on regime features and state statistics. Modules may be numerical sub-solvers or learned components, enabling hybrid surrogates evaluated at arbitrary query times without autoregressive rollout. Across diverse PDE benchmarks, HyCOP produces interpretable programs, delivers order-of-magnitude OOD improvements over monolithic neural operators, and supports modular transfer through dictionary updates (e.g., boundary swaps, residual enrichment). Our theory characterizes expressivity and gives an error decomposition that separates composition error from module error and doubles as a process-level diagnostic.
format Preprint
id arxiv_https___arxiv_org_abs_2605_00820
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle HyCOP: Hybrid Composition Operators for Interpretable Learning of PDEs
Zhao, Jinpai
Panda, Nishant
Lin, Yen Ting
Valseth, Eirik
Oyen, Diane
Dawson, Clint
Computational Engineering, Finance, and Science
Machine Learning
Numerical Analysis
We introduce HyCOP, a modular framework that learns parametric PDE solution operators by composing simple modules (advection, diffusion, learned closures, boundary handling) in a query-conditioned way. Rather than learning a monolithic map, HyCOP learns a policy over short programs - which module to apply and for how long - conditioned on regime features and state statistics. Modules may be numerical sub-solvers or learned components, enabling hybrid surrogates evaluated at arbitrary query times without autoregressive rollout. Across diverse PDE benchmarks, HyCOP produces interpretable programs, delivers order-of-magnitude OOD improvements over monolithic neural operators, and supports modular transfer through dictionary updates (e.g., boundary swaps, residual enrichment). Our theory characterizes expressivity and gives an error decomposition that separates composition error from module error and doubles as a process-level diagnostic.
title HyCOP: Hybrid Composition Operators for Interpretable Learning of PDEs
topic Computational Engineering, Finance, and Science
Machine Learning
Numerical Analysis
url https://arxiv.org/abs/2605.00820