Automatic discovery of optimal meta-solvers for time-dependent nonlinear PDEs

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Lee, Youngkyu, Liu, Shanqing, Darbon, Jerome, Karniadakis, George Em
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866918076709076992
author Lee, Youngkyu
Liu, Shanqing
Darbon, Jerome
Karniadakis, George Em
author_facet Lee, Youngkyu
Liu, Shanqing
Darbon, Jerome
Karniadakis, George Em
contents We present a general and scalable framework for the automated discovery of optimal meta-solvers for the solution of time-dependent nonlinear partial differential equations after appropriate discretization. By integrating classical numerical methods (e.g., Krylov-based methods) with modern deep learning components, such as neural operators, our approach enables flexible, on-demand solver design tailored to specific problem classes and objectives. The fast solvers tackle the large linear system resulting from the Newton--Raphson iteration or by using an implicit-explicit (IMEX) time integration scheme. Specifically, we formulate solver discovery as a multi-objective optimization problem, balancing various performance criteria such as accuracy, speed, and memory usage. The resulting Pareto optimal set provides a principled foundation for solver selection based on user-defined preference functions. When applied to problems in reaction--diffusion, fluid dynamics, and solid mechanics, the discovered meta-solvers consistently outperform conventional iterative methods, demonstrating both practical efficiency and broad applicability.
format Preprint
id arxiv_https___arxiv_org_abs_2507_00278
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Automatic discovery of optimal meta-solvers for time-dependent nonlinear PDEs
Lee, Youngkyu
Liu, Shanqing
Darbon, Jerome
Karniadakis, George Em
Numerical Analysis
We present a general and scalable framework for the automated discovery of optimal meta-solvers for the solution of time-dependent nonlinear partial differential equations after appropriate discretization. By integrating classical numerical methods (e.g., Krylov-based methods) with modern deep learning components, such as neural operators, our approach enables flexible, on-demand solver design tailored to specific problem classes and objectives. The fast solvers tackle the large linear system resulting from the Newton--Raphson iteration or by using an implicit-explicit (IMEX) time integration scheme. Specifically, we formulate solver discovery as a multi-objective optimization problem, balancing various performance criteria such as accuracy, speed, and memory usage. The resulting Pareto optimal set provides a principled foundation for solver selection based on user-defined preference functions. When applied to problems in reaction--diffusion, fluid dynamics, and solid mechanics, the discovered meta-solvers consistently outperform conventional iterative methods, demonstrating both practical efficiency and broad applicability.
title Automatic discovery of optimal meta-solvers for time-dependent nonlinear PDEs
topic Numerical Analysis
url https://arxiv.org/abs/2507.00278