Hamiltonian Matching for Symplectic Neural Integrators

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Canizares, Priscilla, Murari, Davide, Schönlieb, Carola-Bibiane, Sherry, Ferdia, Shumaylov, Zakhar
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916451519037440
author Canizares, Priscilla
Murari, Davide
Schönlieb, Carola-Bibiane
Sherry, Ferdia
Shumaylov, Zakhar
author_facet Canizares, Priscilla
Murari, Davide
Schönlieb, Carola-Bibiane
Sherry, Ferdia
Shumaylov, Zakhar
contents Hamilton's equations of motion form a fundamental framework in various branches of physics, including astronomy, quantum mechanics, particle physics, and climate science. Classical numerical solvers are typically employed to compute the time evolution of these systems. However, when the system spans multiple spatial and temporal scales numerical errors can accumulate, leading to reduced accuracy. To address the challenges of evolving such systems over long timescales, we propose SympFlow, a novel neural network-based symplectic integrator, which is the composition of a sequence of exact flow maps of parametrised time-dependent Hamiltonian functions. This architecture allows for a backward error analysis: we can identify an underlying Hamiltonian function of the architecture and use it to define a Hamiltonian matching objective function, which we use for training. In numerical experiments, we show that SympFlow exhibits promising results, with qualitative energy conservation behaviour similar to that of time-stepping symplectic integrators.
format Preprint
id arxiv_https___arxiv_org_abs_2410_18262
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Hamiltonian Matching for Symplectic Neural Integrators
Canizares, Priscilla
Murari, Davide
Schönlieb, Carola-Bibiane
Sherry, Ferdia
Shumaylov, Zakhar
Machine Learning
Numerical Analysis
Computational Physics
Hamilton's equations of motion form a fundamental framework in various branches of physics, including astronomy, quantum mechanics, particle physics, and climate science. Classical numerical solvers are typically employed to compute the time evolution of these systems. However, when the system spans multiple spatial and temporal scales numerical errors can accumulate, leading to reduced accuracy. To address the challenges of evolving such systems over long timescales, we propose SympFlow, a novel neural network-based symplectic integrator, which is the composition of a sequence of exact flow maps of parametrised time-dependent Hamiltonian functions. This architecture allows for a backward error analysis: we can identify an underlying Hamiltonian function of the architecture and use it to define a Hamiltonian matching objective function, which we use for training. In numerical experiments, we show that SympFlow exhibits promising results, with qualitative energy conservation behaviour similar to that of time-stepping symplectic integrators.
title Hamiltonian Matching for Symplectic Neural Integrators
topic Machine Learning
Numerical Analysis
Computational Physics
url https://arxiv.org/abs/2410.18262