Learning Lagrangian Interaction Dynamics with Sampling-Based Model Order Reduction

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Viswanath, Hrishikesh, Chang, Yue, Panas, Aleksey, Berner, Julius, Chen, Peter Yichen, Bera, Aniket
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911479996874752
author Viswanath, Hrishikesh
Chang, Yue
Panas, Aleksey
Berner, Julius
Chen, Peter Yichen
Bera, Aniket
author_facet Viswanath, Hrishikesh
Chang, Yue
Panas, Aleksey
Berner, Julius
Chen, Peter Yichen
Bera, Aniket
contents Simulating physical systems governed by Lagrangian dynamics often entails solving partial differential equations (PDEs) over high-resolution spatial domains, leading to significant computational expense. Reduced-order modeling (ROM) mitigates this cost by evolving low-dimensional latent representations of the underlying system. While neural ROMs enable querying solutions from latent states at arbitrary spatial points, their latent states typically represent the global domain and struggle to capture localized, highly dynamic behaviors such as fluids. We propose a sampling-based reduction framework that evolves Lagrangian systems directly in physical space over the particles themselves, reducing the number of active degrees of freedom via data-driven neural PDE operators. To enable querying at arbitrary spatial locations, we introduce a learnable kernel parameterization that uses local spatial information from time-evolved sample particles to infer the underlying solution manifold. Empirically, our approach achieves a 6.6x to 32x reduction in input dimensionality while maintaining high-fidelity evaluations across diverse Lagrangian regimes, including fluid flows, granular media, and elastoplastic dynamics. We refer to this framework as GIOROM (Geometry-Informed Reduced-Order Modeling). All code and data are available at: https://github.com/HrishikeshVish/GIOROM
format Preprint
id arxiv_https___arxiv_org_abs_2407_03925
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Learning Lagrangian Interaction Dynamics with Sampling-Based Model Order Reduction
Viswanath, Hrishikesh
Chang, Yue
Panas, Aleksey
Berner, Julius
Chen, Peter Yichen
Bera, Aniket
Machine Learning
Simulating physical systems governed by Lagrangian dynamics often entails solving partial differential equations (PDEs) over high-resolution spatial domains, leading to significant computational expense. Reduced-order modeling (ROM) mitigates this cost by evolving low-dimensional latent representations of the underlying system. While neural ROMs enable querying solutions from latent states at arbitrary spatial points, their latent states typically represent the global domain and struggle to capture localized, highly dynamic behaviors such as fluids. We propose a sampling-based reduction framework that evolves Lagrangian systems directly in physical space over the particles themselves, reducing the number of active degrees of freedom via data-driven neural PDE operators. To enable querying at arbitrary spatial locations, we introduce a learnable kernel parameterization that uses local spatial information from time-evolved sample particles to infer the underlying solution manifold. Empirically, our approach achieves a 6.6x to 32x reduction in input dimensionality while maintaining high-fidelity evaluations across diverse Lagrangian regimes, including fluid flows, granular media, and elastoplastic dynamics. We refer to this framework as GIOROM (Geometry-Informed Reduced-Order Modeling). All code and data are available at: https://github.com/HrishikeshVish/GIOROM
title Learning Lagrangian Interaction Dynamics with Sampling-Based Model Order Reduction
topic Machine Learning
url https://arxiv.org/abs/2407.03925