Scalable physical source-to-field inference with hypernetworks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: James, Berian, Pollok, Stefan, Peis, Ignacio, Baker, Elizabeth Louise, Frellsen, Jes, Bjørk, Rasmus
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908812182552576
author James, Berian
Pollok, Stefan
Peis, Ignacio
Baker, Elizabeth Louise
Frellsen, Jes
Bjørk, Rasmus
author_facet James, Berian
Pollok, Stefan
Peis, Ignacio
Baker, Elizabeth Louise
Frellsen, Jes
Bjørk, Rasmus
contents We present a generative model that amortises computation for the field and potential around e.g.~gravitational or electromagnetic sources. Exact numerical calculation has either computational complexity $\mathcal{O}(M\times{}N)$ in the number of sources $M$ and evaluation points $N$, or requires a fixed evaluation grid to exploit fast Fourier transforms. Using an architecture where a hypernetwork produces an implicit representation of the field or potential around a source collection, our model instead performs as $\mathcal{O}(M + N)$, achieves relative error of $\sim\!4\%-6\%$, and allows evaluation at arbitrary locations for arbitrary numbers of sources, greatly increasing the speed of e.g.~physics simulations. We compare with existing models and develop two-dimensional examples, including cases where sources overlap or have more complex geometries, to demonstrate its application.
format Preprint
id arxiv_https___arxiv_org_abs_2405_05981
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Scalable physical source-to-field inference with hypernetworks
James, Berian
Pollok, Stefan
Peis, Ignacio
Baker, Elizabeth Louise
Frellsen, Jes
Bjørk, Rasmus
Machine Learning
Computational Engineering, Finance, and Science
Computational Physics
We present a generative model that amortises computation for the field and potential around e.g.~gravitational or electromagnetic sources. Exact numerical calculation has either computational complexity $\mathcal{O}(M\times{}N)$ in the number of sources $M$ and evaluation points $N$, or requires a fixed evaluation grid to exploit fast Fourier transforms. Using an architecture where a hypernetwork produces an implicit representation of the field or potential around a source collection, our model instead performs as $\mathcal{O}(M + N)$, achieves relative error of $\sim\!4\%-6\%$, and allows evaluation at arbitrary locations for arbitrary numbers of sources, greatly increasing the speed of e.g.~physics simulations. We compare with existing models and develop two-dimensional examples, including cases where sources overlap or have more complex geometries, to demonstrate its application.
title Scalable physical source-to-field inference with hypernetworks
topic Machine Learning
Computational Engineering, Finance, and Science
Computational Physics
url https://arxiv.org/abs/2405.05981