High Performance Evaluation of Helmholtz Potentials using the Multi-Level Fast Multipole Algorithm

06/27/2020
by   Michael P. Lingg, et al.
0

Evaluation of pair potentials is critical in a number of areas of physics. The classicalN-body problem has its root in evaluating the Laplace potential, and has spawned tree-algorithms, the fast multipole method (FMM), as well as kernel independent approaches. Over the years, FMM for Laplace potential has had a profound impact on a number of disciplines as it has been possible to develop highly scalable parallel algorithm for these potential evaluators. This is in stark contrast to parallel algorithms for the Helmholtz (oscillatory) potentials. The principal bottleneck to scalable parallelism are operations necessary to traverse up, across and down the tree, affecting both computation and communication. In this paper, we describe techniques to overcome bottlenecks and achieve high performance evaluation of the Helmholtz potential for a wide spectrum of geometries. We demonstrate that the resulting implementation has a load balancing effect that significantly reduces the time-to-solution and enhances the scale of problems that can be treated using full wave physics.

READ FULL TEXT

Authors

06/27/2020

High Performance Evaluation of Helmholtz Potentials usingthe Multi-Level Fast Multipole Algorithm

Evaluation of pair potentials is critical in a number of areas of physic...
08/03/2020

On the Approximation of Local Expansions of Laplace Potentials by the Fast Multipole Method

In this paper, we present a generalization of the classical error bounds...
07/15/2020

Layer potential identities and subtraction techniques

When using boundary integral equation methods, we represent solutions of...
10/19/2017

Applications of potential theoretic mother bodies in Electrostatics

Any polyhedron accommodates a type of potential theoretic skeleton calle...
04/12/2021

Software-Hardware Co-design for Fast and Scalable Training of Deep Learning Recommendation Models

Deep learning recommendation models (DLRMs) are used across many busines...
04/06/2022

The Challenge of Sixfold Integrals: The Closed-Form Evaluation of Newton Potentials between Two Cubes

The challenge of explicitly evaluating, in elementary closed form, the w...
10/28/2020

Fast localization of eigenfunctions via smoothed potentials

We study the problem of predicting highly localized low-lying eigenfunct...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.