Trefftz Functions for Nonlocal Electrostatics

TL;DR

This paper introduces Trefftz approximations as an efficient computational tool to localize and solve nonlocal electrostatic models, reducing computational costs and improving accuracy in biophysical simulations.

Contribution

The paper develops Trefftz-based methods specifically for nonlocal electrostatics, enabling more accurate and computationally feasible solutions compared to traditional polynomial approaches.

Findings

Trefftz approximations improve accuracy over polynomial methods.

The approach effectively localizes nonlocal electrostatic problems.

Potential extension to other nonlocal problems in science and engineering.

Abstract

Electrostatic interactions in solvents play a major role in biophysical systems. There is a consensus in the literature that the dielectric response of aqueous solutions is nonlocal: polarization depends on the electric field not only at a given point but in the vicinity of that point as well. This is typically modeled via a convolution of the electric field with an appropriate integral kernel. A primary problem with nonlocal models is high computational cost. A secondary problem is restriction of convolution integrals to the solvent, as opposed to their evaluation over the whole space. The paper develops a computational tool alleviating the "curse of nonlocality" and helping to handle the integration correctly. This tool is Trefftz approximations, which tend to furnish much higher accuracy than traditional polynomial ones. In the paper, Trefftz approximations are developed for problems of nonlocal electrostatics, with the goal of numerically "localizing" the original nonlocal problem. This approach can be extended to nonlocal problems in other areas of computational mathematics, physics and engineering.