Horizon tracking for asynchronous parallel black hole simulations

TL;DR

This paper introduces novel asynchronous horizon tracking methods for black hole merger simulations, enhancing efficiency and accuracy in numerical relativity to support next-generation gravitational wave detectors.

Contribution

It presents new algorithms for dynamically tracking apparent horizons in asynchronous parallel black hole simulations, improving computational efficiency and accuracy.

Findings

Effective horizon tracking in asynchronous simulations

Enhanced simulation efficiency for black hole mergers

Potential application to neutron star simulations

Abstract

In the field of gravitational wave science, next-generation detectors will be substantially more accurate than the current suite of detectors. Numerical relativity simulations of binary black hole (BBH) gravitational waveforms must become faster, more efficient, and more accurate to be used in analyses of these next-generation detections. One approach, which the $\texttt{SpECTRE}$ code employs, is using spectral methods for accuracy along with asynchronous task-based parallelism to avoid idle time in simulations and make the most efficient use of computational resources. When writing an asynchronous application, algorithms must be redesigned compared to their synchronous counterparts. To illustrate this process, we present novel methods for dynamically tracking the apparent horizons in evolutions of BBH mergers using a feedback control system, all in the context of asynchronous parallelism. We also briefly detail how these methods can be applied to binary neutron star simulations performed with asynchronous parallelism.