GRMHD simulations and analysis of polarized emission from black hole accretion systems

Prather, Benjamin

Permalink

https://hdl.handle.net/2142/117717 Copy

Description

Title

GRMHD simulations and analysis of polarized emission from black hole accretion systems

Author(s)

Prather, Benjamin

Issue Date

2022-09-08

Director of Research (if dissertation) or Advisor (if thesis)

Gammie, Charles

Doctoral Committee Chair(s)

Holder, Gilbert

Committee Member(s)

• Filippini, Jeffrey
• Witek, Helvi
• Ricker, Paul

Department of Study

Physics

Discipline

Physics

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• black holes
• accretion
• magnetohydrodynamics
• MHD
• GRMHD
• GPUs
• performance portability
• radiative transfer
• polarization

Abstract

An unprecedented and fast-growing number of high-precision observations of certain supermassive black holes are now available, including images at the scale of the black hole event horizon. These provide powerful combined constraints which can be used in understanding both the particular systems under study, and general principles of accretion and jet physics. Interpretation of these results in concert requires sophisticated global models of accretion, provided by general-relativistic magnetohydrodynamic (GRMHD) simulations. At the same time, in the pursuit of increased power efficiency, computer hardware manufacturers and system designers are exploring new architectural avenues in place of the traditional in-band increases to performance. A diverse set of increasingly specialized hardware demands more portable and flexible software, but offers dramatically increased computing resources in return. After providing some background on the problem and on the equations and methods of GRMHD, I describe the simulation code iharm3D, focusing on speed and scaling improvements implemented by myself in service of running more and longer simulations on specialized CPU architectures. I then introduce a new performance-portable GRMHD code I have written, KHARMA, which much more comprehensively addresses the need for longer and more expansive simulations by allowing users to leverage the capability of many different processor architectures. It makes use of existing tools and modular architecture to remain extensible and easy to modify while performing well. I address the computer hardware environment and design goals which motivate writing a new GRMHD code, and describe how KHARMA is designed to meet these goals. I provide validation test results, as well as performance and scaling data from several supercomputers using various architectures. I then discuss libraries of GRMHD simulations created for the Event Horizon Telescope (EHT) Collaboration using iharm3D and now KHARMA. I describe the simulation properties, parameter coverage, coordinates, and provide some general measurements of the library simulations. Simulations from a new and much larger library with increased parameter coverage are compared with existing results. In order to evaluate a GRMHD model against observations, simulated images and spectral energy distributions must be produced which predict the observable consequences of the model’s fluid configuration. Of special interest is the 230GHz emission measured by the EHT, a global network of millimeter-wavelength observatories capable of resolving certain systems at the event horizon scale. I discuss improvements made to ipole, a code implementing the general-relativistic ray-tracing (GRRT) process for producing simulated images suitable for direct comparisons to EHT observations. I also discuss patterns in simulated images which may aid in model discrimination, including an easy-to-measure pattern in polarized emission which I helped to validate, and now apply to new and broader sets of simulated images. Finally, I discuss a project I’ve led comparing several GRRT codes used by the EHT Collaboration, including ipole, showing that despite using several different transport schemes, all codes agree to excellent accuracy in comparison to detector and modeling uncertainties.

Graduation Semester

2022-12

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Benjamin Prather

Owning Collections