Repository logo
 

Computationally Intensive Problems of Physics and Astronomy: Oscillator Strengths and Departure Coefficients of the Hydrogen Atom in the Interstellar Medium

Files

Whole thesis(983.35 KB)

Date

2013

Authors

Supervisor

Gulyaev, Sergei
Ensor, Andrew

Item type

Thesis

Degree name

Master of Science

Journal Title

Journal ISSN

Volume Title

Publisher

Auckland University of Technology

Abstract

Calculating departure coefficients, bn, as well as bnl, for non-LTE gases/plasmas, is a fundamental computational problem in radio astronomy, physics of the interstellar medium, and for diagnostics of plasmas of nuclear reactors. Most work in this area was done in the 1960s and 1970s. Recent advances in computing technology have rendered the technology used in these two decades obsolete. Hence we ask if the approximate techniques developed to compensate for the technological limitations of the 1960s and 1970s are still needed. In this thesis we introduce modern computational techniques to solve, exactly, the computational problems relating to departure coefficients. Specifically, we have made use of arbitrary precision arithmetic as well as introducing GPU & parallelization techniques to already established solutions. We investigated the problem of the hypergeometric function which arises as the solution of the wave equation for hydrogen, which is the key component in Einstein coefficients, radiative recombination rates and the Stark broadening theory. Furthermore, we implemented, optimized and compared two different techniques for calculating bn coefficients and developed a matrix approach for dealing with the bnl problem. We hope that the solutions resulting from this thesis will pave the way for further development in the outlined area, allowing for exact solutions up to n = 1000 and greater.

Description

Keywords

High Performance Computing, GPU, Radio Astronomy, Computer Science, Departure Coefficients, Hypergeometric Function

Source

DOI

Publisher's version

Rights statement

Permanent link

https://hdl.handle.net/10292/5605

Collections

Masters Theses