Talks and presentations

Dissertation Defense for PhD in Mathematics

April 03, 2024

Talk, University of Tennessee, Knoxville, Knoxville, TN, USA

These are the slides I presented during the defense of my doctoral research and dissertation. My research involved extending a method that my advisor and his collaborators had introduces for numerically solving the spectral fractional Laplacian. There were two major accomplishments that I contributed. The first was to avoid the problem of numerically computing the eigenvalues of an ODE needed by the method with analytically solving the ODE. This overcomes the stability issue in finding the eigenvalues and eigenfunctions. The second was showing that the method with analytic eigenvalues and eigenfunctions was essentially a quadrature for a Balakrishnan integral formulation of the spectral fractional Laplacian. This discovery was made while demonstrating the convergence rate of the algorithm. I implemented the method in C++ using the deal.ii FEM library and was able to match the theoretical convergence order as well as demonstrate the parallel efficiency of the algorithm.

A Computational Mathematician’s Guide to High Performance Computing

February 18, 2015

Talk, University of Tennessee, Knoxville, Knoxville, TN, USA

These are the slides I presented during a Computational and Applied Mathematics (CAM) seminar for the mathematics department at the University of Tennessee, Knoxville. I provided a brief overview of some relevant topics from a computing side such as computer architecture, mathematics libraries, parallel computing libraries, and code optimization tools.

Thesis Defense for MS in Engineering

May 29, 2012

Talk, UT Simcenter, University of Tennessee at Chattanooga, Chattanooga, TN, USA

These are the slides I presented during the defense of my Masters research and thesis. For this, I analyzed an existing high-order reconstruction algorithm for the finite volume method on unstructured meshes and then implemented it code. I verified the convergence orders and made performance measurements relative to existing methods. The impetus of this project was to study the suitability of the specific reconstruction technique for integration in the department’s flow solver codebase. Of particular interest was whether or not this method could resolve vortical structures while remaining computational tractable.