Mike Kirby - Research

Research

Computational Mathematics - Nektar++

The goal of this effort is the development (theoretical and algorithmic) and implementation of high-order finite element methods.

Mapping Scientific Computing Algorithms to GPUs

The goal of this effort is the development (theoretical and algorithmic) and implementation of different scientific computing algorithms to streaming multiprocessors. We have examined the fast iterative method (Eikonal solver) in 2D and 3D and the finite element method in 2D and 3D on NVidia Hardware. Our FEM work involves a focussed effort on maximal independent sets (MIS) and mesh-informed algebraic multigrid (AMG) on GPUs.

Uncertainty Quantification in Biomedical Problems

The goal of this effort is to employ computational uncertainty modeling techniques in biomedical problems.

Visualization Techniques for High-Order Finite Element Methods and Stochastic Finite Element Methods

The goal of this effort is to develop visualization techniques which exploit the mathematical structure of high-order and stochastic finite element methods. Three specific efforts are: (1) the development and implementation of ElVis -- our high-order finite element visualizer; (2) our work on Smoothness-Increasing Accuracy-Conserving Filtering Methods for discontinuous Galerkin methods (DGM); and (3) computation and visualization of stochastic finite element data.
Collaborators:

Application of Formal Methods to High-Performance Computing

The goal of this effort is to apply formal methods to the debugging and optimizing MPI libraries and MPI-based high-performance computing codes.

Last Updated: 01/17/13