Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

Yes, I did! But currently I don’t.

This section includes some material that I have delivered in the past, particularly in the COMP6032: High Performance Computing in Science and Engineering course as part of Manchester’s Advanced Computer Science MSc programme. It should be noted that this module is no longer available and the material it consists of is not part of the current curriculum. It is presented here in the hope that it proves useful to students on similar numerical methods courses.

COMP6032

The labs for this course demonstrated a Successive Over-Relaxation variant of the Gauss-Seidel method for solving a linear system representing Laplace’s equation.

Before implementing and evaluating alternate approaches for the parallelisation of this routine, students were encouraged to develop a firm understanding of the principles of the code. The following material aims to assist in this.

What Happens

The SORgrid program presents a visualisation of the convergence process of Successive Over-Relaxation on a finite-difference discretisation of Laplace’s equation. Note that it is a steady-state equation and does not vary with respect to time - the changes in element values represent the gradual convergence towards the desired tolerance.

Why it Happens

One of the most common questions about this routine was why the series of operation in the code should result in something resembling a solution of the given equation. The brief tutorial “Why the Code Solves the PDE” was written to provide an introduction to discretisation schemes.