Computational Methods for Partial Differential Equations

The Main page for this module is on Blackboard which is not uniformaly accessible; this page contains a brief summary which should be more permanently available to all.

Most of the world is governed by partial differential equations. Nowadays it is very important to be able to solve them numerically. This module discuuses several important techniques for doing this using Finite Difference Methods.

The module has no prerequisites and is assessed solely by projects. Model codes will be supplied, illustrating suitable algorithms for simple problems, which then require modification for application in the projects. The idea is to build up towards an impressive final project of Research level. Thus in 2016-2017, various codes were supplied for solving problems in two dimensions. The projects required solving similar problems in the annular region between two circles. This culminated in the previously unsolved problem of Rayleigh-Benard convection in an annulus. For details see Project 3 below.

Explicitly the projects were:

2016-7 projects

  • Project 1 (10%)
  • Project 2 (40%)
  • Project 3 (50%)

    2017-8 projects

  • Project 1 (20%)
  • Project 2 (40%)
  • Project 3 (40%)

    For those on the MSci or MSc codings, there was a final (Mastery) project, to produce an A1 poster of the full problem. Here is one such creditable poster for each year:

  • Poster for Project 4 2017
  • Poster for Project 4 2018

    In 2017, students were supplied with codes n Matlab which solve the 2-D Convection problem, but could use anyi sensible language in their programs. The picture below illustrates the time-evolution of conducting fluid heated below and on its side walls. The velocity vector is on the left, while the temperature contours are on the right (red being hot). Note how the system begins by setting up 5 convective rolls, but then changes its mind and settles on 4.