Student projects
The EPM group offers projects to Bachelor's and Master's students spanning experimental, observational and theoretical aspects of Earth and planetary magnetism. Some example projects and staff contact details are listed below. Other projects can be designed for students interested in particular topics.
Please contact us if you are interested or require further information.
Bachelor's projects
One of the canonical problem of fluid dynamics is the flow around an obstacle. It has been extensively studied in a wide range of parameters and is often use to benchmark experimental and numerical techniques. It is well known that when the velocity exceeds a certain value, the down stream flow becomes unstable forming the so-called Karman vortex street. What happens when the whole system is subject to a global rotation ? This simple question is fundamental to understand some of the dynamics of the lower atmosphere as well as the dynamics that may occur in the subsurface oceans of icy moons, such as Europa. In this project you will explore the simple problem of a uniform flow passing a cylinder with and without rotation to characterise the onset of the Von-Karman instability. To do so, you will use the 2D numerical code DEDALUS and the more advance code ComSol Multiphysics. In addition, you will participate to an experimental study on the rotating table of the new laboratory FlowLab@GMA.
Master's projects
A number of new algorithms have presented themselves that are able to solve inverse problems by variations on random walk or Monte Carlo approaches. Two such algorithms are in everyday use in electromagnetic induction and in seismology. We plan to study their applicability to inverse problems in general, but will begin on an inverse problem with only three dimensions, in order to compare performance.
This parameter estimation problem is concerned with retrieval of principle absorbances of a crystal species using infra-red spectroscopy, and is one of the few inverse problems that is truly finite-dimensional. Algorithms that present themselves as general parameter-space sampling methods include the Neighbourhood Algorithm of Sambridge, the Covariance Matrix Adaption Evolution Strategy (CMAES) and various flavours of MCMC including Hamiltonian Monte-Carlo.
The ``no free lunch theorem’’ suggests that there is no general algorithm that is able to solve a variety of problems. Despite that, it may be
possible to draw some conclusions regarding the applicability of various algorithms to typical geophysical inverse problems arising in seismology, electromagnetic induction and geomagnetism.
The topic will be of interest to students with a good background and interest in mathematics and probability, together with good computing skills.
Contact: Prof. Andrew Jackson, Prof. Alexander Grayver, Prof. Andreas Fichtner