Postgraduate Studentship
Structural model error in the context of an evolving (non-stationary) physical system
A NERC funded PhD studentship* is currently available at the Centre for the Analysis of Time Series (Department of Statistics) at the London School of Economics. The thesis topic will focus on structural model error in the context of an evolving (non-stationary) physical system, and will be supervised by Professor Leonard Smith and Dr David Stainforth.
Model inadequacy (structural model error) is arguably the most devastating source of failure in science based decision making, as it is arguably impossible to see it coming. Aspects of the system that lie outside the dynamics accessible to any member of the current model class can lead to misleading simulations of the future even when the model is tuned to achieve a high degree of consistency with observations. Current methods, including sampling (over a small set of initial conditions, parameter values, and/or existing models), enlarging the model class (as by introducing "stochastic physics"), and exploiting background knowledge of the system will be considered in a computationally tractable case, and then applied in the context of actual decision making.
Interested candidates should send a CV and a covering letter explaining your interest and expertise in this topic to Ian Marshall, [log in to unmask] These will be reviewed on 11 June 2012 and a full application via the Graduate Admissions Office will be invited as appropriate. CVs and covering letters received late will be considered until the position is filled.
Further information about the Centre for the Analysis of Time Series can be found at:
http://www2.lse.ac.uk/CATS/home.aspx
* The studentship is to start in October 2012 and will cover fees and living expenses for UK students, or fees only for students from member states of the European Union. For full eligibility details of the studentship please refer to NERC guidelines at: http://www.nerc.ac.uk/funding/available/postgrad/eligibility.asp
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|