Five post-doctoral research assistants: National Centre for
Statistical Ecology
_________________________________________
N A T I O N A L C E N T R E F O R
S T A T I S T I C A L E C O L O G Y
_________________________________________
The National Centre for Statistical Ecology (NCSE) is a new joint venture
between the Universities of Cambridge, Kent and St Andrews, funded under
the EPSRC multidisciplinary critical mass in Mathematics initiative. It
links the research groups in statistical ecology at the three
Universities.
We are currently seeking five post-doctoral research assistants to
work in statistical ecology. As well as conducting new research, the
post-docs will take part in the conference and workshop activities
of the Centre. They will work with a large group of research
students and will spend time at each of the three university groups.
The post-docs will liaise with ecologists, and collaborate with
visitors to the Centre. Previous experience of working in the areas
of applied statistics, computational statistics or ecological
statistics would be desirable, although not essential. All five
positions are available from the 1st October, 2005, but a later
start is possible if necessary. All positions are of three-years
duration in the first instance.
All candidates should have (or expect shortly to receive) a PhD in
statistics, mathematical modelling or other related field. The
ability to work as part of a team and to communicate with ecologists
is essential to all posts.
Stipends for all posts will be on the RA1A scale (19460 - 29127
pounds per annum). General enquiries should be addressed to the
Centre Director, Prof. Byron Morgan ([log in to unmask]). To
discuss individual projects please contact the relevant supervisor
(see below).
Applications should be made by completing the application
form available for download from: http://www.ncse.org.uk/. The
closing date for applications is 10th June, 2005.
OUTLINE DESCRIPTIONS OF THE PROJECTS
Modelling via MCMC- Automated Procedures
Post: Statistician
Location: Cambridge
Supervisor: Steve Brooks
Recent work at all three groups within the Centre has shown how
varying levels of complexity can be incorporated into many common
models. The work in Cambridge will focus primarily on problems of
model fitting and model discrimination and will complement current
and ongoing work within the Centre on model construction. Novel
computational methods will be developed in order to exploit fully
the complex modelling structures, and research will focus on the
development and application of MCMC-based methodology, including in
particular reversible-jump procedures.
Parameter Redundancy and State-Space Modelling using Kalman Filters
Post: Statistician
Location: Canterbury, Kent
Supervisor: Byron Morgan
Many ecological studies are now resulting in extensive data sets for
modelling and analysis. A relatively recent development involves
integrating data sets from a variety of different sources by fitting
population dynamics models; a useful tool has been the application
of Kalman Filter procedures for constructing component likelihoods
for time-series data. Quite elaborate models may be devised for
integrated data sets. A necessary check on the development of
complex models arises from the need to demonstrate that proposed
models can realistically be fitted to data, and do not contain
parameters that are poorly identified. This is a problem common to
both classical and Bayesian paradigms. Research at Kent has
established a number of useful computational tools for fitting
complex models to large and related data sets, and for checking for
parameter redundancy. The post-doc working on this project will join
a research team developing these tools.
Automated Model Fitting using Sequential Importance Sampling
Post: Statistician
Location: St Andrews
Supervisor: Steve Buckland
Continuing research at St Andrews involves the formulation and
fitting of complex state-space models. Typically, realistic models
might incorporate density dependence, movement between different
sites of a metapopulation, or species interactions in the case of
multi-species systems. The resulting Leslie or Lefkovitch matrices
can be correspondingly complicated, and much is gained through the
decomposition of these matrices into simpler component processes.
This general approach may be extended, for instance by adding
processes such as sex or genotype assignment. The St Andrews
post-doc will develop methods of analysing time series of survey
data, with an embedded population dynamics model, using hidden
process models. Models will be checked for parameter-redundancy, and
fitted to data using sequential importance sampling (SIS) methods,
with which the group already has extensive experience.
Computational Statistical Ecology
Post: Statistical Programmer
Location: Cambridge
Supervisor: Steve Brooks
A primary aim of the Centre's research programme is to produce
software that enables users to implement the new methodology
developed in the Centre. The work of this post-doc will dovetail
with the work of all of the other members of the Centre. This
post-doc will need to combine excellent computational skills with
the ability to work in a range of different areas of statistics and
at several interfaces with ecology. The post-doc will facilitate the
implementation of the other researchers' work and it is therefore
inevitable that the post-doc will play an important role in
methodological development across the board. The end-product will be
a suite of interlinked research-based modules exploiting both
Bayesian and classical methodology and sharing a single use
interface and common instruction manual. The post-doc will be
responsible for developing this software suite which is intended to
be platform-independent, open-source and freely available.
Modelling Dispersal and Life-History Strategy of Linyphiid Spiders
Post: Statistical/Mathematical Modeller
Location: Cambridge
Supervisor: Steve Brooks and George Thomas, Plymouth
This post forms part of a large BBSRC-funded project with
collaborating groups at the Universities of Cambridge, East Anglia
and Plymouth. Using linyphiid spiders as a model of wind-dispersed
invertebrates in fragmented agricultural landscapes subject to
catastrophic disturbances from farming operations, the project aims
to produce a model that will predict the effects of agriculture on
spider population abundance and diversity in order to aid the
development of sustainable agricultural production. The postholder
will develop a two-dimensional dispersal model incorporating
meteorological data. The model will be integrated with a
spatially explicit population dynamics model and Bayesian methods
will be used for both parameter estimation and prediction.
More details of NCSE and of the projects are given at
http://www.ncse.org.uk/
You may also contact the Centre Director Byron Morgan
([log in to unmask]) or Co-directors Steve Brooks
([log in to unmask]) and Steve Buckland
([log in to unmask]) directly.
|