Location: Institut Mathématiques de Bourgogne, Université de Bourgogne, France.
Duration: 12 months.
Dates: October 2010 to September 2011
Net Salary: around 2000 euros per month
Contact: Hervé Cardot, Peggy Cénac, Pierre-André Zitt

The aim of this post-doctoral position is to develop stochastic algorithms in the context of functional data analysis, more precisely the prediction of continuous time processes on a time interval.

Linear processes in function spaces are well suited to these questions. Among many others, we can cite the following applications:

• climatology,
• traffic prediction,
• electricity consumption,
• pollution prediction.

Stochastic algorithms are particularly useful when data arrive sequentially. They allow to make estimations and predictions with simple and fast procedures which do not need the complete re-estimation of parameters at each step. There is an extensive literature on the properties of these algorithms in the finite dimensional setting (rates of convergence, asymptotic normality, asymptotic efficiency).

However, the behaviour of these algorithms in the infinite dimensional case has been less studied. The first task of the applicant would be to review the literature. According to the candidate's personal tastes, it will be possible to study the asymptotic properties of such procedures and/or develop more applied aspects, such as implementation of the algorithms (for example in R, Matlab or scilab).

The applicant should have a PhD thesis in applied mathematics or mathematics, with a working knowledge of probability/statistics, and have a marked taste for numerical experiments. He/She will join the research team SPAN of the IMB, where he will be supervised by Hervé Cardot, Peggy Cénac and Pierre-André Zitt.

Working language : English or French

To apply, send an e-mail to Peggy Cénac before June 15th, 2010, with the following:

• cover letter,
• a detailed CV with a list of publications,
• a copy of recent works (the PhD thesis dissertation, and/or the corresponding papers, if the PhD has not been defended yet).

Support letters can be joined, or sent separately.