Probabilistic Numerical Methods
About
Probabilistic numerical methods are a set of tools to solve numerical analysis problems from the point of view of statistical inference. Our work mostly focuses on Bayesian numerical methods, which are motivated by their uncertainty quantification properties for the output of numerical methods. The group has significantly advanced the field in research years, with work on the foundations of probabilistic numerics, on methodology for quadrature, ODE solvers and PDE solvers, and on applications to complex engineering problems.
An overview of the field can be found on the following website: http://www.probabilistic-numerics.org/
News
- Mark Girolami, Jon Cockayne, Alessandro Barp and Francois-Xavier Briol are currently involved in the 2016-2017 SAMSI working group on probabilistic numerics. This includes extensive collaborations with other members of the working group, the organisation of a regular reading group and a research visit to Caltech by Alessandro Barp and Francois-Xavier Briol (hosted by Prof. Andrew Stuart and Prof. Houman Owhadi).
Collaborators
- Chris J. Oates (Newcastle University)
- Tim Sullivan (Free University Berlin)
- Philipp Hennig (MPI Tuebingen)
- Michael A. Osborne (University of Oxford)
- Dino Sejdinovic (University of Oxford)
- Andrew Stuart (California Institute of Technology)
- Houman Owhadi (California Institute of Technology)