Clément Chevalier - Workshop on GRF simulation, Bern, 2014

Context

Gaussian Random Field (GRF) simulation plays a major role in applied probability and statistics, with a wide range of application domains. Various Monte-Carlo estimators and Uncertainty Quantification methods rely of GRF simulation. However, a well-known drawback of standard GRF simulation methods, which often thwarts their intensive use in application, is the high computational complexity and memory usage involved when the number of simulation locations is large or infinite.

A number of methods have been proposed to ease GRF simulation in various frameworks. Some methods allow simulating GRFs on a large number of points provided that these points lie on a regular grid (Circulant Embedding). Other methods aim at finding the simulation locations in a smart way in order to solve the problem at hand without having to simulate at thousands of locations. Spectral methods provide an elegant – but usually challenging in application – setup to simulate in infinitely many locations. Some recent approaches enable to incorporate expert knowledge on the problem at hand by taking constraints on the GRF sample paths into account. Finally algorithmic solutions have been recently proposed for optimally chosing simulation points or significantly reducing computational costs when observations are assimilated sequentially.

The GRF-sim workshop

It aims at summing up state-of-the-art methods that have been developed in various problems involving GRF simulation and to present new approaches and applications both from various academic domains (probability and statistics, geosciences, etc.) and inspired by real-world problems (climate, engineering, etc.). The goal is to bring together researchers working actively on different aspects of these problems and discuss the ongoing research and challenges.

Program

The GRF-sim workshop will take place across 3 days, from Monday 24th November 2014 to Wednesday 26th November 2014.

It will consist of four half days of presentations, starting on monday afternoon and finishing Wednesday noon. The complete program can be found here.

Three keynote lectures are scheduled, given by the following speakers:

Prof. Christian Lantuéjoul (Mines-ParisTech, Fontainebleau, France)
Prof. Finn Lindgren (University of Bath, UK)
Prof. Martin Schlather (University of Mannheim, Germany)

In addition, 15 presentations will be given by speakers invited to the workshop.

The topic of the workshop includes general methods for simulating Gaussian and related Random Fields (spectral and algebraic methods, conditional or non-conditional simulations, etc).

Participants

The following people have confirmed their participation to the workshop.

Denis Allard (INRA, Avignon, France) abstract presentation

Alina Astrakova (Univ. Bergen, Norway) abstract presentation

Dario Azzimonti (Univ. Bern, Switzerland) abstract presentation

Felix Ballani (TU - Freiberg, Germany) abstract

Mickaël Binois (Ecole des Mines, Saint-Etienne, France) abstract presentation

Clément Chevalier (Univ. Bern, Switerland) abstract presentation

Reinhard Furrer (Univ. Zurich, Switzerland) abstract presentation R examples

Christian Lantuéjoul (Mines-ParisTech, Fontainebleau, France) abstract presentation

Finn Lindgren (Univ. Bath, UK) abstract presentation

Hassan Maatouk (Ecole des Mines, Saint-Etienne, France) abstract presentation