ENBIS-8 in Athens

21 – 25 September 2008 Abstract submission: 14 March – 11 August 2008

A new look at Kriging for the Approximation of Noisy Simulators with Tunable Fidelity

22 September 2008, 11:15 – 11:35


Submitted by
David Ginsbourger
David Ginsbourger, Victor Picheny, Olivier Roustant, Yann Richet
Ecole des Mines de Saint-Etienne et Institut de Radioprotection et de Surete Nucleaire
Kriging is intensively used to interpolate costly deterministic computer experiments. Since many recent advances in simulation rely on probabilistic methods (Monte-Carlo, etc...), it becomes necessary to take their randomness into account. Here we present an adaptation of the geostatistical "observational noise" to the frame of computer experiments with tunable probabilistic noise. We propose a statistical model for such simulators, give a derivation of the associated Kriging equations, and discuss the estimation of the covariance parameters. Perspectives include the use of Kriging for noisy-simulator-based uncertainty propagation or optimization strategies. We finally rise some new challenges about optimal design of experiments for Kriging, in the input-allocation space.

Return to programme