Düsseldorf 2007 – scientific programme
Parts | Days | Selection | Search | Downloads | Help
P: Fachverband Plasmaphysik
P 7: Poster
P 7.25: Poster
Monday, March 19, 2007, 16:30–18:30, Poster A
Data adaptive control parameter estimation for scaling laws — Heiko Dreier1, Roland Preuss2, •Andreas Dinklage1, and Volker Dose2 — 1Max-Planck-Institut für Plasmaphysik, Teilinstitut Greifswald, Wendelsteinstraße 1, D-17491 Greifswald — 2Max-Planck-Institut für Plasmaphysik, Boltzmannstraße 2, D-85748 Garching
Bayesian experimental design quantifies the utility of data expressed by the information gain. Data adaptive exploration determines the expected utility of a single new measurement using existing data and a data descriptive model. In other words, the method can be used for experimental planning.
As an example for a multivariate linear case, we apply this method for constituting scaling laws of fusion devices. In detail, the scaling of the stellarator W7-AS is examined for a subset of ι=1/3 data. The impact of the existing data on the scaling exponents is presented. Furthermore, in control parameter space regions of high utility are identified which improve the accuracy of the scaling law.
This approach is not restricted to the presented example only, but can also be extended to non-linear models.