Integrated Data Analysis in Fusion

Andreas DINKLAGE
MPI for Plasmaphysics, Greifswald

Data analysis in fusion poses challenges for individual
diagnostics like tackling outliers, the treatment of  systematic and
different statistical errors. The actual inference process, moreover,
requires the combination of these heterogeneous results and the link to
physical models. This process is called integrated data analysis.

Intergrated data analysis is a concept to link different results and the
concept employs Bayesian probablity theory. A brief overview on methods
and a revision of selected results will be presented. The impact on
software architecture allowing for integrated analyses will be discussed.