The aim of nuclear data evaluation is the provision of best estimates and associated uncertainties of cross sections and related quantities in form of evaluated nuclear data (END) files. These files are indispensable requisites for burn-up and transport calculations. The reliability of such calculations is therefore to a large extent determined by the quality of the employed END files. The construction of END files requires an evaluation method to combine results of model calculations with experimental data. Contemporary methods usually make use of Bayesian statistics.
In this seminar, I will give a brief introduction to Bayesian statistics and present an overview and an in-depth discussion of existing evaluation methods. Finally, I will present a new evaluation method that in contrast to most existing methods takes into account the possibility of model deficiency. The model deficiency can be explicitly estimated and important constraints such as the conservation of sum rules of cross sections are preserved by the new method. The latter two features distinguish this evaluation method from the other methods.
