Hybrid data assimilation methods for nuclear-data-induced uncertainties

Le résumé fourni par la source

This study proposes new hybrid data assimilation (DA) methods to effectively use experimental data represented in two different ways in the DA process for reducing nuclear-data-induced uncertainties. Conventional DA methods often assume a linear model, where data in a DA database are represented by their differential coefficients, i.e. nuclear data sensitivity coefficients. In contrast, more rigorous and versatile DA methods based on sampling techniques have been proposed recently. These sampling-based DA methods describe data for DA using results derived from direct transport calculations performed with perturbed nuclear data samples. The proposed hybrid DA methods simultaneously use both types of data described by sensitivity coefficients and sampling results and thus are called ‘hybrid.’ Two hybrid DA methods are proposed herein: the simple hybrid DA and efficient hybrid DA methods. Among these, the simple method is based on a straightforward concept that uses a random sampling technique. In contrast, the efficient method employs a more technical method that effectively combines a new deterministic sampling technique with sensitivity coefficients to reduce statistical uncertainties associated with the simple method. The efficient method is expected to be a candidate to achieve high precision and rigorous DA with a realistic sample size.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.