The quality of evaluated nuclear data and its covariances is affected by the choice of the evaluation algorithm. The evaluator can choose to evaluate in the observable domain or the parameter domain and choose to use a Monte Carlo- or deterministic techniques[1]. The evaluator can also choose to model potential model-defects using, e.g., Gaussian Processes [2]. In this contribution, the performance of different evaluation techniques is investigated by using synthetic data. Different options for how to model the model-defects are also discussed.
In addition, the example of a new Ni-59 is presented where different co-variance driven evaluation techniques are combined to create a final file for JEFF-3.3 [3].
Keywords: Total Monte Carlo, Nuclear data evaluation
AMS subject classifications. 62P35; 81V35; 62-07;
References
[1] P.Helgesson, D.Neudecker, H.Sjöstrand, M.Grosskopf, D.Smith, R.Capote; Assessment of Novel Techniques for Nuclear Data Evaluation, 16th International Symposium of Reactor Dosimetry (ISRD16) (2017)
[2] G. Schnabel, Large Scale Bayesian Nuclear Data Evaluation with Consistent Model Defects, Ph.D. thesis, Technishe Universitätt Wien (2015)
[3] P.Helgesson, H.Sjöstrand, D.Rochman; Uncertainty driven nuclear data evaluation including thermal (n,alpha): applied to Ni-59; Nuclear Data Sheets 145 (2017) 1–24
2017.
The Fourth DAE-BRNS Theme Meeting on Generation and use of Covariance Matrices in the Applications of Nuclear Data