Evaluation of sampling methods for nuclear data uncertainty quantification

In recent years, various tools for nuclear data sampling have been developed to quantify the uncertainty of reactor modeling and simulation (M&S) results. Since M&S consumes numerous computing resources, some advanced sampling methods have been proposed to reduce the sample size. Against this backdrop, this paper extends the concept of FOM to the field of uncertainty analysis to evaluate the quality of the sampled nuclear data. And the rationality is validated based on the numerical results of various sampling methods. The statistic metric thus developed and designed works in three steps. First, four sampling methods are employed and their performance is evaluated based on conventional statistic metrics, such as mean, standard deviation, correlation coefficient, and normality. Then, the sampling error is defined as the difference between the nuclear-data samples and the population. Finally, FOM, a novel statistic metric based on the sampling error and sample size, is proposed for assessing the sampling procedures. Results of our numerical experiments show that FOM can comprehensively evaluate the efficiency of sampling methods and save computing resources by evaluating the moments of nuclear-data samples rather than the results of uncertainty quantification (UQ).