Stochastic vs. sensitivity-based integral parameter and nuclear data adjustments

Developments in data assimilation theory allow to adjust integral parameters and cross sections with stochastic sampling. This work investigates how two stochastic methods, MOCABA and BMC, perform relative to a sensitivity-based methodology called GLLS. Stochastic data assimilation can treat integral parameters that behave non-linearly with respect to nuclear data perturbations, which would be an advantage over GLLS. Additionally, BMC is compatible with integral parameters and nuclear data that have non-Gaussian distributions. In this work, MOCABA and BMC are compared to GLLS for a simple test case: JEZEBEL-Pu239 simulated with Serpent2. The three methods show good agreement between the mean values and uncertainties of their posterior calculated values and nuclear data. The observed discrepancies are not statistically significant with a sample size of 10000. BMC posterior calculated values and nuclear data have larger uncertainties than MOCABA's at equivalent sample sizes.