Authors

Abstract

We address the calibration of a computationally expensive nuclear physicsmodel for which derivative information with respect to the fit parameters is not readilyavailable. Of particular interest is the performance of optimization-based trainingalgorithms when dozens, rather than millions or more, of training data are availableand when the expense of the model places limitations on the number of concurrentmodel evaluations that can be per​formed​.As a case study, we consider the Fayans energy density functional model, whichhas characteristics similar to many model fitting and calibration problems in nuclearphysics. We analyze hyperparameter tuning considerations and variability associatedwith stochastic optimization algorithms and illustrate considerations for tuning indifferent computational settings.