Automated few-group cross-section parameterization based on quasi-regression

This work addresses the quasi-regression technique for parameterization of few-group neutron cross-sections. The objective of this work is to create a consistent and flexible methodology of parameterization, allowing the creation of an automated tool for such a parameterization. The quasi-regression methodology discussed in this work includes the following essential parts: approximation of a smooth function of an arbitrary number of dimensions; built-in approximation error characterization; capability of analysis for the obtained dependencies; and library optimization including error-controlled reduction of the number of approximation terms and number of state parameters.