Stay Informed:
Baskin Engineering COVID-19 Information and Resources
Campus Roadmap to Recovery
Zoom Links: Zoom Help | Teaching with Zoom | Zoom Quick Guide

UCSC-SOE-16-02: Bayesian semiparametric methods for emulation and calibration of stochastic computer simulators

Marian Farah, Athanasios Kottas and Robin Morris
01/14/2016 08:47 PM
Applied Mathematics & Statistics
Stochastic computer simulators are increasingly used in technology and science to model random systems, e.g., in population dynamics, biological processes, and nuclear interactions. We propose a framework for emulation and calibration of complex stochastic simulators. The emulator is built from a general nonparametric mixture model for the joint distribution of the simulator inputs and output(s). For calibration, we develop a semiparametric method to link the simulator data with field observed data using a modular, two-stage approach. In the first stage, the posterior distribution of the emulator parameters is estimated based solely on the simulator data. Then, in the second stage, the posterior distribution of the calibration parameters is approximated based on the field data as well as the posterior distribution of the emulator parameters obtained in the first stage. The methodology is applied to PROPSET, a stochastic simulator that models the bombardment of a device with high-energy protons in order to study the effect of radiation on spaceborne microelectronics devices.

This report is not available for download at this time.