Uncertainty quantification for equation of state using heterogeneous data sources

• Main Authors: Lindquist, Beth A., Rudin, Sven P., Rehn, Daniel A.
• Format: Conference Proceeding
• Language: English
• Subjects: Calibration; Datasets; Density functional theory; Equations of state; Extreme values; Statistical analysis; Uncertainty
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/12.0028532

Uncertainty quantification (UQ) in the context of Equation of State (EOS) generally implies the generation of ensembles of EOS models that are both consistent with calibration data and properly statistically weighted given the uncertainties on the calibration data (as well as any other sources of uncertainty included in the UQ). With respect to generating a calibration data set, high-quality experimental data with known uncertainties are ideal, but the EOS tables need to access extreme conditions where it might be difficult or even impossible to acquire experimental measurements. Density Functional Theory (DFT) calculations can give some information about materials at extreme conditions. However, heterogeneous data sets including both experimental and calculated data can be challenging to incorporate into a rigorous statistical framework for UQ analyses. In this work, we demonstrate a novel strategy for constructing a probability distribution based on DFT calculations, and we pass that distribution forward as the prior in a UQ calculation calibrated to experimental data.