Development of the uncertainty quantification toolkit's python interface and surrogate construction tutorial

Technical Report · Mon Aug 01 04:00:00 EDT 2022

DOI:https://doi.org/10.2172/1884927· OSTI ID:1884927

Curry, Caitlin ^[1]

---

1. Sandia National Laboratories (SNL), Albuquerque, NM, and Livermore, CA (United States)

The uncertainty quantification toolkit (UQTk) is a collection of c++ libraries that assess the confidence of numerical models. Surrogate approximations, often polynomial chaos expansions (PCEs), lessen the computational cost of these assessments. I developed a Python interface in UQTk for regression and Bayesian compressive sensing to add to the existing Galerkin projection method. These methods receive an object containing the polynomial basis information and NumPy arrays of sample points, call c++ methods, and return the PCE coefficients in a NumPy array. To demonstrate these methods, I wrote a tutorial in which I use them to construct surrogates for Genz functions and calculate the resulting error.

• Research Organization: Sandia National Laboratories (SNL-NM), Albuquerque, NM (United States)
• Sponsoring Organization: USDOE National Nuclear Security Administration (NNSA)
• DOE Contract Number: NA0003525
• OSTI ID: 1884927
• Report Number(s): SAND2022-11323R; 709384
• Country of Publication: United States
• Language: English

Related Subjects

97 MATHEMATICS AND COMPUTING