Sandia National Labs

Uncertainty Quantification Toolkit (UQTk) version 3.1.4

Bert Debusschere, Caitlin Curry, Cosmin Safta, Katherine Johnston, Kenny Chowdhary, Khachik Sargsyan, Luke Boll, Mohammad Khalil, Prashant Rai, Tiernan Casey, Xiaoshu Zeng

Overview

The UQ Toolkit (UQTk) is a collection of libraries and tools for the quantification of uncertainty in numerical model predictions. Version 3.1.4 offers Polynomial Chaos Expansions to represent random variables, intrusive and non-intrusive methods for propagating uncertainties through computational models, tools for sensitivity analysis, methods for sparse surrogate construction, and Bayesian inference tools for inferring parameters and model uncertainties from experimental data.

Documentation

For documentation on how to install and use UQTk, please refer to the manual, which is included as a PDF file in the directory docs/UQTk_manual.pdf. For more detailed documentation on the actual source code for development purposes, please refer to the C++ doxygen documentation. If you are familiar with UQTk and would like just a high level overview of where to find everything and how to install, see the sections on Directory Structure and Installation below.

Directory Structure

On a high level UQTk is organized as follows:

• config: Example CMake configuration scripts
• cpp/lib: Core C++ libraries
• cpp/app: Standalone apps that make UQTk functionality available to the command line
• cpp/tests: CMake Unit Tests
• dep: Third party libraries that UQTk depends on
• examples: short tutorial style examples that illustrate key UQTk capabilities
• PyUQTk: Python wrappers for the core C++ libraries as well as additional Python tools

In many key directories, README files have been included to further lay out the contents of their subdirectories.

Capabilities

Below is a list of key UQTk capabilities, along with examples that illustrate those capabilities:

• Intrusive Forward UQ: examples/ops (C++), examples/surf_rxn/SurfRxnISP.cpp (C++)
• Non-Intrusive Forward UQ: examples/surf_rxn/SurfRxnNISP.cpp (C++), examples/fwd_prop (Python), examples/window (Python), examples/uqpc (Command Line/Python)
• Non-Intrusive Surrogate Construction: examples/uqpc (Command Line/Python)
• Bayesian Compressive Sensing (BCS): examples/pce_bcs (C++)
• Global Sensitivity Analysis: examples/uqpc (Command Line/Python), examples/pce_bcs (C++), examples/sensMC (Command Line)
• Bayesian Inference: examples/line_infer (C++), examples/iuq (Command Line/Python), examples/polynomial (Python)
• Bayesian model selection: examples/polynomial (Python)
• Transitional Markov chain Monte Carlo (TMCMC): examples/tmcmc_bimodal (C++/Python/Command Line)
• Karhunen-Loève decompositions: examples/kle_ex1 (C++)
• Data Free Inference (Inference based on summary statistics): examples/dfi (C++)
• Forward Propagation with Basis Adaptation: examples/d_spring_series (Python)
• Numerical Integration (Quadrature): examples/num_integ (Python)

For more details on these capabilities, please refer to the UQTk manual in PDF format.

Installation

To install UQTk, first create a build directory outside of the UQTk repository. From within the build directory, configure the distribution via CMake. See example CMake configuration scripts in the directory config Then build via make, and test with ctest. Install with make install For example:

% mkdir build
% cd build
% ../UQTk/config/config-gcc-Python.sh
% make -j 8
% ctest
% make install

For more details, please refer to the UQTk manual in PDF format.

How to Cite

To cite UQTk, please use the following publications:

@ARTICLE{DebusscherePCE:2004,
  author   =  {B.J. Debusschere and H.N. Najm and P.P. P\'ebay and O.M. Knio
               and R.G. Ghanem and O.P. {Le Ma{\^\i}tre}},
  title    =  {Numerical challenges in the use of polynomial chaos representations
               for stochastic processes},
  journal  =  {{SIAM} Journal on Scientific Computing},
  year     =  {2004},
  volume   =  {26},
  pages    =  {698-719},
  number   =  {2},
  url      =  {http://dx.doi.org/10.1137/S1064827503427741}
}

@InCollection{DebusschereUQTk:2017,
  author    = {B. Debusschere and K. Sargsyan and C. Safta and K. Chowdhary},
  title     = {The Uncertainty Quantification Toolkit (UQTk)},
  booktitle = {Handbook of Uncertainty Quantification},
  editor    = {R. Ghanem and D. Higdon and H. Owhadi},
  year      = {2017},
  pages     = {1807--1827},
  publisher = {Springer},
  url       = {http://www.springer.com/us/book/9783319123844}
}

Contact Us

For more information, visit the UQTk website at https://www.sandia.gov/UQToolkit/ or contact the UQTk developers through the github discussions site