You are here

Software Tools for Uncertainty Quantification and Machine Learning with Applications to Computational Science

Som Dhulipala, Idaho National Laboratory
Zachary Prince, Idaho National Laboratory
Peter German, Idaho National Laboratory
Dewen Yushu, Idaho National Laboratory
Yifeng Che, Idaho National Laboratory
There has been a significant increase in the use of machine learning (ML) techniques to accelerate numerical simulations and uncertainty quantification (UQ) to quantify the trustworthiness of the numerical or ML model predictions. As a result, software tools to perform UQ and ML for modeling and simulations are gaining interest due to their potential to: (1) translate research methods in ML/UQ to applications for practical computational problems; (2) motivate new research on ML/UQ, especially for complex problems like coupled/multi-scale systems, systems with high-dimensional input-output spaces, and systems that take a significant amount of time to simulate; and (3) incorporate technologies like massively parallel computing, exascale computing, and GPU/TPU-based computing into research methods and practical applications. This mini-symposium focuses on recently developed software tools and advancements in existing packages that promote UQ/ML with applications to modeling and simulations. Of specific interest are software tools that are open source. All topics directly addressing or supporting software tools for UQ/ML are welcome in this mini-symposium. These include, but are not limited to, the architecture and usage of a software package, application of a software package to practical computational problems, future directions for software packages to better support UQ/ML of modeling and simulation tasks, etc.