Functional Tensor Network Approximations for Earth System Models


Sensitivity analysis and model calibration studies for large scale models are challenged by both the large computational cost and large number of parameters typically associated with these models. These challenges are exacerbated by the non-linear input-output dependencies that limit the number of reduced-order techniques that could be leveraged in these studies. In this work we focus on the E3SM land component, and we exploit its internal structure to construct low-rank functional tensor network surrogates that model the spatio-temporal dependencies for select quantities of interest. We present a set of functional representations and model construction techniques to create parsimonious approximations commensurate with the flow of information between various model components. We investigate the efficiency of this approach for uncertainty quantification studies at both regional and global scales.

Jul 29, 2021 12:00 AM — 12:00 AM
Chicago, IL

I am contributing to the following presentations at this conference:

  • Kelli McCoy, Cosmin Safta, Roger Ghanem, Manifold-Based Optimization for Constrained Trajectoriess
  • Aniket Jivani, Xun Huan, Cosmin Safta, Beckett Y. Zhou, Nicolas R. Gauger, Uncertainty Quantification for Random Field Quantities Using Multifidelity Karhunen-Loeve Expansions
  • Arun Hegde, Elan Weiss, Wolfgang Windl, Habib Najm, Cosmin Safta, Bayesian Calibration of Interatomic Potential Models for Binary Alloys
Cosmin Safta
Cosmin Safta
Distinguished Member of Technical Staff

My research interests include uncertainty quantification, machine learning, and statistics.