Accelerating Time-integrators with Surrogate Models

Executive Summary

We will develop and implement new techniques to incorporate machine learning and other surrogate models into classical time integration algorithms. Such methods can accelerate the time evolution of computational simulations in climate, energy, materials science and other important missions.

Publications, Presentations, and Patents

Steven Roberts, Daniel Martin, Hans Johansen, David Gardner, and Carol Woodward,“Improved Time-Stepping for the BISICLES Ice-Sheet Model" (Presentation, CASC Division Meeting, Livermore, CA, July 13, 2023).

Steven Roberts, Daniel Martin, Hans Johansen, David Gardner, Carol Woodward, Benjamin Seibold, David Ketcheson, Abhijit Biswas, and David Shirokoff. “Overcoming First Order” (Presentation, Los Alamos Workshop on Time Integration for Multiphysics, Los Alamos, NM, August 10, 2023).

Steven Roberts, Daniel Martin, Hans Johansen, David Gardner, and Carol Woodward, “Improved Time-Stepping for the BISICLES Ice-Sheet Model” (Presentation, FASTMath Seminar, Virtual, August 16, 2023).