This is a past event.

Danny Perez from Los Alamos National Laboratory will present at a special Physics Colloquium this week.

Perez's talk is titled "Bridging the Scales in Materials Simulations Using Exascale Computing and Machine

Learning."

The seminar will be presented in person at 11 a.m. on Friday (Sep. 13) in Fisher 125.

ABSTRACT:

The advent of exascale computers, which can carry out in excess of 10¹⁸ operations per second, has the potential to revolutionize the way we understand, predict, and optimize the properties of materials. However, the characteristics of these architectures also pose formidable challenges that require novel algorithmic solutions. I will illustrate how machine-learning approaches can address some of the challenges facing the multiscale materials modeling community by 1) bridging the gap between quantum and classical mechanics though machine-learned approximations of so-called interatomic potentials, enabling accurate atomic-scale simulations of materials at a low computational cost, and 2) by systematically orchestrating complex scale-bridging workflows to autonomously derive larger-scale models, e.g., in the form of reaction-diffusion equations, from these lower scale simulations.

BIO:

Dr. Danny Perez is a staff scientist in the Theoretical Division of Los Alamos National Laboratory, which he joined in 2007 after receiving a Ph.D. in Physics at the Université de Montréal, in Canada. His research centers on the development of novel methodologies for the simulation of materials with atomic-scale resolution, with a particular emphasis on ultra-large-scale computing. He applies these methods to a range of energy-relevant problems, most notably to the understanding of materials in extreme conditions found in fusion reactors and particle accelerators.