7–10 Nov 2022
Europe/Berlin timezone

The NOMAD (Novel Materials Discovery) Center of Excellence (CoE): Exploring Exascale Computing and Extreme-Scale Data

Not scheduled
20m

Speaker

Matthias Scheffler

Description

Exascale computing will have a profound impact on everyday life in the coming decades. At $10^{18}$ operations per second, exascale supercomputers will be able to quickly analyze massive volumes of data and more realistically simulate complex processes. The goal of the NOMAD Center of Excellence CoE https://www.nomad-coe.eu/ is to bring computational materials science to the next level of supercomputing.

The NOMAD CoE assesses and exploits the characteristics of extreme-scale data and exascale computing for computational materials science, to enable investigations of systems of higher complexity (space and time), consideration of metastable states and temperature, and all this at significantly higher accuracy and precision than what is possible today. Systematic studies and predictions of novel materials to solve urgent energy, environmental, and societal challenges require such significant methodological advancements targeting the upcoming exascale computers. Key NOMAD examples are catalytic water splitting for hydrogen production and the transformation of waste heat into useful electricity.

The infrastructure developed with the NOMAD CoE builds on three pillars: “Codes” extends and develops ab initio computational materials science for entire code families to be able to exploit exascale for attacking new problem categories that are not feasible on today’s top supercomputers. “Workflows” develops tools to manage high-throughput computations that take full advantage of exascale resources. “Big-Data Analytics” advances the existing big-data tools and brings them towards near-real-time performance. Here, we present the first achievements of the NOMAD CoE: upgrade of the ELPA eigensolver to the exascale, development of an interface to CC4S for coupled-cluster calculations in solids, creation of a libraries of “recipes” for massive ab initio calculations, applied to creation of a large data-set of vibrational calculations, and deployment of a massive parallel framework for symbolic regression calculations.

Addresses
(a) Physics Department and IRIS Adlershof, Humboldt-Universität zu Berlin, Berlin, Germany
(b) For the affiliations of the PIs of the NOMAD CoE see: https://www.nomad-coe.eu

Abstract Number (department-wise) SG 02
Department Scheffler Group

Primary authors

Presentation materials

There are no materials yet.