Michael Herbst: Algorithmic differentiation (AD) for plane-wave DFT
Автор: MICDE University of Michigan
Загружено: 2025-10-03
Просмотров: 52
Michigan Institute for Computational Discovery and Engineering - Materials Science and Engineering joint seminar.
Prof. Michael Herbst is jointly affiliated with the Mathematics and the Materials Science and Engineering departments at EPFL. He has a unique perspective on computational materials science that spans from fundamental mathematical principles to high-throughput applications and algorithm development with modern programming languages. His expertise also extends to data science and error propagation, and appeals to a broad interdisciplinary audience across science and engineering.
At this seminar, he will talk about recent advances in applying algorithmic differentiation (AD) to plane-wave DFT using the Density-Functional Toolkit (DFTK). Topics include efficient derivative computation in metallic systems, challenges unique to AD in DFT, and applications to inverse design, uncertainty quantification, and error estimation.
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
