BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Inaugural lecture: Robust error-controlled materials simulations
DTSTART:20240313T171500
DTEND:20240313T180000
DTSTAMP:20260502T093958Z
UID:6b6ccbe4ddd5cd9db28080c33c96ecc769985747c59eb26eb1e99aed
CATEGORIES:Inaugural lectures - Honorary Lecture
DESCRIPTION:Prof. Michael Herbst\nSystematic first-principle simulations a
 re nowadays a key component when developing novel materials. Usually the r
 esulting simulation data is not directly used to drive the search\, but in
 stead employed to train a considerable cheaper statistical surrogate.  In
  this setting of potentially millions of simulations as well as these mult
 iple layers of approximations (physical\, numerical\, statistical) obtaini
 ng robust computational workflows and tracking simulation errors remains c
 hallenging.\n\nIn this talk I will report on progress along two axes of re
 search to tackle these challenges.  The first concerns the development of
  robust numerical algorithms for density-functional theory (DFT) --- the m
 ost widely employed family of first-principle models in the field. The foc
 us of the development here is to obtain black-box methods that are able to
  automatically adapt to the physics of the simulated system. Secondly\, I 
 will discuss first results on employing multi-task statistical surrogate m
 odels\, a surrogatisation technique\, which enables the use of data of het
 erogeneous quality when training a single surrogate. By combining material
 s simulation approaches of different cost/accuracy balances this not only 
 unlocks computational savings to generate training data\, but also allows 
 to opportunistically exploit heterogeneous databases of already existing s
 imulation data.\n\nIn both efforts software has played a key role to provi
 de an accessible platform fostering such interdisciplinary developments. I
 n our work we develop and extend the density-functional toolkit (DFTK)\, a
  Julia-based DFT code\, suitable to mathematical research (only 7500 lines
  of code)\, but at the same time integrated into standard tools for materi
 als discovery such as AiiDA.\n\nBio: Michael Herbst obtained a PhD in Theo
 retical Chemistry from Heidelberg University in 2018\, after which he move
 d on to two postdoctoral research stays in Applied Mathematics. From 2019 
 till 2021 he worked with Éric Cancès (École des Ponts\, Paris\, France)
  and from 2021 till 2023 he stayed in the group of Benjamin Stamm (RWTH Aa
 chen). Since March 2023 he is an assistant professor in the Institute of M
 athematics and the Institute of Materials at EPFL. His current research sp
 ans broadly in the field of materials simulations concerning numerical err
 or control and uncertainty quantification of DFT models as well as the dev
 elopment of efficient and robust algorithms for high-throughput materials 
 screening.
LOCATION:ELA 1 https://plan.epfl.ch/?room==ELA%201 https://epfl.zoom.us/j/
 62519319142
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR