BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:"Machine learning in chemistry and beyond" (ChE-650) seminar by Al
 exandre Tkatchenko: On Electrons and Machine Learning Force Fields
DTSTART:20211116T151500
DTEND:20211116T161500
DTSTAMP:20260510T054623Z
UID:a748fee88d8ecd9ba31aeb5679180aee94d452f7f485a2afbb1409a5
CATEGORIES:Conferences - Seminars
DESCRIPTION:Alexandre Tkatchenko is a professor at the Department of Physi
 cs and Materials Science (and head of this department since January 2020) 
 at the University of Luxembourg\, where he holds a chair in Theoretical Ch
 emical Physics. Tkatchenko also holds a distinguished visiting professor p
 osition at Technical University of Berlin. His group develops accurate and
  efficient first-principles computational models to study a wide range of 
 complex materials\, aiming at qualitative understanding and quantitative p
 rediction of their structural\, cohesive\, electronic\, and optical proper
 ties at the atomic scale and beyond. He has delivered more than 250 invite
 d talks\, seminars and colloquia worldwide\, published 180 articles in pre
 stigious journals (h-index of 69 with more than 24\,000 citations\; Top 1%
  ISI highly cited researcher in 2018-2020)\, and serves on the editorial b
 oards of Science Advances and Physical Review Letters. Tkatchenko has rece
 ived a number of awards\, including APS Fellow from the American Physical 
 Society\, Gerhard Ertl Young Investigator Award of the German Physical Soc
 iety\, Dirac Medal from the World Association of Theoretical and Computati
 onal Chemists (WATOC)\, van der Waals prize of ICNI-2021\, and three flags
 hip grants from the European Research Council: a Starting Grant in 2011\, 
 a Consolidator Grant in 2017\, and Proof-of-Concept Grant in 2020.\nOn Ele
 ctrons and Machine Learning Force Fields\n\nMachine Learning Force Fields 
 (MLFF) should be accurate\, efficient\, and applicable to molecules\, mate
 rials\, and interfaces thereof. The first step toward ensuring broad appli
 cability and reliability of MLFFs requires a robust conceptual understandi
 ng of how to map interacting electrons to interacting "atoms". Here I disc
 uss two aspects: (1) how electronic interactions are mapped to atoms with 
 a critique of the "electronic nearsightedness" principle\, and (2) our dev
 elopments of symmetry-adapted gradient-domain machine learning (sGDML) fra
 mework for MLFFs generally applicable for modeling of molecules\, material
 s\, and their interfaces. I highlight the key importance of bridging funda
 mental physical priors and conservation laws with the flexibility of non-l
 inear ML regressors to achieve the challenging goal of constructing chemic
 ally-accurate force fields for a broad set of systems. Applications of sGD
 ML will be presented for small and large (bio/DNA) molecules\, pristine an
 d realistic solids\, and interfaces between molecules and 2D materials. \
 n\n[Refs] Sci. Adv. 3\, e1603015 (2017)\; Nat. Commun. 9\, 3887 (2018)\; C
 omp. Phys. Comm. 240\, 38 (2019)\; J. Chem. Phys. 150\, 114102 (2019)\; Sc
 i. Adv. 5\, eaax0024 (2019).
LOCATION:https://epfl.zoom.us/j/64473017589?pwd=Vmpnd1pleGhEb1hFb3kxUlNIUW
 JyQT09
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR