A symposium at the DPG Spring meeting, Regensburg, 4-11 September 2022
Abstract submissions for contributed talks and posters, on the general theme of “Frontiers of Electronic Structure Theory” as well as on the focus topic, are open until June 1st 2022
https://www.dpg-tagung.de/r22/submission.html?language=en
NB: Select “SYES: Symposium Frontiers of Electronic-Structure Theory” as the part of the conference you are applying to, and make sure to book accommodation early, Regensburg gets fully booked for the DPG meeting.
Invited Speakers:
Eun-Ah Kim (Cornell, USA)
Kieron Burke (UC Irvine, USA)
Cecilia Clementi (FU, Germany)
Volker Deringer (Oxford, UK)
Jörg Behler (Göttingen, Germany)
Machine learning methods have gained a prominent spot in the research of materials and molecules, especially in the context of the atomic-scale modeling of their properties. The growing understanding of how machine-learning methods should be adapted to the specific requirements of the field is making them progressively more effective and easy to use. Machine-learning techniques use the predictions of electronic-structure theory to train surrogate models that can compute the same properties with similar accuracy at much reduced cost. The combination of physics-based and data-driven paradigms is extending dramatically the reach of electronic-structure theory, as its predictive accuracy can now be applied to more complex, larger-scale problems and longer timescales. The field has been evolving so fast that in the past years we have witnessed considerable breakthroughs enabled by this combination: first-principles accuracy assessment of finite-temperature thermodynamics, including also subtle effects such as quantum nuclear fluctuations, has become commonplace; predictions of microscopic quantities beyond the interatomic potential energy are making it possible to incorporate functional properties into a fully-predictive machine-learning framework; inverse design and generative models are simplifying the search of configurational and composition spaces for compounds with optimal performance; including information and training data calculated from methods that go beyond density functional theory allows to make predictions systematically improvable.
This symposium will cover recent progress in the broad area of artificial intelligence applied to real materials, with invited talks representing several different fronts of research in the areas that have particularly profited from these techniques. The event will provide a forum to report on the most recent developments and will feature talks by authoritative experts in the field as well as emerging leaders in this area. We anticipate that the symposium will attract a broad and diverse audience, ranging from researchers working on methods as well as applications to advanced materials, including interdisciplinary approaches linking chemistry, physics, material sciences and biology.
Organizers:
Michele Ceriotti (EPFL, Lausanne)
Georg Kresse (University of Vienna)
Mariana Rossi (MPI for the Structure and Dynamics of Matter, Hamburg)