Artificial Intelligence for Materials Science
Abstract
Properties and functions of materials are typically governed by a high intricacy of various processes, making explicit atomistic or electronic-structure modeling highly challenging or even impractical. As a result, the integration of artificial intelligence tools has been a game-changer and a paradigm shift. Often, material functions are triggered by rare events that may be absent in the training data, smoothed away by regularization, or AI may predict fake events. I will discuss how this challenge can be addressed, also considering the frequently overlooked uncertainty in AI predictions. I will explore how machine learning can identify “rules” and “materials genes”, enabling active learning for systematic, efficient predictions of new or even novel materials with improved performance.
About the speaker
Matthias Scheffler is a German theoretical physicist whose research focuses on condensed matter theory, materials science, and artificial intelligence. He is particularly known for his contributions to density-functional theory and many-electron quantum mechanics and for his development of multiscale approaches. In the latter, he combines electronic-structure theory with thermodynamics and statistical mechanics, and also employs numerical methods from engineering. Matthias Scheffler studied physics at Technische Universität (TU) Berlin. He carried out his doctoral work in the field of theoretical solid-state physics at the Fritz Haber Institute of the Max Planck Society (FHI) and received his Ph.D. from the TU Berlin in 1978. He then moved to the Physikalisch-Technische Bundesanstalt in Braunschweig, where he was employed as a research associate from 1978 to 1987. From 1979 to 1980, he was also a visiting scientist at the IBM T.J. Watson Research Center, Yorktown Heights, USA. He received his habilitation in 1984 from the TU Berlin. In 1988, he was appointed as a scientific member of the Max Planck Society and the founding director of the Theory Department of the Fritz Haber Institute of the Max Planck Society in Berlin. The following year, he received an honorary professorship at the TU Berlin. This was followed by further honorary professorships at Freie Universität Berlin (2006), Humboldt-Universität zu Berlin (2016), and in Hokkaido, Japan (2016). He has also been a Distinguished Visiting Professor of Computational Materials Science and Engineering at the University of California, Santa Barbara since 2005. Since 2015, he has headed the European Center of Excellence NOMAD (Novel Materials Discovery) since 2020, the NOMAD Laboratory at the FHI, and since 2021, he has been Deputy Spokesperson of the FAIRmat project at the Humboldt-Universität zu Berlin.
Awards and honors
2001 Max Planck Research Award jointly awarded by the Alexander von Humboldt Foundation and the Max Planck Society
2003 Medard W. Welch Award of the AVS (association for science and technology of materials, interfaces and processing)2004 Max Born Medal and Prize jointly awarded by the British Institute of Physics (IOP) and the German Physical Society (DPG)
2007 Honorary doctorate from the Lund University, Sweden
2010 Rudolf Jaeckel Prize of the German Vacuum Society (DVG)
Since 1998 Fellow of the American Physical Society
Since 2002 Member of the Berlin-Brandenburg Academy of Sciences and Humanities
Since 2017 Member of the German National Academy of Sciences Leopoldina
On-site at Campus Limpertsberg, you will have the opportunity to meet the speaker over a walking lunch from noon to all registered participants (Hall Bâtiment des Sciences). The talk will be at 1 PM in BS 0.03.
- Online – through Webex – to allow you to listen to the talk in case you are traveling or cannot make it on-site at Campus Limpertsberg. Here are the details to join the talk online:
https://unilu.webex.com/unilu/j.php?MTID=m40c6b78c21fd16df034cae014761910a
Meeting number: 2791 168 0633
Meeting password: N8DaMU3qwm7
For organizational reasons, please register for lunch by 21 April 2025