NOMAD Laboratory


Jul 31, 2018

IMPRESS 2018 - Interfacing Machine Learning and Experimental Methods for Surface Structures

The advent of machine learning methods has drastically changed the way structure determination is performed, since it facilitates the rational design of (new) experiments and the analysis of large amounts of data. The aim of the IMPRESS 2018 workshop, held 11 - 13 July 2018 at the Graz University of Technology, was to bring experimentalists and theorists together, so that both can learn and benefit from each other’s expertise. 

About 50 scientists from Asia, America, and Europe came together in Graz, making the workshop, which was sponsored by NOMAD, CECAM and the Psi-k, a great success!

Invited talks from international experts covered a wide range of topics:

  • Alexander Hinderhofer (Univ. Tübingen) opened the event with a talk on monitoring growth of thin films experimentally,
  • Chiara Panosetti (TU Munich) and Katrin Forster-Tonigold presented "traditional“ approaches to computationally find surface or interface geometries,
  • Milica Torodović (Aalto University) and Andreas Jeindl (TU Graz) introduced their respective machine learning approaches for surface structure search,
  • Peter Zeppenfeld (JKU Linz) discussed thermodynamics and the importance of transition states for the dynamics of polymorph formation,
  • Daniel Packwood (Kyoto University) introduced a Bayes‘ Learning approach to obtain the parameters necessary for kinetic Monte-Carlo simulation,
  • Roland Resel (TU Graz) presented x-ray diffraction results to follow the polymorphism of thin films experimentally,
  • Torsten Fritz (Univ. Jena), Tim Frolov (LLNW) and Antti Karttunen (Aalto University) focused on interfaces and boundaries,
  • Julian Pilz (TU Graz) presented experimental challenges in substrate and parameter control,
  • Benedikt Hartl (TU Vienna) provided an overview over the problems at interfaces under electrochemical control,
  • Daniel Wegner (The Netherlands) and Takanori Fukushima (Tokio Institute of Technology) highlighted the challenge of structure-to-structure relationships from experimental viewpoints, while Talat Rahman (University of Florida) provided the computational viewpoint, and
  • Bernd Meyer (FAU Erlangen-Nürnberg) spoke about recent advances in methodology to predict the structure at solid/liquid interfaces.


Opening session, Alexander Hinderhofer

Engaging poster session

Discussions bringing experimentalists and theorists together