The PRACE Spring school on Modeling Materials using HPC and AI/ML

UTC
Description

The spring PRACE Seasonal School on Modeling Materials using HPC and AI/ML will be organized by the Centre of Operations, Slovak Academy of Sciences in Bratislava, Slovakia. The registration is free of charge.

The workshop is organized as an online event utilizing the Zoom platform. Locally based attendants may join us in Bratislava, depending on the actual SARS-CoV-2 virus/COVID-19 pandemic situation and valid restrictions. For those who will be attending online, enjoy a short tour of Bratislava HERE.

The mission of the PRACE Seasonal School on Modelling Materials using HPC and AI/ML is to get together experts and users in the field of material science in order to pass along knowledge and experience in the underlying theoretical concepts, ways of (parallel) computer implementation of pertinent algorithms, as well as efficient use of existing material science/periodic systems codes and program packages in HPC environment.

We aim to offer an end-to-end crash course covering basic theory, HPC implementation, showcasing typical applications and efficient ways of their (parallel) runs. Lectures as well as hands-on sessions are planned for various topics of the workshop (the exact mode of presentation can be decided by the respective speakers).

In this workshop we address also the utilization of Machine Learning tools and techniques that are being progressively applied to real chemical problems, such as discovery of new chemical compounds and materials with specific properties.

Who can apply: For this workshop, you will need background in theoretical and computational chemistry or physics.

    • 13:00 13:15
      Opening 15m
      Speaker: Dr Lukáš Demovič (Slovak Academy of Sciences)
    • 13:15 13:45
      Brief introduction to HPC environment 30m
      Speaker: Dr Filip Holka (Centre of Operations, Slovak Academy of Sciences)
    • 13:45 14:00
      break 15m
    • 14:00 14:30
      onDemand: how to & accounts activation 30m
      Speaker: Dr Michal Pitoňák
    • 14:30 15:30
      Jupyter notebooks interface & ML with Python 1h
      Speaker: Miloslav Valčo
    • 08:45 09:45
      Introduction to ML: Representations 1h

      In the “Introduction to ML” lectures, we will review the workflows in classical machine learning applied to materials science, and in particular, to the structure and properties of molecules. We will focus on molecular representation for machine learning, as well as the widely-used kernel methods for machine learning molecular properties. In the hands-on tutorial, we will get acquainted with descriptor building and simple machine learning applications.

      Speaker: Prof. Milica Todorovic (University of Turku)
    • 09:45 10:15
      break 30m
    • 10:15 11:15
      Introduction to ML: Kernel methods 1h
      Speaker: Prof. Milica Todorovic (University of Turku)
    • 11:15 11:45
      break 30m
    • 11:45 12:45
      Hands-on: Kernel methods and representations 1h
      Speaker: Prof. Milica Todorovic (University of Turku)
    • 12:45 14:00
      break 1h 15m
    • 14:00 15:00
      Chemical structure search using ML 1h

      In the “Chemical structure search using ML”, we will compare active learning against classical machine learning, and consider how it could be applied to molecular conformer search. The accompanying tutorial will serve to gain practical experience with this application. In the final lecture on “Advanced applications of ML” we will consider more complex ML application to spectral properties of molecules and to molecular design.

      Speaker: Prof. Milica Todorovic (University of Turku)
    • 15:00 15:15
      break 15m
    • 15:15 16:15
      Hands-on: Chemical structure search using ML 1h
      Speaker: Prof. Milica Todorovic (University of Turku)
    • 16:15 16:30
      break 15m
    • 16:30 17:30
      Advanced applications of ML 1h
      Speaker: Prof. Milica Todorovic (University of Turku)
    • 08:30 09:45
      Introduction to ab-initio simulation in VASP 1h 15m
      Speaker: Dr Martijn Marsman (University of Vienna)
    • 09:45 10:15
      break 30m
    • 10:15 11:30
      Electronic convergence 1h 15m
      Speaker: Dr Martijn Marsman (University of Vienna)
    • 11:30 12:00
      break 30m
    • 12:00 13:00
      Symmetry and sampling in reciprocal space 1h
      Speaker: Dr Martin Schlipf (University of Vienna)
    • 13:00 14:00
      break 1h
    • 14:00 15:30
      VASP hands-on session 1: Atoms and molecules 1h 30m
      Speakers: Dr Martijn Marsman (University of Vienna) , Dr Martin Schlipf (University of Vienna) , Henrique Miranda (University of Luxembourg) , Dr Marie-Therese Huebsch
    • 15:30 16:00
      break 30m
    • 16:00 17:30
      VASP hands-on session 2: Bulk systems 1h 30m
      Speakers: Dr Martijn Marsman (University of Vienna) , Dr Martin Schlipf (University of Vienna) , Henrique Miranda (University of Luxembourg) , Dr Marie-Therese Huebsch
    • 08:30 09:45
      Introduction to molecular dynamics 1h 15m
      Speaker: Prof. Tomáš Bučko (Comenius University in Bratislava)
    • 09:45 10:15
      break 30m
    • 10:15 11:30
      Machine learning force fields 1h 15m
      Speaker: Prof. Georg Kresse (University of Vienna)
    • 11:30 12:00
      break 30m
    • 12:00 13:00
      High performance computing with VASP 1h
      Speaker: Dr Martin Schlipf (University of Vienna)
    • 13:00 14:00
      break 1h
    • 14:00 15:30
      VASP hands-on session 3: Surface science 1h 30m
      Speakers: Dr Martin Schlipf (University of Vienna) , Dr Marie-Therese Huebsch, Dr Ferenc Karsai (University of Vienna) , Dr Andreas Singraber (University of Vienna)
    • 15:30 16:00
      break 30m
    • 16:00 17:30
      VASP hands-on session 4: Molecular dynamics 1h 30m
      Speakers: Dr Martin Schlipf (University of Vienna) , Dr Marie-Therese Huebsch, Dr Ferenc Karsai (University of Vienna) , Dr Andreas Singraber (University of Vienna)