Topic: machine learning approaches in materials science, descriptors

Download optional Lab 5 for water adsorption. Be prepared: SISSO is needed

Lecture slides

Practice

  • Lab 5 for water adsorption (SISSO)

Important: Special rules for Lab 5

  1. If students generate or collect the dataset by themselves, they are allowed to team up with max 3 students per team.
  2. If students use provided but raw dataset, they are allowed to team up with max 2 students per team.
  3. If students use fully processed dataset (all features and target properties are already given), they are not allowed to team up.

* Contribution of every team member must be clarified in the report; increase of team size must be approved by Instructors

Sample public data sources for Lab 5 (option 1 in the above list)

Raw datasets for Lab 5 provided by Instructors (option 2)

  • Intercalation potential from host crystal properties (contact DA, relevant article)
  • Optoelectronic properties of functionalized distyrylbenzene from functionalized benzene (relevant article)
  • Excited states of infinite polymer from dimer calculations (contact AZ, relevant article)
  • Dihedral PES from monomers (contact AZ, relevant article)

Reading

  • Zhong-Kang Han, Debalaya Sarker, Sergey V. Levchenko, Hands-On Compressed Sensing: SISSO (2020) – pdf
  • Z Han, D Sarker, R Ouyang, A Mazheika, Y Gao, S Levchenko, Single-atom alloy catalysts designed by first-principles calculations and artificial intelligence, Nat Commun 12, 1833 (2021) – pdf
  • L Ward, A Agrawal, A Choudhary, C Wolverton, A general-purpose machine learning framework for predicting properties of inorganic materials, npj Comp Mater 2, 16028 (2016)
  • L M Ghiringhelli, J Vybiral, E Ahmetcik, R Ouyang, S V Levchenko, C Draxl, M Scheffler, Learning physical descriptors for materials science by compressed sensing, New J Phys 19, 023017 (2017)

Resources