Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 68: Topical Session: Data Driven Materials Science - Descriptors (joint session MM/CPP)

CPP 68.6: Vortrag

Mittwoch, 18. März 2020, 13:00–13:15, BAR 205

Machine-learning descriptors with domain knowledge of the interatomic bond — •Thomas Hammerschmidt, Jan Jenke, Aparna P.A. Subramanyam, Jörg Koßmann, Yury Lysogorskiy, and Ralf Drautz — ICAMS, Ruhr-Universität Bochum, Germany

The performance of machine-learning depends critically on the quality of the descriptors. In the case of learning atomic-scale properties, like formation energies obtained from density-functional theory (DFT) calculations, the descriptors typically measure the atomistic geometry and the distribution of chemical elements. Here, we construct descriptors that additionally include prior knowledge of the interatomic bond from a hierarchy of coarse-grained electronic-structure methods. In particular, we use tight-binding (TB) and analytic bond-order potentials (BOPs) that are derived from a second-order expansion of DFT. We demonstrate that a recursive solution of the TB problem and the closely related moments of the electronic density-of-states at the BOP level establish a smooth structure-energy relation. This first level of domain knowledge of the interatomic bond shows highly descriptive power in machine-learning applications already with simple, qualitative TB models. As second level of domain knowledge we include the bond chemistry in terms of bond-specific TB Hamiltonians that are obtained from downfolding the DFT eigenspectrum of molecular dimers. In the third level of domain knowledge we include the role of the valence electrons by determining non-selfconsistent bond energies with the bond-specific TB Hamiltonians.