Dresden 2020 – wissenschaftliches Programm

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

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

SYBD: Symposium Big data driven materials science

SYBD 1: Big Data Driven Materials Science

SYBD 1.3: Hauptvortrag

Dienstag, 17. März 2020, 10:30–11:00, HSZ 02

Verification and error estimates for ab initio data — •Claudia Draxl — Humboldt-Universität zu Berlin, Germany — Fritz-Haber-Institut der Max-Planck-Gesellschaft, Berlin, Germany

Veracity (uncertainty of data quality), one of the 4V challenges of Big Data, is an issue for the FAIRness of (computational) materials-science results. Creating benchmark data and estimating errors are prerequisites for the interoperability of our research data. The precision of the many different computer codes used in the community has been investigated thoroughly by evaluating the equation of state of 71 monoatomic crystals [1]. More recently, it has been demonstrated how ultimate precision for molecules and solids in DFT calculations can been reached [2] and how different methodology impacts the results [3]. We also address code-specific uncertainties that come from numerical settings commonly used in practice [4]. We do so by systematically investigating total and relative energies as a function of computational parameters, employing four popular DFT codes. Based on this, we propose an analytical model for quantifying errors associated with the basis-set incompleteness and predicting converged results. It will be discussed how our approach enables comparison and interoperability of the heterogeneous data present in computational materials databases [5], for the purpose of data-driven research.

[1] K. Lejaeghere et al., Science 351, aad3000 (2016). [2] A. Gulans, A. Kozhevnikov, and C. Draxl, Phys. Rev. B 97, 161105(R) (2018). [3] A. Gulans and C. Draxl, preprint. [4] C. Carbogno, et al., preprint. [5] https://nomad-repository.eu