DPG Phi
Verhandlungen
Verhandlungen
DPG

SurfaceScience21 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

O: Fachverband Oberflächenphysik

O 71: Mini-Symposium: Machine learning applications in surface science I

O 71.1: Invited Talk

Wednesday, March 3, 2021, 13:45–14:15, R1

Machine learning for novel functional materials — •Pascal Friederich — Karlsruhe Institute of Technology, Germany

During the last decade, machine learning (ML) algorithms were increasingly applied to questions in the physical sciences, e.g. to automate labs, to accelerate simulations, and to solve inverse problems such as the design of new materials. This talk will show our recent work on combining ML models with conventional tools to accelerate simulations and to obtain new scientific insight. Firstly, we show that ML enables the analysis of energy disorder in amorphous organic semiconductors which is of high relevance to understand charge transport in devices such as OLEDs.[1] Secondly, we will show how ML models can accelerate ab-initio photodynamics simulations of small molecules to unprecedented simulation times of 10 ns and more.[2] Thirdly, we will show how a combination of graph representations and basic ML regression models can provide scientific insight into organic electronics as well as quantum optical experiments in a highly intuitive and human interpretable way.[3]

[1] The influence of sorbitol doping on aggregation and electronic properties of PEDOT:PSS, P. Friederich, S. Leon, J. D. Perea Ospina, L. Roch and A. Aspuru-Guzik, MLST, 2020. [2] Nanosecond Photodynamics Simulations of a cis-transIsomerization are Enabled by Machine Learning, J. Li, P. Reiser, A. Eberhard, P. Friederich, and S. A. Lopez, DOI: 10.26434/chemrxiv.13047863.v1, 2020. [3] Scientific intuition inspired by machine learning generated hypotheses, P. Friederich, M. Krenn, I. Tamblyn, A. Aspuru-Guzik, arXiv:2010.14236, 2020.

100% | Mobile Layout | Deutsche Version | Contact/Imprint/Privacy
DPG-Physik > DPG-Verhandlungen > 2021 > SurfaceScience21