Learning Conductance: Gaussian Process Regression for Molecular Electronics

• DOI: 10.1021/acs.jctc.2c00648
• Publication Date: 2023-01-24T17:16:12Z
• Authors: Michael Deffner; Marc Philipp Weise; Haitao Zhang; Maike Mücke; Jonny Proppe; Ignacio Franco; Carmen Herrmann
• Keywords: 02 engineering and technology; 0210 nano-technology; 01 natural sciences; 0104 chemical sciences

Experimental studies of charge transport through single molecules often rely on break junction setups, where molecular junctions are repeatedly formed and broken while measuring the conductance, leading to a statistical distribution conductance values. Modeling this experimental situation resulting histograms is challenging for theoretical methods, as computations need capture structural changes in experiments, including statistics formation rupture. This type extensive sampling implies that even when evaluating from computationally efficient electronic structure which typically reduced accuracy, evaluation too expensive be routine task. Highly accurate quantum only feasible few selected conformations thus necessarily ignore rich conformational space probed experiments. To overcome these limitations, we investigate potential machine learning modeling histograms, particular by Gaussian process regression. We show selecting specific parameters features, regression can used efficiently predict zero-bias structures, reducing computational cost simulating an order magnitude. enables calculation basis first-principles approaches effectively number necessary calculations, paving way toward their evaluation.