Unsupervised landmark analysis for jump detection in molecular dynamics simulations

Abstract

Molecular dynamics is a versatile and powerful method to study diffusion in solid-state ionic conductors, requiring minimal prior knowledge of equilibrium or transition states of the system's free energy surface. However, the analysis of trajectories for relevant but rare events, such as a jump of the diffusing mobile ion, is still rather cumbersome, requiring prior knowledge of the diffusive process in order to get meaningful results. In this work, we present a novel approach to detect the relevant events in a diffusive system without assuming prior information regarding the underlying process. We start from a projection of the atomic coordinates into a landmark basis to identify the dominant features in a mobile ion's environment. Subsequent clustering in landmark space enables a discretization of any trajectory into a sequence of distinct states. As a final step, the use of the smooth overlap of atomic positions descriptor allows distinguishing between different environments in a straightforward way. We apply this algorithm to ten Li-ionic systems and perform in-depth analyses of cubic Li7La3Zr2O12, tetragonal Li10GeP2S12, and the β-eucryptite LiAlSiO4. We compare our results to existing methods, underscoring strong points, weaknesses, and insights into the diffusive behavior of the ionic conduction in the materials investigated.

Enhanced Li-Ion Conductivity in Nanosized Li10GeP2S12

“…First, the values beyond the first maximum peak for Li-Li indicate that the Li distribution is disordered in LGPS indicative of a superionic conductor. We note that the main maximum peaks of the bulk RDFs at 2.36 and 3.40 Å for Li-S and Li-Li, respectively, are in excellent agreement with the experimental structure6 , as well as RDFs calculated using ab initio MD47 .…”

How Machine Learning Will Revolutionize Electrochemical Sciences

“…Unsupervised learning is suitable to parse through the simulation data and identify such representative events. 31 , 57 , 58 Alternatively, the dimensionality of the data can be reduced to correlate the most essential features. 59 …”

Li4–xGe1–xPxO4, a Potential Solid-State Electrolyte for All-Oxide Microbatteries

“…In addition, we analyzed the trajectory using the SITATOR package for an unsupervised analysis of the diffusive pathways in the system. 35 The tracer diffusion matrix D Li was computed from the NVT trajectories via the mean square displacement (MSD) of lithium, according to the Einstein relation:…”