Detecting dynamic domains and local fluctuations in complex molecular systems via timelapse neighbors shuffling

Abstract

Many complex molecular systems owe their properties to local dynamic rearrangements or fluctuations that, despite the rise of machine learning (ML) and sophisticated structural descriptors, remain often difficult to detect. Here we show an ML framework based on a new descriptor, named Local Environments and Neighbors Shuffling (LENS), which allows identifying dynamic domains and detecting local fluctuations in a variety of systems via tracking how much the surrounding of each molecular unit changes over time in terms of neighbor individuals. Statistical analysis of the LENS time-series data allows to blindly detect different dynamic domains within various types of molecular systems with, e.g., liquid-like, solid-like, or diverse dynamics, and to track local fluctuations emerging within them in an efficient way. The approach is found robust, versatile, and, given the abstract definition of the LENS descriptor, capable of shedding light on the dynamic complexity of a variety of (not necessarily molecular) systems.

References in corpus (4)

• Machine Learning Unifies the Modelling of Materials and Molecules
• Recognizing molecular patterns by machine learning: an agnostic structural definition of the hydrogen bond
• Recognizing Local and Global Structural Motifs at the Atomic Scale
• Detecting dynamic domains and local fluctuations in complex molecular systems via timelapse neighbors shuffling

Cited by in corpus (9)

• Detecting dynamic domains and local fluctuations in complex molecular systems via timelapse neighbors shuffling
• TimeSOAP: Tracking high-dimensional fluctuations in complex molecular systems via time-variations of SOAP spectra
• "Layer-by-layer" Unsupervised Clustering of Statistically Relevant Fluctuations in Noisy Time-series Data of Complex Dynamical Systems
• Reconstructions and Dynamics of $\beta$-Lithium Thiophosphate Surfaces
• Relevant, hidden, and frustrated information in high-dimensional analyses of complex dynamical systems with internal noise
• Data-driven assessment of optimal spatiotemporal resolutions for information extraction in noisy time series data
• LeaPP: Learning Pathways to Polymorphs through machine learning analysis of atomic trajectories
• A data driven approach to classify descriptors based on their efficiency in translating noisy trajectories into physically-relevant information
• Maximum Information Extraction Via Clustering and Minimization of Shannon Entropy