Recent advances in describing and driving crystal nucleation using machine learning and artificial intelligence
arXiv:2304.13815 · doi:10.1016/j.cossms.2023.101093
Abstract
With the advent of faster computer processors and especially graphics processing units (GPUs) over the last few decades, the use of data-intensive machine learning (ML) and artificial intelligence (AI) has increased greatly, and the study of crystal nucleation has been one of the beneficiaries. In this review, we outline how ML and AI have been applied to address four outstanding difficulties of crystal nucleation: how to discover better reaction coordinates (RCs) for describing accurately non-classical nucleation situations; the development of more accurate force fields for describing the nucleation of multiple polymorphs or phases for a single system; more robust identification methods for determining crystal phases and structures; and as a method to yield improved course-grained models for studying nucleation.
arXiv comment15 pages; 1 figure
References in corpus (40)
- Denoising Diffusion Probabilistic Models
- On representing chemical environments
- Normalizing Flows: An Introduction and Review of Current Methods
- Robust Structural Identification via Polyhedral Template Matching
- Identification of slow molecular order parameters for Markov model construction
- Comparing molecules and solids across structural and alchemical space
- Discovering Phase Transitions with Unsupervised Learning
- How van der Waals interactions determine the unique properties of water
- Unsupervised learning of phase transitions: from principal component analysis to variational autoencoders
- Machine learning a general purpose interatomic potential for silicon
- Time-lagged autoencoders: Deep learning of slow collective variables for molecular kinetics
- The Phase Diagram of a Deep Potential Water Model
- freud: A Software Suite for High Throughput Analysis of Particle Simulation Data
- Caliber based spectral gap optimization of order parameters (SGOOP) for sampling complex molecular systems
- Data-driven model reduction and transfer operator approximation
- Molecular enhanced sampling with autoencoders: On-the-fly collective variable discovery and accelerated free energy landscape exploration
- Coarse Graining Molecular Dynamics with Graph Neural Networks
- Nonlinear Discovery of Slow Molecular Modes using State-Free Reversible VAMPnets
- Silicon liquid structure and crystal nucleation from ab-initio deep Metadynamics
- A Topological Framework for Local Structure Analysis in Condensed Matter
- State Predictive Information Bottleneck
- Neural network based path collective variables for enhanced sampling of phase transformations
- Phase equilibrium of water with hexagonal and cubic ice using the SCAN functional
- Unsupervised learning for local structure detection in colloidal systems
- Homogeneous nucleation of ice
- An analytical coarse-graining method which preserves the free energy, structural correlations, and thermodynamic state of polymer melts from the atomistic to the mesoscale
- Variational Selection of Features for Molecular Kinetics
- Unsupervised machine learning for detection of phase transitions in off-lattice systems I. Foundations
- Splitting probabilities as a test of reaction coordinate choice in single-molecule experiments
- Overcoming timescale and finite-size limitations to compute nucleation rates from small scale Well Tempered Metadynamics simulations
- Driving and characterizing nucleation of urea and glycine polymorphs in water
- Collective Variables for the Study of Crystallization
- Unsupervised machine learning for detection of phase transitions in off-lattice systems II. Applications
- Capabilities and Limitations of Time-lagged Autoencoders for Slow Mode Discovery in Dynamical Systems
- Reaction coordinates and rate constants for liquid droplet nucleation: quantifying the interplay between driving force and memory
- Measuring transferability issues in machine-learning force fields: The example of Gold-Iron interactions with linearized potentials
- GCIceNet: A Graph Convolutional Network for Accurate Classification of Water Phases
- A general-purpose machine learning Pt interatomic potential for an accurate description of bulk, surfaces and nanoparticles
- How to quantify and avoid finite size effects in computational studies of crystal nucleation: The case of homogeneous crystal nucleation
- Evaluating the Transferability of Machine-Learned Force Fields for Material Property Modeling