Random sampling versus active learning algorithms for machine learning potentials of quantum liquid water
arXiv:2410.10698 · doi:10.1021/acs.jctc.4c01382
Abstract
Training accurate machine learning potentials requires electronic structure data comprehensively covering the configurational space of the system of interest. As the construction of this data is computationally demanding, many schemes for identifying the most important structures have been proposed. Here, we compare the performance of high-dimensional neural network potentials (HDNNPs) for quantum liquid water at ambient conditions trained to data sets constructed using random sampling as well as various flavors of active learning based on query by committee. Contrary to the common understanding of active learning, we find that for a given data set size, random sampling leads to smaller test errors for structures not included in the training process. In our analysis we show that this can be related to small energy offsets caused by a bias in structures added in active learning, which can be overcome by using instead energy correlations as an error measure that is invariant to such shifts. Still, all HDNNPs yield very similar and accurate structural properties of quantum liquid water, which demonstrates the robustness of the training procedure with respect to the training set construction algorithm even when trained to as few as 200 structures. However, we find that for active learning based on preliminary potentials, a reasonable initial data set is important to avoid an unnecessary extension of the covered configuration space to less relevant regions.
References in corpus (39)
- CP2K: An Electronic Structure and Molecular Dynamics Software Package -- Quickstep: Efficient and Accurate Electronic Structure Calculations
- E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentials
- Machine learning for molecular simulation
- Less is more: sampling chemical space with active learning
- On-the-fly machine learning force field generation: Application to melting points
- Machine-learning based interatomic potential for amorphous carbon
- Competing quantum effects in the dynamics of a flexible water model
- Active Learning of Uniformly Accurate Inter-atomic Potentials for Materials Simulation
- How van der Waals interactions determine the unique properties of water
- The Phase Diagram of a Deep Potential Water Model
- Automatic Selection of Atomic Fingerprints and Reference Configurations for Machine-Learning Potentials
- The first-principles phase diagram of monolayer nanoconfined water
- A neural network interatomic potential for the phase change material GeTe
- A deep potential model with long-range electrostatic interactions
- Deep Potentials for Materials Science
- Evaluation of the MACE Force Field Architecture: from Medicinal Chemistry to Materials Science
- Committee neural network potentials control generalization errors and enable active learning
- Machine Learning Inter-Atomic Potentials Generation Driven by Active Learning: A Case Study for Amorphous and Liquid Hafnium dioxide
- Structure-based Sampling and Self-correcting Machine Learning for Accurate Calculations of Potential Energy Surfaces and Vibrational Levels
- Raman Spectrum and Polarizability of Liquid Water from Deep Neural Networks
- Automated Fitting of Neural Network Potentials at Coupled Cluster Accuracy: Protonated Water Clusters as Testing Ground
- Ab initio quality neural-network potential for sodium
- The Interplay of Structure and Dynamics in the Raman Spectrum of Liquid Water over the Full Frequency and Temperature Range
- How to validate machine-learned interatomic potentials
- Strategies for the Construction of Machine-Learning Potentials for Accurate and Efficient Atomic-Scale Simulations
- Single-model uncertainty quantification in neural network potentials does not consistently outperform model ensembles
- Tutorial: How to Train a Neural Network Potential
- Fast Uncertainty Estimates in Deep Learning Interatomic Potentials
- Accurate prediction of heat conductivity of water by a neuroevolution potential
- Uncertainty-biased molecular dynamics for learning uniformly accurate interatomic potentials
- High-dimensional neural network potentials for solvation: The case of protonated water clusters in helium
- Converged Colored Noise Path Integral Molecular Dynamics Study of the Zundel Cation down to Ultra-low Temperatures at Coupled Cluster Accuracy
- ArcaNN: automated enhanced sampling generation of training sets for chemically reactive machine learning interatomic potentials
- Machine-learning potentials for nanoscale simulations of deformation and fracture: example of TiB ceramic
- Reducing the cost of neural network potential generation for reactive molecular systems
- A machine learning potential for simulating infrared spectra of nanosilicate clusters
- Size and Quality of Quantum Mechanical Data Sets for Training Neural Network Force Fields for Liquid Water
- Efficient Generation of Stable Linear Machine-Learning Force Fields with Uncertainty-Aware Active Learning
- Self-learning path integral hybrid Monte Carlo with mixed ab initio and machine learning potentials for modeling nuclear quantum effects in water