An entropy-maximization approach to automated training set generation for interatomic potentials

Karabin, Mariia; Pérez, Danny

doi:10.1063/5.0013059

Cited by 26 publications

(21 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Entropy-maximization techniques 30,31 help to partially overcome these problems by maximizing the structural diversity of a data set. When acquiring new data, these methods are focused on the structural dissimilarity compared to the existing data.…”

Section: Mainmentioning

confidence: 99%

Uncertainty Driven Dynamics for Active Learning of Interatomic Potentials

Kulichenko

Barros

Lubbers

et al. 2022

Preprint

View full text Add to dashboard Cite

Machine learning (ML) models, if trained to datasets of high-fidelity quantum simulations, produce accurate and efficient interatomic potentials. Active learning (AL) is a powerful tool to iteratively generate diverse datasets. In this approach, the ML model provides an uncertainty estimate along with its prediction for each new atomic configuration. If the uncertainty estimate passes a certain threshold, then the configuration is included in the dataset. A key challenge in this process is locating structures for which the model lacks underlying training data. Here, we develop a strategy to more rapidly discover configurations that meaningfully augment the training dataset. The approach, uncertainty driven dynamics for active learning (UDD-AL), modifies the potential energy surface used in molecular dynamics simulations to favor regions of configuration space for which there is large model uncertainty. Performance of UDD-AL is demonstrated for two challenging AL tasks: sampling the conformational space of glycine and sampling the promotion of proton transfer in acetylacetone. The method is shown to efficiently explore chemically relevant configuration space, which may be inaccessible using regular dynamical sampling at target temperature conditions.

show abstract

Section: Mainmentioning

confidence: 99%

Uncertainty Driven Dynamics for Active Learning of Interatomic Potentials

Kulichenko

Barros

Lubbers

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Generation of reference structures suitable for tuning machine learning models has been explored in numerous studies [26,27,37,43,124,137,138,139,140,141,142,123,125,126,143,127,144]. Ab initio MD has been a particularly popular approach to sample physically meaningful configurations [26,27].…”

Section: Maise-net: Automated Generator Of Neural Networkmentioning

confidence: 99%

“…A similar approach was developed by Dolgirev et al [145]. Several strategies to improve the mapping of configuration spaces have been developed in recent years, e.g., normal mode sampling [142], active learning-based models [125,126], enhanced sampling [127], ab initio random structure searching [123,124], and entropy-maximization approach [143]. A number of studies have shown the benefit of iterating the generation of data and the parameterization of models [26,34,41,123,124,125,126,144].…”

Section: Maise-net: Automated Generator Of Neural Networkmentioning

confidence: 99%

MAISE: Construction of neural network interatomic models and evolutionary structure optimization

Hajinazar,

Thorn,

Sandoval

et al. 2020

Preprint

View full text Add to dashboard Cite

MAISE is an open-source package for materials modeling and prediction. The code's main feature is an automated generation of neural network (NN) interatomic potentials for use in global structure searches. The systematic construction of Behler-Parrinello-type NN models approximating ab initio energy and forces relies on two approaches introduced in our recent studies. An evolutionary sampling scheme for generating reference structures improves the NNs' mapping of regions visited in unconstrained searches, while a stratified training approach enables the creation of standardized NN models for multiple elements. A more flexible NN architecture proposed here expands the applicability of the stratified scheme for an arbitrary number of elements. The full workflow in the NN development is managed with a customizable 'MAISE-NET' wrapper written in Python. The global structure optimization capability in MAISE is based on an evolutionary algorithm applicable for nanoparticles, films, and bulk crystals. A multitribe extension of the algorithm allows for an efficient simultaneous optimization of nanoparticles in a given size range. Implemented structure analysis functions include fingerprinting with radial distribution functions and finding space groups with the SPGLIB tool. This work overviews MAISE's available features, constructed models, and confirmed predictions.

show abstract

“…We will discuss in this paper how to construct and utilize structures that are not energetically near the equilibrium structure. An approach in that direction is the recent work from M. Karabin and D. Perez [16]. Their work aims to sample the descriptor space of the targeted potential model as widely and unbiased as possible.…”

Section: Introductionmentioning

confidence: 99%

Systematic Atomic Structure Datasets for Machine Learning Potentials: Application to Defects in Magnesium

Poul¹,

Huber²,

Bitzek³

et al. 2022

Preprint

View full text Add to dashboard Cite

We present a physically motivated strategy for the construction of training sets for transferable machine learning interatomic potentials. It is based on a systematic exploration of all possible space groups in random crystal structures, together with deformations of cell shape, size, and atomic positions. The resulting potentials turn out to be unbiased and generically applicable to studies of bulk defects without including any defect structures in the training set or employing any additional Active Learning. Using this approach we construct transferable potentials for pure magnesium that reproduce the properties of hexagonal closed packed (hcp) and body centered cubic (bcc) polymorphs very well. In the process we investigate how different types of training structures impact the properties and the predictive power of the resulting potential.

show abstract

An entropy-maximization approach to automated training set generation for interatomic potentials

Cited by 26 publications

References 33 publications

Uncertainty Driven Dynamics for Active Learning of Interatomic Potentials

Uncertainty Driven Dynamics for Active Learning of Interatomic Potentials

MAISE: Construction of neural network interatomic models and evolutionary structure optimization

Systematic Atomic Structure Datasets for Machine Learning Potentials: Application to Defects in Magnesium

Contact Info

Product

Resources

About