Impact of training and validation data on the performance of neural network potentials: A case study on carbon using the CA-9 dataset

Hedman, Daniel; Rothe, Tom; Johansson, Gustav; Sandin, Fredrik; Larsson, J. Andreas; Miyamoto, Yoshiyuki

doi:10.1016/j.cartre.2021.100027

Cited by 4 publications

(2 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In recent years, machine learning (ML) has been applied to the development of interatomic potentials. [41][42][43][44][45][46][47][48][49][50][51][52] Unlike the CBOPs mentioned above, machine learning interatomic potentials (ML-IAPs) do not rely on fixed mathematical expressions but instead learn the mathematical representation of the potential energy surface (PES) via training, which allows for more accurate predictions of energy, forces, and so on. For example, Deringer and Csányi built a Gaussian approximation potential (GAP) for liquid and amorphous carbon that showed close-to-DFT accuracy.…”

Section: Introductionmentioning

confidence: 99%

“…51 Hedman et al used deep learning to train neural network potentials on the nine-carbon allotrope dataset (CA-9), which reproduced ab initio results with high accuracy. 52 Finally, Rowe et al developed the GAP-17 potential for graphene using the GAP model 46 and, most recently, the GAP-20 potential for various crystalline phase carbon and amorphous carbon, 50 including dispersion corrections and long-range interactions. Although current ML-IAPs have shown excellent accuracy in predicting static and dynamic properties of carbon allotropes, there is still a clear lack of testing related to the prediction of mechanical properties.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A comprehensive assessment of empirical potentials for carbon materials

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A comprehensive assessment of empirical potentials for carbon materials

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

Carbon science perspective in 2022: Current research and future challenges

Meunier

Ania

Bianco

et al. 2022

Carbon

View full text Add to dashboard Cite

ColabFit exchange: Open-access datasets for data-driven interatomic potentials

Vita,

Fuemmeler,

Gupta

et al. 2023

The Journal of Chemical Physics

View full text Add to dashboard Cite

Data-driven interatomic potentials (IPs) trained on large collections of first principles calculations are rapidly becoming essential tools in the fields of computational materials science and chemistry for performing atomic-scale simulations. Despite this, apart from a few notable exceptions, there is a distinct lack of well-organized, public datasets in common formats available for use with IP development. This deficiency precludes the research community from implementing widespread benchmarking, which is essential for gaining insight into model performance and transferability, and also limits the development of more general, or even universal, IPs. To address this issue, we introduce the ColabFit Exchange, the first database providing open access to a large collection of systematically organized datasets from multiple domains that is especially designed for IP development. The ColabFit Exchange is publicly available at https://colabfit.org, providing a web-based interface for exploring, downloading, and contributing datasets. Composed of data collected from the literature or provided by community researchers, the ColabFit Exchange currently (September 2023) consists of 139 datasets spanning nearly 70 000 unique chemistries, and is intended to continuously grow. In addition to outlining the software framework used for constructing and accessing the ColabFit Exchange, we also provide analyses of the data, quantifying the diversity of the database and proposing metrics for assessing the relative diversity of multiple datasets. Finally, we demonstrate an end-to-end IP development pipeline, utilizing datasets from the ColabFit Exchange, fitting tools from the KLIFF software package, and validation tests provided by the OpenKIM framework.

show abstract

Impact of training and validation data on the performance of neural network potentials: A case study on carbon using the CA-9 dataset

Cited by 4 publications

References 44 publications

A comprehensive assessment of empirical potentials for carbon materials

A comprehensive assessment of empirical potentials for carbon materials

Carbon science perspective in 2022: Current research and future challenges

ColabFit exchange: Open-access datasets for data-driven interatomic potentials

Contact Info

Product

Resources

About