AFLOW-ML: A RESTful API for machine-learning predictions of materials properties

Gossett, Eric; Toher, Cormac; Oses, Corey; Isayev, Olexandr; Legrain, Fleur; Rose, Frisco; Zurek, Eva; Carrete, Jesús; Mingo, Natalio; Tropsha, Alexander; Curtarolo, Stefano

doi:10.1016/j.commatsci.2018.03.075

Cited by 86 publications

(48 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Herein, we illustrate that the Vickers hardnesses, H v , of a wide variety of crystalline materials predicted by using a macroscopic hardness model in conjunction with ML-derived bulk and shear moduli obtained via the RESTful API [32] available on AFLOW are in excellent agreement with results obtained from first-principles calculations. Both are in good agreement with experiment.…”

Section: Introductionsupporting

confidence: 59%

Predicting superhard materials via a machine learning informed evolutionary structure search

et al. 2019

Self Cite

View full text Add to dashboard Cite

Good agreement was found between experimental Vickers hardnesses, Hv, of a wide range of materials and those calculated by three macroscopic hardness models that employ the shear and/or bulk moduli obtained from: (i) first principles via AFLOW-AEL (AFLOW Automatic Elastic Library), and (ii) a machine learning (ML) model trained on materials within the AFLOW repository. Because H ML v values can be quickly estimated, they can be used in conjunction with an evolutionary search to predict stable, superhard materials. This methodology is implemented in the XTALOPT evolutionary algorithm. Each crystal is minimized to the nearest local minimum, and its Vickers hardness is computed via a linear relationship with the shear modulus discovered by Teter. Both the energy/enthalpy and H ML v, Teter are employed to determine a structure's fitness. This implementation is applied towards the carbon system, and 43 new superhard phases are found. A topological analysis reveals that phases estimated to be slightly harder than diamond contain a substantial fraction of diamond and/or lonsdaleite. arXiv:1906.05886v1 [cond-mat.mtrl-sci]

show abstract

Section: Introductionsupporting

confidence: 59%

Predicting superhard materials via a machine learning informed evolutionary structure search

et al. 2019

Self Cite

View full text Add to dashboard Cite

show abstract

“…Our trained model and AFLOW-ML [31][32][33] predicted 490 carbon structures, the results are shown in Figure 4. In Figure 4 To confirm the stability of the novel phase, the phonon spectra of the new carbon phase were calculated.…”

Section: Resultsmentioning

confidence: 98%

Accelerating inverse crystal structure prediction by machine learning: A case study of carbon allotropes

et al. 2020

View full text Add to dashboard Cite

Based on structure prediction method, the machine learning method is used instead of the density function theory (DFT) method to predict the material properties, thereby accelerating the material search process. In this paper, we established a data set of carbon materials by high-throughput calculation with available carbon structures obtained from the Samara Carbon Allotrope Database. We then trained an ML model that specifically predicts the elastic modulus (bulk modulus, shear modulus, and the Young's modulus) and confirmed that the accuracy is better than that of AFLOW-ML in predicting the elastic modulus of a carbon allotrope. We further combined our ML model with the CALYPSO code to search for new carbon structures with a high Young's modulus. A new carbon allotrope not included in the Samara Carbon Allotrope Database, named Cmcm-C24, which exhibits a hardness greater than 80 GPa, was firstly revealed. The Cmcm-C24 phase was identified as a semiconductor with a direct bandgap. The structural stability, elastic modulus, and electronic properties of the new carbon allotrope were systematically studied, and the obtained results demonstrate the feasibility of ML methods accelerating the material search process.

show abstract

“…The main features include a search bar where information such as ICSD reference number, AFLOW unique identifier (AUID) or the chemical formula can be entered in order to retrieve specific materials entries. Below are buttons linking to several different online applications such as the advanced search functionality, convex hull phase diagram generators, machine learning applications [45,160,161] and AFLOW-online data analysis tools. The link to the advanced search application is highlighted by the orange square, and the application page is shown in Figure 6(b).…”

Section: A Computational Materials Data Web Portalsmentioning

confidence: 99%

“…The use of APIs to provide programmatic access is being extended beyond materials data retrieval, to enable the remote use of pre-trained machine learning algorithms. The AFLOW-ML API [161] facilitates access to the two machine learning models that are also available online at aflow.org/aflow-ml [45,160]. The API allows users to submit structural data for the material of interest using a utility such as cURL, and then returns the results of the model's predictions in JSON format.…”

Section: B Programmatically Accessible Online Repositories Of Computmentioning

confidence: 99%

Automated Computation of Materials Properties

Toher

Oses

Curtarolo

2019

Materials Informatics

Self Cite

View full text Add to dashboard Cite

Materials informatics offers a promising pathway towards rational materials design, replacing the current trial-and-error approach and accelerating the development of new functional materials. Through the use of sophisticated data analysis techniques, underlying property trends can be identified, facilitating the formulation of new design rules. Such methods require large sets of consistently generated, programmatically accessible materials data. Computational materials design frameworks using standardized parameter sets are the ideal tools for producing such data. This work reviews the state-of-the-art in computational materials design, with a focus on these automated ab-initio frameworks. Features such as structural prototyping and automated error correction that enable rapid generation of large datasets are discussed, and the way in which integrated workflows can simplify the calculation of complex properties, such as thermal conductivity and mechanical stability, is demonstrated. The organization of large datasets composed of ab-initio calculations, and the tools that render them programmatically accessible for use in statistical learning applications, are also described. Finally, recent advances in leveraging existing data to predict novel functional materials, such as entropy stabilized ceramics, bulk metallic glasses, thermoelectrics, superalloys, and magnets, are surveyed. * stefano@duke.edu 1 S. Curtarolo, G. L. W. Hart, M. Buongiorno Nardelli, N. Mingo, S. Sanvito, and O. Levy, The high-throughput highway to computational materials design, Nat. Mater. 12, 191-201 (2013).

show abstract

AFLOW-ML: A RESTful API for machine-learning predictions of materials properties

Cited by 86 publications

References 62 publications

Predicting superhard materials via a machine learning informed evolutionary structure search

Predicting superhard materials via a machine learning informed evolutionary structure search

Accelerating inverse crystal structure prediction by machine learning: A case study of carbon allotropes

Automated Computation of Materials Properties

Contact Info

Product

Resources

About