Data-Driven Learning of Total and Local Energies in Elemental Boron

Deringer, Volker L.; Pickard, Christopher; Csänyi, Gábor

doi:10.1103/physrevlett.120.156001

Cited by 188 publications

(185 citation statements)

References 45 publications

Supporting

Mentioning

180

Contrasting

Order By: Relevance

“…1 Overview of the GAP-RSS protocol as introduced in ref. 37 and extended here by a selection step. We start from a set of randomised seed structures for a given chemical composition (in generations 0-3, 250 each; in generation 4, 5000).…”

Section: Gap Ttingmentioning

confidence: 99%

“…found to be suitable for GAP-RSS in ref. 37. For a more detailed walk-through of the underlying ML framework, we refer the reader to ref.…”

Section: Gap Ttingmentioning

confidence: 99%

“…The initial choice for these have been many-body descriptors, most importantly the Smooth Overlap of Atomic Positions (SOAP), which includes all neighbours of an atom up to a cut-off radius. 55 To improve the stability of the t, similar to our previous work on amorphous materials modelling 40 and structure searching, 36,37 we combine the manybody SOAP expansion with non-parametric two-body ("2b") and three-body ("3b") terms that encode interatomic distances and bond angles, respectively. The 2b and SOAP descriptors have radial cut-offs of 5.0Å, whereas that for the 3b term is 2.6Å (to include only "true" bond angles involving nearest-neighbour contacts).…”

Section: Gap Ttingmentioning

confidence: 99%

“…Very recently, the use of ML to speed up global searching and crystal structure optimization itself has been reported, 41,42 but we focus here on the use of explicit interatomic potentials ("force elds") for this purpose. What is more, in recent work, 37 we have argued that RSS can be used to construct the interatomic potential from scratch, exploring and tting a complex potential energy surface (PES) at the same time. This points towards a more general "data-driven" strategy for atomistic materials modelling.…”

Section: Introductionmentioning

confidence: 99%

“…44 With this in mind, we decided to explore what effect external pressure would have on the result of GAP-RSS for this element; doing so has already been suggested and proven to be benecial for exploring the PES of elemental boron. 37 To sample higher-pressure structures, we apply hydrostatic external pressure during the GAP-RSS relaxation runs, with a randomised value in each run. The values are drawn from an exponential distribution with the probability density P of nding a given pressure p,…”

mentioning

confidence: 99%

See 4 more Smart Citations

Data-driven learning and prediction of inorganic crystal structures

et al. 2018

Self Cite

View full text Add to dashboard Cite

Crystal structure prediction algorithms, including ab initio random structure searching (AIRSS), are intrinsically limited by the huge computational cost of the underlying quantum-mechanical methods. We have recently shown that a novel class of machine learning (ML) based interatomic potentials can provide a way out: by performing a highdimensional fit to the ab initio energy landscape, these potentials reach comparable accuracy but are orders of magnitude faster. In this paper, we develop our approach, dubbed Gaussian approximation potential-based random structure searching (GAP-RSS), towards a more general tool for exploring configuration spaces and predicting structures. We present a GAP-RSS interatomic potential model for elemental phosphorus, which identifies and correctly "learns" the orthorhombic black phosphorus (A17) structure without prior knowledge of any crystalline allotropes. Using the tubular structure of fibrous phosphorus as an example, we then discuss the limits of free searching, and discuss a possible way forward that combines a recently proposed fragment analysis with GAP-RSS. Examples of possible tubular (1D) and extended (3D) hypothetical allotropes of phosphorus as found by GAP-RSS are discussed. We believe that in the future, ML potentials could become versatile and routine computational tools for materials discovery and design.

show abstract

Section: Gap Ttingmentioning

confidence: 99%

“…found to be suitable for GAP-RSS in ref. 37. For a more detailed walk-through of the underlying ML framework, we refer the reader to ref.…”

Section: Gap Ttingmentioning

confidence: 99%

Section: Gap Ttingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Data-driven learning and prediction of inorganic crystal structures

et al. 2018

Self Cite

View full text Add to dashboard Cite

show abstract

Machine Learning Interatomic Potentials as Emerging Tools for Materials Science

2019

Self Cite

View full text Add to dashboard Cite

In pursuit of this goal, atomic-scale computer simulations have long been a central approach, and two major families of methods are routinely used today. On the one hand, there are quantum-mechanical simulations, in which we solve Schrödinger's equation for the electronic structure of molecular and periodic systems, most widely based on density-functional theory (DFT). [6][7][8] These methods provide (largely) reliable results for structural models of materials that normally contain a few tens or hundreds of atoms. State-of-the-art DFT methods can be applied to many material classes, and they are increasingly used for high-throughput screening and "in silico" (computer-based) design of materials: new compositions and previously unknown structures have been identified in DFT searches and subsequently experimentally realized. [5,[9][10][11] On the other hand, interatomic potential models ("force fields"), parameterizing interactions between atoms with (relatively) simple functional forms, are widely used in materials science to describe matter in molecular dynamics (MD) simulations. These simulations grant access to larger time and length scales, reaching system sizes of up to hundreds of thousands of atoms. [12] In parameterizing these potentials, a certain physical form of the atomic interactions is assumed, often in terms of bond distances, angles, and so on, and physical properties such as equilibrium lattice parameters or elastic constants enter the fitting of the potential. For this reason, such potentials are often called "empirical." They are several orders of magnitude faster than DFT, but necessarily less accurate and less easily transferable.In this Progress Report, we highlight recent developments in "machine-learned" interatomic potentials, which represent a rapidly growing field that promises to do away with the aforementioned trade-offs. Over the last year, there has been a surge of interest in machine learning (ML) methodology: part of it is due to the dramatic growth of ML throughout the scientific disciplines, and part of it is due to tangible success stories of ML-based interatomic potentials that are now beginning to emerge. We will argue that this is an exciting development with very practical implications, currently on the verge of moving from a somewhat specialized new technology to everyday applicability, poised to enhance and complement the communities' existing strengths in computational materials modeling. We will show selected applications of ML potentials to problems in materials science, discuss the current limitations (and possible pitfalls), and outline what we expect to be interesting directions for the development of the field in the coming years.Atomic-scale modeling and understanding of materials have made remarkable progress, but they are still fundamentally limited by the large computational cost of explicit electronic-structure methods such as density-functional theory. This Progress Report shows how machine learning (ML) is currently enabling a new degree of realism in materi...

show abstract

Atomically Thin 2D Transition Metal Oxides: Structural Reconstruction, Interaction with Substrates, and Potential Applications

Yang

Song

Callsen

et al. 2018

Adv Materials Inter

117

View full text Add to dashboard Cite

Most of the large variety of 2D materials are derived from their parent van der Waals (vdWs) crystals, thus their atomic struc tures are identical to the bulk counterparts with some minor lattice constant differ ence due to the reduced dimensionality and vanished interlayer interaction. Simi larly, some 2D monolayers, of which a typical example would be silicene, [10] stem from bulk materials with covalent bonds. Although their bonding may significantly change (silicene: sp 2 ; bulk silicon: sp 3 ) when they go from the bulk material to the 2D monolayer, their atomic structures are analogous. However, a slight atomic relaxation may occur in order to balance the reduced bonding coordination in the 2D forms. Therefore, in these cases, the determination of the 2D atomic structures is quite straightforward.In contrast, most metal oxides feature strong interlayer ionic bonds. The lack of a strong interlayer interaction in their 2D forms usually introduces dangling bonds, leading to strong surface polarization which induces surface instability of 2D metal oxides. Pronounced lattice relaxation, prominent struc tural reconstruction and substrate effects have been identified as the main mechanisms for compensating such strong sur face polarization in 2D metal oxides, as have been observed for a Pd 5 O 4 overlayer on Pd(111), [22] a strained PdO(101) layer on Pd (100), [23] a Ag 1.83 O trilayer on Ag(111), [24] a RhO 2 trilayer on Rh(111), [25] multiple phases of 2D Mn oxides on Pd(100), [26] and TiO 2 on rutile TiO 2 (011). [27] All of these significant changes increase the difficulty of synthesizing 2D metal oxides, as well as pose a challenge to computational structure prediction methods. Notwithstanding, more recently, spectacular progress has been made in prediction, design, preparation, and charac terization of oxide monolayers owing to the advancement of growth technologies and novel synthesis routes, as well as the development of computational and theoretical methods. [28][29][30][31][32][33] The structural reconstructions in combination with the elec tron confinement in 2D and the large surfacetovolume ratio endow 2D transition metal oxides (TMOs) with stunning physical/chemical properties. Moreover, the 2D TMOs showThe discovery of graphene has stimulated dramatic research interest on other 2D materials including transition metal oxide (TMO) monolayers in order to realize novel functionalization and applications. Due to reduced bonding coordination and strong surface polarization, the structures of most TMOs in the monolayer limit are very different from their bulk counterparts, as well as their physical and chemical properties. In this brief review, the authors sum marize recent research progress on atomically thin TMO layers. The focus is on the structural properties of the TMOs and their interaction with the sub strates from the computational point of view. The authors also introduce the potential applications of the TMO 2D materials on supercapacitors, photo catalysts, batteries, and sensors.

show abstract

Data-Driven Learning of Total and Local Energies in Elemental Boron

Cited by 188 publications

References 45 publications

Data-driven learning and prediction of inorganic crystal structures

Data-driven learning and prediction of inorganic crystal structures

Machine Learning Interatomic Potentials as Emerging Tools for Materials Science

Atomically Thin 2D Transition Metal Oxides: Structural Reconstruction, Interaction with Substrates, and Potential Applications

Contact Info

Product

Resources

About