Crystal structure prediction accelerated by Bayesian optimization

Yamashita, Tomoki; Sato, Nobuya; Kino, Hiori; Miyake, Takashi; Tsuda, Koji; Oguchi, Tamio

doi:10.1103/physrevmaterials.2.013803

Cited by 132 publications

(97 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To perform the framework of BO, we adopted the fingerprint of Oganov and Valle 21 as the descriptor of structures in the previous work. 15 A fingerprint is a vector representation of a structure. The vector is calculated by a fingerprint function that is invariant with respect to shifts in the coordinate system, rotations, and reflections.…”

Section: Resultsmentioning

confidence: 99%

“…20 In the CSP using BO, 15 a stable structure is searched by repeating a selection of an initial structure among candidate initial ones and relaxation of it. Unlike LAQA, sLAQA, and Greedy, a selected structure is fully optimized.…”

Section: Resultsmentioning

confidence: 99%

“…Initial structure generation and local optimization We randomly generated initial structures with specific space groups, using CrySPY, 22 as described by Yamashita et al 15 Once the space group is specified, some lattice parameters are fixed by the symmetry. The remaining unfixed lattice parameters are taken at random.…”

Section: Resultsmentioning

confidence: 99%

“…15 In the RS approach, initial structures are randomly generated and fully relaxed. Here, we first generate a number of initial structure candidates, and perform RS-based CSP by randomly selecting a structure to relax them.…”

Section: Resultsmentioning

confidence: 99%

“…To date, several searching algorithms in CSP have been successfully developed such as random search, [2][3][4] simulated annealing, 5,6 basin hopping, 7 minima hopping, 8,9 evolutionary algorithm (EA), [10][11][12] particle-swarm optimization (PSO), 13,14 and Bayesian optimization (BO). 15 Ab initio random structure searching by Pickard and Needs is quite simple but still one of the most efficient approaches even today. EA and PSO are also popular and efficient algorithms as implemented in USPEX [10][11][12] and CALYPSO.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Fine-grained optimization method for crystal structure prediction

et al. 2018

Self Cite

View full text Add to dashboard Cite

Crystal structure prediction based on first-principles calculations is often achieved by applying relaxation to randomly generated initial structures. Relaxing a structure requires multiple optimization steps. It is time consuming to fully relax all the initial structures, but it is difficult to figure out which initial structure leads to the optimal solution in advance. In this paper, we propose a optimization method for crystal structure prediction, called Look Ahead based on Quadratic Approximation, that optimally assigns optimization steps to each candidate structure. It allows us to identify the most stable structure with a minimum number of total local optimization steps. Our simulations using known systems Si, NaCl, Y 2 Co 17 , Al 2 O 3 , and GaAs showed that the computational cost can be reduced significantly compared to random search. This method can be applied for controlling all kinds of local optimizations based on first-principles calculations to obtain best results under restricted computational resources.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Fine-grained optimization method for crystal structure prediction

et al. 2018

Self Cite

View full text Add to dashboard Cite

show abstract

Machine Learning Guided Synthesis of Flash Graphene

et al. 2022

View full text Add to dashboard Cite

nology, nanomaterials must be synthesized by rapid and scalable processes that do not deleteriously affect their properties. To address this challenge, we and others recently reported the synthesis of graphene, [1][2][3] as well as mixed-phase MoS 2 and WS 2 , [4] high-entropy alloy NPs, [5,6] nanodiamond, [7] and other nanomaterials using the electrothermal flash Joule heating effect. The graphene product was called "flash graphene" after the intense black body radiation produced during the electrical discharge. Flash Joule heating permits the conversion of amorphous carbon, including waste such as pyrolyzed rubber tires, [8] ash by-products from plastic recycling, [9] or landfill-grade mixed plastic waste, [10] into graphene crystals. Furthermore, flash graphene crystals are turbostratic and exhibit varying degrees of layer-to-layer misorientation along the c-axis. [1] Such turbostratic graphene possesses nanostructure-dependent properties, including enhanced solubility in surfactant solutions [1] and altered band structure. [11] The scalable and environmentally friendly nature of the Joule heating process, as well as the turbostratic nature of the synthesized product, make flash Joule heating an intriguing synthetic technique that warrants further study and analysis.Although flash Joule heating has immense practical utility, it is intrinsically difficult to study. The flash graphene formation process occurs in just hundreds of milliseconds. Furthermore, present-day flash Joule heating reactors do not offer control over the current discharge profile, adding a stochastic element to each reaction that depends on momentary fluctuations in circuit-to-sample contact. These fluctuations are difficult to control experimentally, making it challenging to map processstructure-property relationships by a traditional grid-search. Due to these factors, the parameters that drive bulk nanocrystal formation during flash Joule heating remain ambiguous.At the same time, an emerging body of literature indicates that machine learning (ML) is a powerful tool for fundamental studies in materials science. [12][13][14][15][16][17][18] While ML is classically considered an industrial tool for process failure prevention, the use of ML to interrogate large parameter spaces can yield insights on new technologies at a low time-cost. For example, Tang et al. used ML to explore the process-structure-property relationships governing well-understood processes, such as chemical vapor deposition and quantum dot synthesis, and argued based on their results that ML would allow researchers to investigate Advances in nanoscience have enabled the synthesis of nanomaterials, such as graphene, from low-value or waste materials through flash Joule heating. Though this capability is promising, the complex and entangled variables that govern nanocrystal formation in the Joule heating process remain poorly understood. In this work, machine learning (ML) models are constructed to explore the factors that drive the transformation of amorphous carbon into gra...

show abstract

Optimization of Core–Shell Nanoparticles Using a Combination of Machine Learning and Ising Machine

Urushihara,

Karube,

Yamaguchi

et al. 2023

Advanced Photonics Research

View full text Add to dashboard Cite

Machine‐learning‐based optimization techniques are widely used for designing complex materials. However, an efficient search for the complex systems, where a combinatorial explosion occurs in the materials search space, is still challenging. Core–shell nanoparticles (CSNPs) are an example of a complex system with a high degree of freedom owing to their complicated structure and multiple constituent materials. In this study, a new black box optimization technique is developed. In this method, the structure of the CSNPs is optimized using an Ising machine and their constituent materials are selected using Bayesian optimization with the optical properties of the materials as the “descriptors”. Aiming for applications to i‐line photolithography, the authors search for CSNPs that are transparent to ultraviolet light of wavelength 355–375 nm and opaque to visible light of wavelength 400–830 nm. The transmittance spectra of the nanoparticles are obtained using a Mie theory calculator. The proposed nanoparticles with the best optical properties have a multilayered structure with a radius of approximately 40 nm and an outer shell composed of either Mg or Pb. The results indicate that a combination of various optimization techniques is more efficient in discovering the better complex materials.

show abstract

Crystal structure prediction accelerated by Bayesian optimization

Cited by 132 publications

References 35 publications

Fine-grained optimization method for crystal structure prediction

Fine-grained optimization method for crystal structure prediction

Machine Learning Guided Synthesis of Flash Graphene

Optimization of Core–Shell Nanoparticles Using a Combination of Machine Learning and Ising Machine

Contact Info

Product

Resources

About