Bayesian optimization of chemical composition: A comprehensive framework and its application to<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mml:mrow><mml:mi>R</mml:mi><mml:msub><mml:mi>Fe</mml:mi><mml:mn>12</mml:mn></mml:msub></mml:mrow></mml:math>-type magnet compounds

First-principles calculation based on density functional theory is a powerful tool for understanding and designing magnetic materials. It enables us to quantitatively describe magnetic properties and structural stability, although further methodological developments for the treatment of strongly-correlated 4f electrons and finite-temperature magnetism are needed. Here, we review recent developments of computational schemes for rare-earth magnet compounds, and summarize our theoretical studies on Nd 2 Fe 14 B and R Fe 12 -type compounds. Effects of chemical substitution and interstitial dopants are clarified. We also discuss how data-driven approaches are used for studying multinary systems. Chemical composition can be optimized with fewer trials by the Bayesian optimization. We also present a data-assimilation method for predicting finitetemperature magnetization in wide composition space by integrating computational and experimental data.

show abstract

“…In Ref. [100], the Bayesian optimization has been adopted to optimize the chemical composition of the R Fe 12 -type compound. More precisely, the…”

Section: Materials Informaticsmentioning

confidence: 99%

Understanding and optimization of hard magnetic compounds from first principles

Miyake

Harashima

Fukazawa

et al. 2021

Science and Technology of Advanced Materials

Self Cite

View full text Add to dashboard Cite

show abstract

“…Based on existing data, machine learning can learn the correlation between material parameters {y i } and material features {x i } and infer a relation y i =f({x i }) ( Fig. 1) [26]. The features should be easily accessed and can be quantified as a dataset {x i }, which can be used as numerical representation of material parameters {y i }.…”

Section: Introductionmentioning

confidence: 99%

“…As shown in Fig. 2b, Bayesian optimization (BO) is a powerful technique for finding optimal values of function y i =f({x i }) within a given {x i } set [26,27]. Different from one-shot machine-learning approaches, BO can be initialized with a small group of data and evolve into a reasonably good model during sampling-modeling iterations.…”

Section: Introductionmentioning

confidence: 99%

Bayesian optimization based on a unified figure of merit for accelerated materials screening: A case study of halide perovskites

Chen

Wang

et al. 2020

Sci. China Mater.

Self Cite

View full text Add to dashboard Cite

The figure of merit is of crucial importance in materials design to search for candidates with optimal functionality. In the field of photovoltaics, the bandgap (E g ) is a well-recognized figure of merit for screening solar cell absorbers subject to the Shockley-Queisser limit. In this paper, the bandgap as the figure of merit is challenged since an ideal solar cell absorber requires multiple criteria such as stability, optical absorption, and carrier lifetime. Multiple criteria make the quantitative description of material candidates difficult and computationally time-consuming. Taking halide perovskites as an example, we combine thermodynamic stability (ΔH d ) and E g into a unified figure of merit and use Bayesian optimization (BO) to accelerate materials screening. We have found that, in comparison to an exhaustive search via multiple parameters, BO based on the unified figure of merit can screen optimal candidates (E g,PBE between 0.6-1.2 eV, ΔH d >−29 meV per atom) more efficiently. Therefore, the proposed method opens a viable route for the search of optimal solar cell absorbers from a large amount of material candidates with less computational cost.

show abstract

“…This active learning approach to finding compositions with Bayesian optimization is quickly gaining popularity in the informatics community for low-dimensional systems. 61 − 63 …”

Section: Introductionmentioning

confidence: 99%

Investigating Active Learning and Meta-Learning for Iterative Peptide Design

Barrett

White

2020

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Often the development of novel functional peptides is not amenable to high throughput or purely computational screening methods. Peptides must be synthesized one at a time in a process that does not generate large amounts of data. One way this method can be improved is by ensuring that each experiment provides the best improvement in both peptide properties and predictive modeling accuracy. Here, we study the effectiveness of active learning, optimizing experiment order, and meta-learning, transferring knowledge between contexts, to reduce the number of experiments necessary to build a predictive model. We present a multitask benchmark database of peptides designed to advance these methods for experimental design. Each task is a binary classification of peptides represented as a sequence string. We find neither active learning method tested to be better than random choice. The metalearning method Reptile was found to improve the average accuracy across data sets. Combining meta-learning with active learning offers inconsistent benefits.

show abstract

Bayesian optimization of chemical composition: A comprehensive framework and its application toRFe12-type magnet compounds

Cited by 37 publications

References 68 publications

Understanding and optimization of hard magnetic compounds from first principles

Understanding and optimization of hard magnetic compounds from first principles

Bayesian optimization based on a unified figure of merit for accelerated materials screening: A case study of halide perovskites

Investigating Active Learning and Meta-Learning for Iterative Peptide Design

Contact Info

Product

Resources

About