arXiv subjects

Kei Terayama

Publications and source records attributed to Kei Terayama.

11 recordsLinked to original sources

arXiv2026

LLM-guided phase diagram construction through high-throughput experimentation

Constructing phase diagrams for multicomponent alloys requires extensive experimental measurements and is a time-consuming task. Here we investigate whether large language models (LLMs) can guide experimental planning for phase diagram construction. In our framework, a general-purpose LLM serves as the experimental planner, suggesting compositions for measurement at each cycle in a closed loop with high-throughput synthesis and X-ray diffraction phase identification. Using this framework, we experimentally constructed the ternary phase diagram of the Co-Al-Ge system at 900 degree C through iterative synthesis and characterization. We compared two strategies that differ in how the initial compositions are selected: one uses predictions from a domain-specific LLM trained on phase diagram data (aLLoyM), while the other relies solely on the general-purpose LLM. The two strategies exhibited complementary strengths. aLLoyM directed the initial measurements toward compositionally complex regions in the interior of the ternary diagram, enabling the earliest discovery of all three novel phases that form only in the ternary system. In contrast, the general-purpose LLM adopted a textbook-like approach which efficiently identified a larger number of phases in fewer cycles. In addition, a simulated benchmark comparing the LLM against conventional machine learning confirmed that the LLM achieves more efficient exploration. The results demonstrate that LLMs have high potential as experimental planners for phase diagram construction.

cond-mat.mtrl-sci↗

arXiv2026

ZEBRA-Prop: A Zero-Shot Embedding-Based Rapid and Accessible Regression Model for Materials Properties

Large language models (LLMs) exhibit substantial potential across diverse scientific disciplines, including materials science. A property prediction framework, ZEBRA-Prop (Zero-Shot Embedding-Based Rapid and Accessible Regression Model for Materials Properties), is presented here as an extension of LLM-Prop. In contrast to LLM-Prop, which requires task-specific fine-tuning of the LLM, ZEBRA-Prop eliminates fine-tuning, thereby reducing computational cost and enabling rapid model training. The framework employs MatTPUSciBERT, an LLM specialized for materials science, to enhance predictive capability. Multiple textual embeddings are incorporated through a learnable weighting mechanism, which alleviates the context-length constraints inherent in LLM-Prop and facilitates effective integration of diverse textual representations. Evaluation is conducted using two datasets: the TextEdge dataset (approximately 140,000 entries) and an in-house dataset (approximately 2,000 entries) derived from the Materials Project database, with physical properties obtained from first-principles calculations. The predictive performance of ZEBRA-Prop is close to that of LLM-Prop, while the training time is reduced by approximately 95%. The performance improvements are attributable to three principal factors: domain-specific LLM utilization, diversified textual descriptions, and systematic text preprocessing. ZEBRA-Prop constitutes a scalable and computationally efficient framework for materials property prediction and supports accelerated materials discovery, particularly under limited computational resources.

cond-mat.mtrl-sci↗

arXiv2026

Update of PHYSBO: Improving Usability and Portability of Bayesian Optimization for Physics and Materials Research

Bayesian optimization (BO) is widely used to accelerate physics and materials research, where objective function evaluations are computationally or experimentally expensive. While many BO frameworks focus on algorithmic efficiency, practical usability and portability are equally critical for sustained use in real research environments. PHYSBO is a Bayesian optimization library designed to address these needs by enabling optimization over user-defined candidate pools and by supporting domain-specific problem settings. This paper presents the major updates introduced in PHYSBO versions 2 and 3, with a focus on improvements in usability, portability, and practical deployment rather than on new optimization algorithms. In PHYSBO version 2, the software license was changed from GPL to MPL to improve compatibility with a wider range of research and software ecosystems. Building on this revision, PHYSBO version 3 introduces a set of implementation-oriented updates aimed at improving usability and portability, without modifying the core optimization algorithms. These updates include improvements in computational performance and scalability, extended support for multi-objective optimization, the introduction of range-based policies for continuous-variable optimization, the removal of environment-dependent components such as tightly coupled Cython modules, and compatibility with NumPy 2. These improvements reduce the technical and organizational burden on users, enabling PHYSBO to be deployed across diverse computing environments and research workflows. By emphasizing portability and ease of integration while maintaining sufficient performance, PHYSBO version 3 is positioned as a sustainable research infrastructure for Bayesian optimization in physics and materials science.

physics.comp-ph↗

arXiv2024

Extracting effective solutions hidden in large language models via generated comprehensive specialists: case studies in developing electronic devices

Recently, many studies have increasingly explored the use of large language models (LLMs) to generate research ideas and scientific hypotheses. However, real-world research and development often require solving complex, interdisciplinary challenges where solutions may not be readily found through existing knowledge related to the problem. Therefore, it is desirable to leverage the vast, comprehensive knowledge of LLMs to generate effective, breakthrough solutions by integrating various perspectives from other disciplines. Here, we propose SELLM (Solution Enumeration via comprehensive List and LLM), a framework leveraging LLMs and structured guidance using MECE (Mutually Exclusive, Collectively Exhaustive) principles, such as International Patent Classification (IPC) and the periodic table of elements. SELLM systematically constructs comprehensive expert agents from the list to generate cross-disciplinary and effective solutions. To evaluate SELLM's practicality, we applied it to two challenges: improving light extraction in organic light-emitting diode (OLED) lighting and developing electrodes for next-generation memory materials. The results demonstrate that SELLM significantly facilitates the generation of effective solutions compared to cases without specific customization or effort, showcasing the potential of SELLM to enable LLMs to generate effective solutions even for challenging problems.

cs.CL↗

arXiv2024

Data-driven study of the enthalpy of mixing in the liquid phase

The enthalpy of mixing in the liquid phase is a thermodynamic property reflecting interactions between elements that is key to predict phase transformations. Widely used models exist to predict it, but they have never been systematically evaluated. To address this, we collect a large amount of enthalpy of mixing data in binary liquids from a review of about 1000 thermodynamic evaluations. This allows us to clarify the prediction accuracy of Miedema's model which is state-of-the-art. We show that more accurate predictions can be obtained from a machine learning model based on LightGBM, and we provide them in 2415 binary systems. The data we collect also allows us to evaluate another empirical model to predict the excess heat capacity that we apply to 2211 binary liquids. We then extend the data collection to ternary metallic liquids and find that, when mixing is exothermic, extrapolations from the binary systems by Muggianu's model systematically lead to slight overestimations of roughly 10% close to the equimolar composition. Therefore, our LightGBM model can provide reasonable estimates for ternary alloys and, by extension, for multicomponent alloys. Our findings extracted from rich datasets can be used to feed thermodynamic, empirical and machine learning models for material development.

cond-mat.mtrl-sci↗

arXiv2022

Individual health-disease phase diagrams for disease prevention based on machine learning

Early disease detection and prevention methods based on effective interventions are gaining attention. Machine learning technology has enabled precise disease prediction by capturing individual differences in multivariate data. Progress in precision medicine has revealed that substantial heterogeneity exists in health data at the individual level and that complex health factors are involved in the development of chronic diseases. However, it remains a challenge to identify individual physiological state changes in cross-disease onset processes because of the complex relationships among multiple biomarkers. Here, we present the health-disease phase diagram (HDPD), which represents a personal health state by visualizing the boundary values of multiple biomarkers that fluctuate early in the disease progression process. In HDPDs, future onset predictions are represented by perturbing multiple biomarker values while accounting for dependencies among variables. We constructed HDPDs for 11 non-communicable diseases (NCDs) from a longitudinal health checkup cohort of 3,238 individuals, comprising 3,215 measurement items and genetic data. Improvement of biomarker values to the non-onset region in HDPD significantly prevented future disease onset in 7 out of 11 NCDs. Our results demonstrate that HDPDs can represent individual physiological states in the onset process and be used as intervention goals for disease prevention.

cs.LG↗

arXiv2022

Bayesian optimization package: PHYSBO

PHYSBO (optimization tools for PHYSics based on Bayesian Optimization) is a Python library for fast and scalable Bayesian optimization. It has been developed mainly for application in the basic sciences such as physics and materials science. Bayesian optimization is used to select an appropriate input for experiments/simulations from candidate inputs listed in advance in order to obtain better output values with the help of machine learning prediction. PHYSBO can be used to find better solutions for both single and multi-objective optimization problems. At each cycle in the Bayesian optimization, a single proposal or multiple proposals can be obtained for the next experiments/simulations. These proposals can be obtained interactively for use in experiments. PHYSBO is available at https://github.com/issp-center-dev/PHYSBO.

physics.comp-ph↗

arXiv2022

A machine learning-based classification approach for phase diagram prediction

Knowledge of phase diagrams is essential for material design as it helps in understanding microstructure evolution during processing. The determination of phase diagrams is thus one of the central tasks in materials science. When exploring new materials for which the phase diagram is unknown, experimentalists often try to determine the key experiments that should be performed by referencing known phase diagrams of similar systems. To enhance this practical strategy, we attempted to estimate unknown phase diagrams based on known phase diagrams using a machine learning-based classification approach. As a proof of concept, we focused on predicting the number of coexisting phases across the 800 K isothermal section of each of the 10 ternaries of the Al-Cu-Mg-Si-Zn system from the other 9 sections. To increase the prediction accuracy, we introduced new descriptors generated from the thermodynamic properties of the elements and CALPHAD extrapolations from lower-order systems. Using the random forest method, the presence of single-, two-, and three-phase domains was predicted with an average accuracy of 84% across all 10 considered sections with a standard deviation of 11%. The proposed approach represents a promising tool for assisting the investigator in developing new materials and determining phase equilibria efficiently.

cond-mat.mtrl-sci↗

arXiv2018

Efficient Construction Method for Phase Diagrams Using Uncertainty Sampling

We develop a method to efficiently construct phase diagrams using machine learning. Uncertainty sampling (US) in active learning is utilized to intensively sample around phase boundaries. Here, we demonstrate constructions of three known experimental phase diagrams by the US approach. Compared with random sampling, the US approach decreases the number of sampling points to about 20%. In particular, the reduction rate is pronounced in more complicated phase diagrams. Furthermore, we show that using the US approach, undetected new phase can be rapidly found, and smaller number of initial sampling points are sufficient. Thus, we conclude that the US approach is useful to construct complicated phase diagrams from scratch and will be an essential tool in materials science.

cond-mat.mtrl-sci↗

arXiv2018

Population-based de novo molecule generation, using grammatical evolution

Automatic design with machine learning and molecular simulations has shown a remarkable ability to generate new and promising drug candidates. Current models, however, still have problems in simulation concurrency and molecular diversity. Most methods generate one molecule at a time and do not allow multiple simulators to run simultaneously. Additionally, better molecular diversity could boost the success rate in the subsequent drug discovery process. We propose a new population-based approach using grammatical evolution named ChemGE. In our method, a large population of molecules are updated concurrently and evaluated by multiple simulators in parallel. In docking experiments with thymidine kinase, ChemGE succeeded in generating hundreds of high-affinity molecules whose diversity is better than that of known inding molecules in DUD-E.

physics.chem-ph↗

arXiv2017

ChemTS: An Efficient Python Library for de novo Molecular Generation

Automatic design of organic materials requires black-box optimization in a vast chemical space. In conventional molecular design algorithms, a molecule is built as a combination of predetermined fragments. Recently, deep neural network models such as variational auto encoders (VAEs) and recurrent neural networks (RNNs) are shown to be effective in de novo design of molecules without any predetermined fragments. This paper presents a novel python library ChemTS that explores the chemical space by combining Monte Carlo tree search (MCTS) and an RNN. In a benchmarking problem of optimizing the octanol-water partition coefficient and synthesizability, our algorithm showed superior efficiency in finding high-scoring molecules. ChemTS is available at https://github.com/tsudalab/ChemTS.

physics.chem-ph↗