AUTHOR=Huang Dongyang , Fu Jiaxing , Yu Chenghao 

TITLE=Machine learning discovery of the dielectric properties of strontium-containing condensed matter

JOURNAL=Frontiers in Physics

VOLUME=Volume 13 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/physics/articles/10.3389/fphy.2025.1599182

DOI=10.3389/fphy.2025.1599182

ISSN=2296-424X

ABSTRACT=The dielectric constant is one of the most important physical properties of dielectrics. The pursuit of materials with high dielectric constants has long been a central focus in both fundamental research and practical applications. Traditional theoretical predictions or first-principles calculations of dielectric constants are often challenging and require significant time and computational resources. Machine learning techniques can establish models that link composition and properties, facilitating the study of dielectric properties in condensed matter and enhancing the efficiency of discovering new dielectrics. Strontium-containing dielectrics constitute a diverse class of materials, some of which exhibit exceptionally high dielectric constants, thereby showing great potential for practical applications. In this work, machine learning models were successfully developed to capture the relationship between composition and dielectric properties of strontium-containing dielectrics using different algorithms, with hyperparameter optimization performed via grid search. The optimal model achieved a correlation coefficient of 0.868 and demonstrated a certain degree of generalization ability on the test set. This model serves as a valuable reference and guide, improving the efficiency of dielectric material selection and the discovery of novel high-performance dielectrics.