Skip to content
SDGs Center UDINUS

Evidence record · UDINUS · SDGs

Predicting Lattice Constants in ABX3 Perovskites Using Quantum Machine Learning

Rektorat UDINUS 2025 Published 2026-10-30

Evidence record · UDINUS · SDGs

Unit
Rektorat UDINUS
Period
2025
Goals
Record
2842dfc3-3ac0-4ea7-a094-1b0157dc144f
Published
2026-10-30

Universitas Dian Nuswantoro, through researchers Akrom, Muhamad; Rustad, Supriadi; Andono, Pulung Nurtantio; Shidik, Guruh Fajar; Dipojono, Hermawan Kresno; Maezono, Ryo; and Kasai, Hideaki, conducted a study on predicting lattice constants in perovskite materials with the chemical structure ABX3. The activity addressed the importance of accurately predicting lattice constants for optimizing the performance of materials that have attracted significant attention because of their versatile applications in optoelectronics, solar cells, and catalysis.

The study compared quantum machine learning with a classical machine learning approach for lattice-constant prediction. In particular, the activity examined Quantum Support Vector Regression, or QSVR, and compared it with XGBoost, or XGB. The prediction task used an ABX3 dataset comprising 122 compounds. The comparison provided a basis for assessing the performance of quantum and classical approaches in materials informatics.

The study reported that the classical XGB model outperformed the other approaches, with R2 of 0.99, RMSE of 0.009, and MAE of 0.007. QSVR demonstrated competitive accuracy, with R2 of 0.97, RMSE of 0.064, and MAE of 0.062. The activity also examined quantum feature map optimization. This optimization improved the ability of the QSVR model to process high-dimensional data, although its performance was slightly hindered by coherence limitations.

By examining lattice-constant prediction for ABX3 materials, the activity contributes to a comparative perspective on classical and quantum approaches. The findings identify the potential of quantum-enhanced methods for material property prediction while also showing the reported strength of the classical XGB model. The study further points toward hybrid methods that combine the computational efficiency of quantum algorithms with the robustness of classical models for materials informatics. The activity is connected to solar-cell materials and computational research on their properties. This work frames UDINUS research within the study of computational approaches for materials associated with solar-cell applications.

Reference https://www.scopus.com/pages/publications/105001831151?origin=resultslist