研究目的
To study the relationship between structure and properties of Pmn21-BxAl1-xNyP1-y compounds using machine learning algorithms based on first-principles calculations, and to establish prediction models for lattice constants and shear modulus.
研究成果
The study successfully establishes machine learning models for predicting lattice constants and shear modulus of Pmn21-BxAl1-xNyP1-y with strong generalization ability. The models show acceptable error ranges on test sets. The electronic properties analysis reveals that several structures have direct band gaps, making them suitable for optoelectronic applications.
研究不足
The study relies on first-principles calculations which are time-consuming and may not converge for some complex structures. The machine learning models are based on a limited dataset of 16 structures.
1：Experimental Design and Method Selection:
The study uses the DFT with the GGA-PBE in the CASTEP code for structural optimization and mechanical properties calculation. The BFGS minimization scheme is used in geometry optimization.
2：Sample Selection and Data Sources:
The study focuses on 16 structures of Pmn21-BxAl1-xNyP1-y, with their component ratios and lattice constants used as training sets for machine learning.
3：List of Experimental Equipment and Materials:
The CASTEP program and machine learning algorithms (LR, SVR, GBDT, Xgboost) are used.
4：Experimental Procedures and Operational Workflow:
The study involves calculating elastic constants, phonon dispersion spectra, and electronic properties using hybrid PBE0 functional. Machine learning models are trained and validated using 7-fold cross-validation.
5：Data Analysis Methods:
The MAPE is calculated for each model to evaluate their performance.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
