Energies, structures descriptors of two-polaron configurations in defected Li4Ti5O12

This dataset contains structures, distance descriptors and relative energies of two-polaron configurations in two different defected supercells of Li₄Ti₅O₁₂. Additionally, it contains jupyter notebooks with XGBoost models used to predict the energies of polarons in each supercell from the given descriptors.

Data Structure

The main directory contains the following two subdirectories:

• lto_cell_A: This directory contains two subdirectories, opt_pol_structures and Notebook_ML_Model.
  • opt_pol_structures: It contains directories for the DFT+U optimized geometry files, including occupation matrices for different polaronic configurations for LTO_Cell_A, used for training and testing the machine learning model.
  • Notebook_ML_Model: It contains the following files: 
    • all_calculated_3.dat: This file contains the DFT calculated relative energies and features corresponding to different polaronic configurations for LTO_Cell_A.
    • polaron_data: This file contains features for all possible 780 polaronic configurations for the LTO simulation cell A.
    • XGB_reg.ipynb: This Jupyter notebook is used for training and evaluating the XGBoost regression model. It includes data preprocessing, model training, and evaluation steps, as well as visualizations of the results.
    • lto_model.json: This file contains the saved XGBoost model in JSON format.
    • training.dat and test.dat: These files contain the training and testing datasets used for the model. 
• lto_cell_B: This directory also contains two subdirectories, opt_pol_structures and Notebook_ML_Model. 
  • opt_pol_structures: Similar to the one in lto_cell_A, it contains directories for the DFT+U optimized geometry files for LTO_Cell_B.
  • Notebook_ML_Model: It contains the following files: 
    • all_calculated.dat: This file contains the DFT+U calculated relative energies and features corresponding to different 196 polaronic configurations for LTO_Cell_B used for training and testing purposes.
    • all_feature.dat: This file contains features for all possible 74880 polaronic configurations for the LTO simulation cell B.
    • XGB_reg.ipynb: This Jupyter notebook is used for training and evaluating the XGBoost regression model for LTO_Cell_B. It includes data preprocessing, model training, and evaluation steps, as well as visualizations of the results.
    • lto_model.json: This file contains the saved XGBoost model in JSON format for LTO_Cell_B.
    • training.dat and test.dat: These files contain the training and testing datasets used for the model in LTO_Cell_B.
• lto_cell_C: This directory also contains two subdirectories, opt_pol_structures and Notebook_ML_Model.
  • opt_pol_structures: Similar to the one in lto_cell_A, it contains directories for the DFT+U optimized geometry files for LTO_Cell_C.
  • Notebook_ML_Model: It contains the following files: 
    • all_calculated.dat: This file contains the DFT+U calculated relative energies and features corresponding to different 188 polaronic configurations for LTO_Cell_C used for training and testing purposes. 
    • XGB_reg.ipynb: This Jupyter notebook is used for training and evaluating the XGBoost regression model for LTO_Cell_C. It includes data preprocessing, model training, and evaluation steps, as well as visualizations of the results.
    • training.dat and test.dat: These files contain the training and testing datasets used for the model in LTO_Cell_C.