Supplementary files for our work in the paper "Lattice thermal conductivity: An accelerated discovery guided by machine learning"
This reposirtry contains three files:
1- Excel, CSV files contains two folders: 1.Training folder contains the intial dataset that was used for the training process containing the temperature and LTC from previous literature (Check supplementrary materials docx file for references) and a Feature excel file contains the features that were obtained using Magpie for 119 compound 2.Prediction folder contains the icsd-prediction.csv file which is the result of the screening process that have been done using our RF-based model, and a prototype prediction folder containing the prediction on prototype structures for 11,488 compounds generated through combination. (Please note that the A (in ABX and A_1BX) refers to the first group in the periodic table and A_1 refers to the second group)
2- DFT files for X = Cs2SnI6 and SrS that includes the X.in file, the header and the displacement files obtained from PHONO3PY (for further information please visit the phono3py documentation page: https://phonopy.github.io/phono3py/)
3- Codes used in this work this includes: 1.ML_Code.py that has the code for the machine learning using RF (Sklearn, for more info please visit: https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html) and 2.screening_code that allows us to generate the files necessary for feature generation using the protoype strucutres and elements (the features are generated by Magpie software, for more info lease visit: https://bitbucket.org/wolverton/magpie/src/master/) 3.prototype_Poscar folder that includes the POSCAR blueprint for 4 prototype AxByXn Structures (Please refer to Fig. 6b in our paper for further info)