Multiscale Modeling and Simulation Platform for Materials and Life Sciences
J-OCTA
Multiscale Modeling and Simulation Platform for Materials and Life Sciences
J-OCTA
Estimation of the χ parameter by machine learning
The χ parameter is an important index in Flory-Huggins theory representing interactions between polymers and solvents. In this case, a machine learning model was constructed using χ value data for 263 compound pairs. Molecular structures were obtained in SMILES format, and AutoCorr3D descriptors were calculated. Features were created by mixing descriptors, and XGBoost was used to train a χ parameter prediction model.
Use Cases Highlights
- Creation of features from descriptors of two molecules and learning their relationship with interaction parameter χ
Creation of features from descriptors of two molecules and learning their relationship with interaction parameter χ
Results of χ parameter learning are shown. For the trained model, prediction accuracy for the training set was R² = 0.937, and for the test set R² = 0.778.
Prediction of χ parameter by machine learning
Details of analysis
Inquiries Regarding Products
Have questions about product implementation? Contact us today.
Related Information
Solutions
Materials Simulation and Materials Informatics (Automotive)
Materials development through the integration of simulation and data science (Materials DX)
Materials Simulation and Materials Informatics (Mobility)
Materials development through the integration of simulation and data science (Materials DX)
Materials Simulation and Materials Informatics (Electrical Equipment/Precision Equipment/Semiconductors/Heavy Electrical Equipment)
Materials development through the integration of simulation and data science (Materials DX)
