High-Throughput Virtual Screening of DHA/VHF

Code for the article: High throughput virtual screening of 230 billion molecular solar heat battery candidates. The code is divided into four directories, each performing one task.

1. run_sqm_calculations: contains code to compute SQM barrier heights and storage densities automatically.
2. train_ml_models: includes two notebooks to train ML models:
   • Train linear regression to predict storage densities for all 230 billion molecules.
   • Train LightGBM models to predict storage densities/barriers for 420 million molecules.
3. run_all_230b: is the Rust code used to run predictions on all 230 billion molecules, using the linear regression model.
4. run_ga_test: is a notebook used to run the Genetic Algorithm (GA) test.