"Predicting the future of excitation energy transfer in light-harvesting complex with artificial intelligence-based quantum dynamics"
https://www.nature.com/articles/s41467-022-29621-w
-
Because of the large size of training data, we couldn't uplaod it here, however you can download quantum_HEOM package from https://github.com/jwa7/quantum_HEOM, and generate the training data. We provide the script LTLME.py
-
The farthest_point.py samples the trajectories based on farthest point sampling (we just need to do it for one case, initial exciation on site-1 or site-6)
-
We choose 500 trajectories for site-1 and 500 trajectories from site-6, in total 1000 trajectories as a training set
-
The validation set is the 100 trajectories from site-1 + 100 trajectories from site-6
-
use prep_input.py to prepare your input files accordingly
-
use train_CNN.py to train the CNN model
-
Run run_dyn.py to predict EET dynamics for test trajectories. We have provided a trained model "trained_ML_model.hdf5" at https://doi.org/10.6084/m9.figshare.16922446 (file size exceeds 25MB, GitHub limit)
-
The respective parameters of test trajectories are in temperature.npy, gamma.npy, initial_site.npy and lambda.npy.
-
The search_optim_eet.py predict population of site-3 at t=0.5ps for different combinations of gamma, lambda and temperature.