1. Charge-transport forecasted via deep learning in the photosystem II reaction center
- Author
-
Zhao, Zi-Ran, Zhao, Shun-Cai, and Huang, Yi-Meng
- Subjects
Physics - Chemical Physics ,Quantum Physics - Abstract
Predicting future physical behavior through the limited theoretical simulation data available is an emerging research paradigm resulted by the integration of artificial intelligence technology and quantum physics. In this work, the charge-transport(CT) behavior was forecasted over a long time by a deep learning model, the long short-term memory (LSTM) network with error threshold training method in the photosynthesis II reaction center (PSII-RC). The theoretical simulation data within 8 fs was fed to the modified LSTM network for training, which brings out a distinct prediction with difference of $10^{-4}$ orders of magnitude over a long time period compared to the collection time for training sets. The results indicate the potential of employing LSTM to reveal the physics governing CT in addition to quantum physical methods. The implications of this work warrant further investigation to fully elucidate the scope and efficacy of LSTM for advancing our understanding of photosynthesis at the molecular scale., Comment: 8 pages, 4 figures
- Published
- 2024