Quantum dynamics evolution predicted by the long short-term memory network in the photosystem II reaction center

TL;DR

This paper demonstrates that a modified LSTM neural network can predict charge transport dynamics in the photosystem II reaction center over extended time scales, surpassing traditional quantum simulation methods in accuracy.

Contribution

The study introduces a novel application of LSTM networks with an error-threshold training method to predict quantum charge transport in photosynthesis, extending predictions beyond initial simulation times.

Findings

LSTM predictions differ by about 10^{-4} over long periods.

The model effectively captures underlying physics of charge transport.

Results suggest LSTM can complement quantum physical methods.

Abstract

Predicting future physical behavior from limited theoretical simulation data is an emerging research paradigm driven by the integration of artificial intelligence and quantum physics. In this work, charge transport (CT) behavior was predicted over extended time scales using a deep learning model-the long short-term memory (LSTM) network with an error-threshold training method-in the photosystem II reaction center (PSII-RC). Theoretical simulation data within 8 fs were used to train the modified LSTM network, yielding distinct predictions with differences on the order of $10^{-4}$ over prolonged periods compared to the training set collection time. The results highlight the potential of LSTM to uncover the underlying physics governing CT beyond conventional quantum physical methods. These findings warrant further investigation to fully explore the scope and efficacy of LSTM in advancing our understanding of photosynthesis at the molecular scale.