Multi-Time-Scale Time Encoding for CNN Prediction of Fenna-Matthews-Olson Energy-Transfer Dynamics.

Machine learning simulations of open quantum dynamics often rely on recursive predictors that accumulate error. We develop nonrecursive convolutional neural networks (CNNs) that map system parameters and a redundant time encoding directly to excitati…

Preview only. Read the full abstract at the source