Prediction of complex turbulent dynamics in liquid crystals with optical reservoir computing

摘要: Reservoir computing (RC) is a powerful machine learning framework derived from Recurrent Neural Networks that maps input signals into higher dimensional computational spaces through the dynamics of a fixed, non-linear system called a reservoir. It has been shown to be a best-in-class algorithm for analyzing dynamical systems based solely on observed timeseries data [1]. Various optical architectures have been developed to run RC, reducing its working time and improving its performance [2-4]. Recent studies have demonstrated that optical RC setups can predict spatiotemporal chaotic datasets obtained from the Kuramoto-Sivashinsky benchmark equation [5, 6].In this work, for the first time, we use an optical setup to analyze real-world spatiotemporal chaos generated inside a liquid crystal (LC) cell due to its turbulent director flows. To this end, we applied an AC voltage with a specific frequency to a …