Learning ergodic averages in chaotic systems

Francisco Huhn, Luca Magri

We propose a physics-informed machine learning method to predict the time average of a chaotic attractor. The method is based on the hybrid echo state network (hESN). We assume tha…