Model-Free Prediction of Chaotic Systems Using High Efficient Next-generation Reservoir Computing
This work addresses the challenge of model-free prediction for chaotic systems, which has applications in fields like weather forecasting and fluid dynamics, but it appears incremental as it builds on reservoir computing with specific efficiency improvements.
The authors tackled the problem of predicting chaotic systems from past observations without a model, achieving superior prediction length, reduced training data, and lower computational cost compared to existing reservoir computing methods, as demonstrated on Lorenz and Kuramoto-Sivashinsky equations.
To predict the future evolution of dynamical systems purely from observations of the past data is of great potential application. In this work, a new formulated paradigm of reservoir computing is proposed for achieving model-free predication for both low-dimensional and very large spatiotemporal chaotic systems. Compared with traditional reservoir computing models, it is more efficient in terms of predication length, training data set required and computational expense. By taking the Lorenz and Kuramoto-Sivashinsky equations as two classical examples of dynamical systems, numerical simulations are conducted, and the results show our model excels at predication tasks than the latest reservoir computing methods.