Prediction of chimera in coupled map networks by means of deep learning

Chimera states are Spatio-temporal patterns in coupled oscillator arrays, in which incoherent domains coexist with coherent ones. To characterize chimeras, however, is a nontrivial problem since it is difficult to distinguish between coherent domains and incoherent domains. A useful tool for this task is machine learning, in particular deep learning techniques like reservoir computing and multilayer perceptrons. In this work we use these quantifiers in order to identify chimera states in logistic map lattices with non-local coupling. We compare our results from machine learning techniques with more conventional characterizations, such as Lyapunov exponents and a local order parameter. (c) 2022 Elsevier B.V. All rights reserved.