Abstract
Digital twins have attracted a great deal of recent attention from a widerange of fields. A basic requirement for digital twins of nonlinear dynamicalsystems is the ability to generate the system evolution and predict potentiallycatastrophic emergent behaviors so as to providing early warnings. The digitaltwin can then be used for system "health" monitoring in real time and forpredictive problem solving. In particular, if the digital twin forecasts apossible system collapse in the future due to parameter drifting as caused byenvironmental changes or perturbations, an optimal control strategy can bedevised and executed as early intervention to prevent the collapse. Twoapproaches exist for constructing digital twins of nonlinear dynamical systems:sparse optimization and machine learning. The basics of these two approachesare described and their advantages and caveats are discussed.