Surrogate Dynamic

Surrogate dynamic modeling aims to create computationally efficient approximations of complex dynamical systems, reducing the need for expensive simulations. Current research focuses on improving the accuracy and generalization capabilities of these surrogates, employing diverse methods such as physics-informed neural networks (PINNs), graph neural networks (GNNs), and sparse identification of nonlinear dynamics (SINDy), often combined with techniques like stochastic collocation and large language models for enhanced interpretability and control. These advancements are crucial for accelerating scientific discovery and engineering design across various fields, from fluid dynamics and robotics to energy systems and materials science, by enabling faster and more accessible simulations.