Abstract
Dynamic Mode Decomposition (DMD) is a data-driven method for extracting spatial structures and their temporal evolution from sequential measurements. Its connection with the Koopman operator provides a framework for interpreting nonlinear dynamics through the evolution of observable functions. This paper reviews the mathematical foundations of DMD and outlines several developments: exact and improved formulations, online algorithms for streaming data, higher-order DMD for systems with temporal memory, and DMD with control for input-driven dynamics. The methods are discussed in terms of their computational structure, assumptions, and potential applications..