Dynamic Mode Decomposition (DMD) is a feature detection algorithm that allows to identify recurring patterns from a temporally resolved data sequence. The technique is compatible with both data either generated from numerical computations and or obtained from experimental measurements.
The two major advantages of DMD-based analysis are:
I. It enforces a data-driven paradigm, which makes innecessary the knowledge of an underlying mathematical model.
II. It provides not only the temporal fate of the patterns identified, but also their spatial support.
Very often, the patterns identified using DMD allow to isolate physical phenomena that are relevant for the process generating the data. In this manner, the DMD eases the interpretation of the acquired data and can even assist in the generation of Reduced Order Models of the process being investigated.