The Dynamic Mode Decomposition - A Data-Driven Algorithm for Complex Systems Analysis

Explore the Dynamic Mode Decomposition (DMD), a powerful data-driven modeling technique for analyzing complex systems, in this comprehensive lecture by Nathan Kutz at the Alan Turing Institute. Delve into the underlying theory of DMD and its ability to reveal coherent spatiotemporal structures, produce reconstructions, and generate future-state predictions from data. Learn about the method's linear algebra-based formulation and various optimizations and extensions that enhance its practicality for real-world data analysis. Discover the PyDMD Python package, which implements DMD and its major variants, designed to handle noisy, multiscale, parameterized, high-dimensional, and strongly nonlinear dynamics. Gain insights into the features of PyDMD, practical usage tips, information for developers, and coding examples. This 1 hour and 29 minute talk provides a thorough overview of DMD's applications across multiple scientific disciplines and its potential as a leading method for equation-free system analysis.