PyDMD: A Python Package for Dynamic Mode Decomposition (DMD)
This video introduces the dynamic mode decomposition (DMD) algorithm and demonstrates how scientists and engineers with a background in Python can get started with the PyDMD Python package for modeling snapshot data.
We provide motivation for data-driven mathematical modeling techniques and review the theoretical underpinnings of the DMD algorithm. We then introduce the PyDMD package as a practical tool for applying DMD and demonstrate its use on a noisy synthetic data set. In the process, we build intuition regarding the output of DMD and highlight some of the major features that PyDMD has to offer. By the end of this video, viewers will be able to (1) build and interact with basic PyDMD models, (2) interpret the results of DMD, and (3) apply DMD to snapshot data in order to obtain system diagnostics, reconstructions, predictions, and governing equations.
Paper: https://www.jmlr.org/papers/v25/24-07...
arXiv: https://arxiv.org/abs/2402.07463
Chapters:
00:00 -- Intro and video overview
01:45 -- Why do we build mathematical models?
04:52 -- Dynamic Mode Decomposition (DMD)
11:18 -- Introduction to PyDMD
Coding Demonstration:
13:30 -- PyDMD installation
14:46 -- Building the toy data set
24:13 -- Building and fitting basic PyDMD models
28:45 -- Accessing and understanding core model attributes
40:38 -- Data reconstruction
44:25 -- Model summary plotting
50:28 -- Building more complex PyDMD models