Machine Learning, Dynamical Systems and Control

Lecture 1

 

Stacks Image 8

 

[ Part 1 ] [ Part 2 ]

 
DYNAMIC MODE DECOMPOSITION: This lecture provides an introduction to the Dynamic Mode Decomposition (DMD). The focus is on approximating a nonlinear dynamical system with a linear system.

 

MATLAB CODE

Lecture 2

 

Stacks Image 23

 

[ view ]

 
KOOPMAN THEORY: This lecture generalizes the DMD method to a function of the state-space, thus potentially providing a coordinate system that is intrinsically linear.

 

Lecture 3

 

Stacks Image 38

 

[ view ]

 
TIME DELAY EMBEDDINGS: This lecture generalizes the Koopman/DMD method to a function of the state-space created by time-delay embedding of the dynamical trajectories.

 

MATLAB CODE

 

 

KEY REFERENCES AND SUPPLEMENTARY VIDEOS

 

Stacks Image 60

Koopman observable subspaces and finite linear representations of nonlinear dynamical systems for control

 

This video highlights the recent innovation of Koopman analysis for representing nonlinear systems and control.
[Part 1], [Part 2], [Part 3]

 

 

Stacks Image 75

Dynamic Mode Decomposition with Control

 

This video highlights the concepts of Dynamic Mode Decomposition which includes actuation and control.
[View]

 

 

Stacks Image 90

Koopman theory for partial differential equations

 

This video highlights the concepts of Koopman theory and how they can be used for partial differential equations.
[View]

 

 

Stacks Image 105

Multi-Resolution Dynamic Mode Decomposition

 

This video highlights the recent innovation of multi-resolution analysis applied to dynamic mode decomposition.
[View]

 

 

Stacks Image 120

Generalizing Koopman Theory to Allow for Inputs and Control

 

This video highlights the new innovations around Koopman theory and data-driven control strategies.
[View]