Utilizing Koopman Theory and Extended DMD to find Linear Representations of Nonlinear Systems

Faculty Sponsor: Tsampikos Kottos

Live Poster Session: Zoom Link Goes Here

Richik Acharya

I am a Physics and Math Double major researching in the Kottos Lab. I am from Pleasantville, NY. Outside of research, I love to play soccer, hike, and rock climb. I also like to sew and create clothes.

Abstract: My project looks to leverage large data models and neural networks to find linear representations of nonlinear systems, which are notoriously ill-behaved. Although this linear representation allows us to predict the state of out system with perfect accuracy, it comes with a trade off: the fact that this linear representation in infinite-dimensional. Even so, the eigen-modes of such a linear operator can provide invaluable information even with a finite dimensional truncation. These modes can give us insight into the conserved quantities of the system, the topology of the phase space, and other properties obscured by the nonlinearity of the system.

QAC-Poster.pptx-1-2