MS Teams
Speaker
William Gilpin | Harvard Quantitative Biology Initiative
Title
Inferring chaos and emergent low-dimensionality in living dynamical systems
Abstract
Dynamical systems theory provides a rich set of tools for inferring underlying mathematical structure from partial observations of complex systems, yet translating these insights to real-world biological datasets remains challenging. In this talk, I will overview my recent work at the intersection of nonlinear dynamics, fluid mechanics, and biology. I will first focus on my recent work developing physics-informed machine learning algorithms that extract dynamical models directly from raw experimental data. I will present a general technique for discovering strange attractors within diverse biological time series, including gene expression, patient electrocardiograms, animal trackers, and neural spiking. Next, I will describe my work on biological fluid dynamics, and the discovery of a beautiful vortex array created as many invertebrates swim—which enables a novel feeding strategy based on chaotic mixing of the local microenvironment. I will relate this work to broader questions at the intersection of nonlinear dynamics and organismal behavior. I will discuss how these insights open up several exciting new avenues at the intersection of dynamical systems theory, systems biology, and machine learning.