An Entropic Framework for Sparse Network Reconstruction in Nonlinear Dynamical Systems
Faculty Mentor Name
Abd AlRahman Rasheed AlMomani
Format Preference
Poster
Abstract
Synchronization phenomena in networks of interacting dynamical systems arise in a wide range of scientific and engineering domains, including neuroscience, power systems, and multi-agent coordination. This project introduces an entropic framework for the reconstruction of sparse interaction networks governed by Kuramoto-type oscillator dynamics.
The central objective of the work is to address the inverse problem of network reconstruction: inferring unknown coupling structures from observed time-series data of oscillator phases. The reconstruction task is formulated as a nonlinear system identification problem that is linear in the unknown interaction weights. To solve this problem, an Entropic Regression framework is implemented, combining information-theoretic feature selection with projection-based estimation to identify sparse interaction structures and estimate coupling strengths.
The resulting framework provides a principled approach for inferring connectivity in complex dynamical systems and establishes a foundation for data-driven analysis of synchronization phenomena in settings where network structure is not directly observable.
An Entropic Framework for Sparse Network Reconstruction in Nonlinear Dynamical Systems
Synchronization phenomena in networks of interacting dynamical systems arise in a wide range of scientific and engineering domains, including neuroscience, power systems, and multi-agent coordination. This project introduces an entropic framework for the reconstruction of sparse interaction networks governed by Kuramoto-type oscillator dynamics.
The central objective of the work is to address the inverse problem of network reconstruction: inferring unknown coupling structures from observed time-series data of oscillator phases. The reconstruction task is formulated as a nonlinear system identification problem that is linear in the unknown interaction weights. To solve this problem, an Entropic Regression framework is implemented, combining information-theoretic feature selection with projection-based estimation to identify sparse interaction structures and estimate coupling strengths.
The resulting framework provides a principled approach for inferring connectivity in complex dynamical systems and establishes a foundation for data-driven analysis of synchronization phenomena in settings where network structure is not directly observable.