Program Summary
Complex, nonlinear, multiscale dynamical systems are ubiquitous. Examples include weather, fluids, materials, biological systems, communication networks, and social systems. These systems often evolve to a critical state built up from a series of irreversible and unexpected events, which severely limits development and implementation of mathematical models to accurately predict formation and evolution of patterns in such systems.
The Models, Dynamics and Learning (MoDyL) aims to build rigorous data-driven models for non-equilibrium dynamics to address this challenge, leveraging existing data to enable robust prediction in complex systems. Collaboration among researchers from disciplines such as dynamical system theory, computational topology, statistics, spectral analysis, as well as domain experts in the various application problems is critical to address such a complex challenge. MoDyL will bring disparate researchers together to develop fundamental mathematics and computational algorithms for extracting models from dynamic data sets.