Thesis Defense: Approximation of Large Stiff Acausal Models

Title: Approximation of Large Stiff Acausal Models

Abstract: Simulations drive mission-critical decision making in many fields, but are prone to computational intractability, which severely limits an engineer's productivity whilst designing practical systems. These issues are alleviated by the use of approximate models called surrogates, which match the full system to high fidelity whilst being feasible to simulate. In this research, we propose a method to generate surrogates of dynamical systems with multiple widely separated timescales, called the Continuous Time Echo State Networks. We also study deployment of such systems to accelerate common tasks such as global optimization and global sensitivity analysis through practical applications from building simulation, quantitative systems pharmacology and electrical circuit design. Lastly, we examine how to data-efficiently sample a model's input space to produce a surrogate with a desired error performance.

Committee: Prof. Alan Edelman, Prof. Steven Johnson, Prof. Gilbert Strang, Dr. Chris Rackauckas, Dr. Chris Laughman