Steffen Werner

Computational modeling and simulations enable the study of natural phenomena in areas in which experimental studies are expensive, dangerous or even impossible. Examples include structural and acoustical dynamics, fluid dynamics, chemical reactors and power networks. Due to the need for high accuracy, for example, in the design of digital twins, the resulting models are generally rather complex with millions of degrees of freedom, making their use in computations a formidable challenge.

Professor Werner's research revolves around the theoretical analysis and the development of computational frameworks for the construction of cheap-to-evaluate approximations for complex dynamical systems using techniques from numerical linear algebra and scientific machine learning. Considered research projects range from the design of structure-preserving model order reduction methods, which allow the inclusion of nonlinear and differential structures into surrogate models, to the construction of reduced-order models purely from data, also known as data-driven modeling, to the design of reliable and robust controllers for dynamical systems.

Beside these topics, Professor Werner studies the use and solvability of matrix equations such as Lyapunov and Riccati equations for applications in systems theory. Some of his major concerns are the availability of mathematical research software and the reproducability of numerical experiments.