DFG-Sonderforschungsbereiche (SFB)
Disciplines
Mathematics (80%); Physics, Astronomy (20%)
Keywords
Maxwell,
Landau-Lifshitz-Gilbert,
Time-Stepping,
Adaptive Mesh-Refinement
Abstract
In recent groundbreaking experiments, physicists managed to manipulate magnetic
materials using ultrashort light pulses. This can be extremely useful for new data storage
systems or magnetic sensors.
This project aims to develop the mathematical theory and computational algorithms to
simulate and predict this new technology.
To that end, cutting edge mathematical theory for time dependent partial differential
equations has to be developed. Those equations describe the behavior of magnets on a
microscopic level and are notoriously hard to solve. The project will develop so-called
adaptive approximation methods for those equations. Such methods automatically focus the
bulk of the computational resources to hard subproblems, while not wasting them on easy
parts of the problem. This will be supported by neural network models which are well suited
to approximate solutions of the problem.