A wide range of real-world applications like instance, the robust operation of energy
networks or the training of large neural networks, lead to optimization problems with non-smooth in
the target functions and or the constraints. Such non-smoothness significantly challenges the analysis
of problems. Even nowadays, there hardly exist off-the-shelf solution algorithms is mainly due to the
lack of computationally tractable optimality and stationarity conditions, respectively.
In this minisymposium, four speakers from four institutions present their approach to tackle the
non-smoothness in their context. This ranges from the optimal control of crystal growth to the solution
of linear complementarity problems.
Mini symposium organizers:
Luise Blank (Universität Regensburg)
Andrea Walther (Humboldt-Universität zu Berlin)
Session 4. Room A6, Thursday 11:30-13:30.
Chair: Andrea Walther (Humboldt-Universität zu Berlin)
Speakers:
Luise Blank (Universität Regensburg) Optimal control of anisotropic Allen-Cahn equations
Anna Lentz (Universität Würzburg) Spatially sparse optimization problems in fractional order Sobolev spaces
Bennet Gebken (Universität Paderborn) First- and second-order models for nonsmooth functions based on derivative sampling
Adrian Schmidt (Humboldt-Universität zu Berlin) On solving complementarity-constrained problems