Viktor Stein

From May 2025 onward, I will be a Postdoc in the Applied Numerical Analysis Group in the Department of Mathematics at Technical University of Munich supervised by Massimo Fornasier. I worked on my PhD thesis under the supervision of Gabriele Steidl at TU Berlin in the Applied Mathematics Group at TU Berlin, and will defend my thesis in late June 2025.

My research develops rigorous analytical foundations and practical algorithms for particle-based sampling methods, aiming to (i) understand properties of gradient flows in probability spaces and (ii) design fast, efficient, and stable algorithms.

Developing methods that are both theoretically grounded and practically robust enables safer, faster, and more interpretable inference methods in modern applications.

Research interests

• Metric gradient flows in the Wasserstein geometry, the Fisher-Rao geometry, and kernelized variants
• Kinetic extensions thereof, like accelerated gradient flows
• Particle methods for generative modeling
• designing and investigating the properties of suitable loss functionals, combining e.g., optimal transport, f-divergences, and kernel distances.
• infinite-dimensional geometry