The research seminar will be held on Thursdays at FU Berlin (Arnimallee 3-7) during Wintersemester 2023/24. The seminar brings together experts on numerical analysis, applied mathematics, statistics and stochastics with particular focus on applications to stochastic and deterministic partial differential equations.

Wintersemester 2024/25

• Thursday 16.01.25, 1 pm

Georg Heinze
Gradient flows on metric graphs with reservoirs: Microscopic derivation and multiscale limits

I will discuss evolution equations on metric graphs with reservoirs, that is graphs where a one-dimensional interval associated to each edge and, in addition, the vertices are able to store and exchange mass with these intervals.Focusing on the case where the dynamics are driven by an entropy functional defined both on the metric edges and vertices, we provide a rigorous understanding of such coupled systems of ordinary and partial differential equations as(generalized) gradient flows in continuity equation format. Approximating the edges by a sequence of vertices, which yields a fully discrete system, we are able to establish existence of solutions in this formalism. Furthermore, we study several scaling limits using the recently developed framework of EDP convergence with embeddings to rigorously show convergence to gradient flows on reduced metric and combinatorial graphs.  Finally, numerical studies confirm our theoretical findings and provide additional insights into the dynamics under rescaling. This is joint work with Jan-Frederik Pietschmann and André Schlichting.

• Thursday 21.11.24, 1 pm

Emil Engström
An abstract approach to the Robin-Robin method

The Robin-Robin method is a common domain decomposition method for numerically approximating solutions to partial differential equations in parallel. While there are convergence proof for specific equations and domains, there are few general results. We have therefore developed an abstract approach to the Robin-Robin method, enabling the treatment of linear and nonlinear elliptic and parabolic equations on Lipschitz domains within one framework. Furthermore, the framework can be extended to initial-value problems on moving domains. This extension is an ongoing joint work with Ana Djurdjevac and Amal Alphonse.
room 108, Arnimallee 6

• Thursday 14.11.24, 1 pm

Chris Oldnall
Estimating causal effects in the instrumental variable framework, without ex-ante knowledge of valid instruments

In the field of Biology, the use of instrumental variables has taken precedence for estimating causal relationships between exposures and outcomes, in the form of Mendelian randomisation. This framework assumes the instrument is a genetic variant, which is essentially inherited at birth and remains unchanged throughout one's life. The framework however also stipulates that the instrument is only valid if it acts on the outcome only through the exposure of interest — a breach of this situation makes the estimation null, and in the biological sphere is referred to as pleiotropy. Many attempts have been made to identify ‘pleiotropic’ instruments and remove them to perform valid estimation, however with no mathematically testable criteria, this procedure is not robust. This work instead looks at the approach by Sun et al.¹ in developing a solution which leverages potentially invalid instruments with valid ones in the form of G-estimators with unconditional moment conditions. This approach proves promising, but is limited by scalability leading to trade-offs between biological accuracy and computational ability. This talk demonstrates the severity of pleiotropic rich systems, and will discuss the computational limitations and trade-offs that one has to make to employ G-estimation in the Mendelian randomisation application.

1. B Sun, Z Liu, E J Tchetgen Tchetgen, Semiparametric efficient G-estimation with invalid instrumental variables, Biometrika, Volume 110, Issue 4, December 2023, Pages 953–971, https://doi.org/10.1093/biomet/asad011
room 108, Arnimallee 6

• Thursday 07.11.24, 1 pm

Alix Leroy
Adaptive stepsize algorithms for Langevin dynamics
The work focuses on weak approximation of stochastic differential equations and develops a method of computing solutions of Langevin dynamics using variable stepsize. The method assumes a knowledge of the problem allowing to establish a good monitor function which locates points of rapid change in solutions of stochastic differential equations. Using time-transformation we show that it is possible to integrate a rescaled system using fixed stepsize numerical discretization effectively placing more timesteps where needed.
room 108, Arnimallee 6

Wednesday 30.10.24, 10:00 Jonas Latz
Data-driven approximation of Koopman operators and generators: Convergence rates and error bounds(joint work with Liam Llamazares-Elias, Samir Llamazares-Elias, and Stefan Klus)

• Tuesday 22.10.24, 14:00 Paolo Villani
  Adaptive training of Gaussian Process based surrogates for Bayesian parameter identification.

Sommersemester 2024

• 16.5.24 Robert Gruhlke
  QMC meets Optimal sampling
  room 210, Arnimallee 6
• 23.5.24 Ilja Klebanov
  

  Sampling in Unit Time with Kernel Fisher-Rao Flow
  Paper von Aimee Maurais und Youssef Marzouk
  room 210, Arnimallee 6

• 30.5.24 Konrad Mundinger (ZIB)
  Learning Operators via Hypernetworks
  room 126, Arnimallee 6

• 6.6.24 Andre Zepernick
  Quasi-Monte Carlo Methods for PDEs on Random Domains
  room 126, Arnimallee 6

• 13.6.24 Matei Hanu
  Optimisation in Bayesian experimental design
  room 126, Arnimallee 6
  
• 20.6.24 Deborah Hendrych (ZIB)
  Solving the Optimal Experiment Design Problem with mixed-integer convex methods
• room 126, Arnimallee 6
  
• 11.7.24
  
  

Wintersemester 2023/24

• Tue 17.10. Angelina Senchukova (LUT University), Edge-preserving inversion with heavy-tailed Bayesian neural networks priors, room A6/108, 3:15 pm
• Thu 19.10. CRC Colloquium, 2:00-6:00 pm
• Thu 26.10. Sven Wang (HU Berlin), On polynomial-time mixing for high-dimensional MCMC in inverse problems, room A6/108, 1:00 pm
• Thu 02.11. Ruben Harris (FU Berlin), Ensemble Kalman Inversion for time-dependent forward operators, room A6/210, 12:00pm
• Thu 02.11. Ali El-Rahal (FU Berlin), Effiziente Synergien durch integrierte Prozessoptimierung – Bedarfsgerechter Einsatz von Produktionskapazitäten unter Berücksichtigung der partiellen Produktionssysteme , room TBA, 1:00pm
• Thu 09.11. No seminar talk
• Thu 16.11. Lorenz Richter (Zuse Institute Berlin), An optimal control perspective on diffusion-based generative modeling leading to robust numerical methods, room A6/108, 1:00 pm
• Thu 23.11. CRC Colloquium, 2:00-6:00 pm
• Thu 30.11 André-Alexander Zepernick (FU Berlin), An introduction to TorchPhysics: Deep Learning for partial differential equations, A3/115, 1:00 pm
• Fri 01.12. Jeremie Houssineau (University of Warwick), Ensemble Kalman filtering for epistemic uncertainty, room A6 108/109, 10:00 am
• Thu 07.12. TBA
• Thu 14.12. CRC Colloquium, 2:00-6:00 pm
• Thu 21.12. Hanno Gottschalk (TU Berlin), From Probabilistic Models of Mechanical Failure to Multi-Objective Shape Optimization, room A3/115, 1:00 pm
• Thu 11.01. Vicky Holfeld (ITWM), TBA, room A3/115, 1:00 pm
• Thu 18.01. No seminar talk
• Thu 25.01. CRC Colloquium, 2:00-6:00 pm
• Thu 01.02. Caroline Geiersbach (WIAS), PDE-Constrained Optimization Problems with Probabilistic State Constraints, room A6/108 12:00 pm
• Thu 01.02. Daniel Walter (HU Berlin), Towards optimal sensor placement for inverse problems in spaces of measures, room A6/108 1:00 pm
• Thu 08.02. Martin Eigel (WIAS), TBA, room A3/115, 1:00 pm
• Thu 15.02. CRC Colloquium, 2:00-6:00 pm

Older talks

Contact

Prof. Dr. Claudia Schillings

c.schillings@fu-berlin.de

Arnimallee 6, room 213 Consulting hours: By appointment