Zum Inhalt springen

RMKS Frankfurt 01.07.2022

Rhein-Main-Kolloquium

Datum: 01.07.2022

Zeit: 15:15–18:00 Uhr

15 :15 Uhr Dr. Julian Gerstenberg, Goethe-Universität Frankfurt Functional Representation Theorems for Exchangeable Laws Functional Representation theorems (FRTs) for exchangeable random objects exist for many types of data structures, for example sequences (de Finetti/Hewitt-Savage), partitions (Kingman), hierarchical structures, graphs or more general array-like data structures (Aldous-Hoover-Kallenberg). In this talk several known FRTs are presented and the language of category theory [vocabulary: cagegory, functor, natural transformation] is used to introduce an abstract concept of data structures, which allows for a unified formulation of many known FRTs. This leads to a conjecture about a "General FRT". No knowledge of category theory is assumed to follow this talk, the concepts will be motivated by statistical practise. This research is funded by the DFG project 502386356.

16:15 Uhr Kaffeepause

16:45 Uhr Dr. Jonathan Warren, University of Warwick At the edge of a cloud of Brownian particles The talk concerns a model for the motion of particles carried in a turbulent fluid in which a single particle moves according to an SDE of the form dX_t= \sigma dB_t + dW(t,X_t). Here, W is a Gaussian field, describing the environment, and B is an independent Brownian motion representing some additional diffusivity. We are interested in the behaviour at large times, but far from the origin. There, we find a transition which is analogous to that between weak and strong disorder for polymer models, and at the transition the stochastic heat equation appears.

Nummer

154

Referenten

Julian Gerstenberg, Goethe-Universität Frankfurt
Jonathan Warren, University of Warwick

Ort

Goethe-Universität Frankfurt, Campus Bockenheim, Raum 711 (groß)
Robert-Mayer-Str. 10, 7. Stock, Frankfurt am Main


Kooperationspartner

Technische Universität Darmstadt, Johannes Gutenberg-Universität Mainz

Für diese Veranstaltung ist keine Anmeldung erforderlich. PDF- Link