FB 6 Mathematik/Informatik

Institut für Mathematik

Navigation und Suche der Universität Osnabrück



WS 2019/2020

05.11.2019 um 14:15 Uhr in Raum 69/E23

Harsha Kumar (TU München)

Dynamics of the consensus problem on directed graphs

A fundamental issue in the area of complex network dynamics and multi-agent systems is the consensus (or agreement) problem, which requires a unanimous decision among processes (or agents) for a data value. In this talk, I will explain consensus dynamics on directed graphs (digraphs) with a non-linear communication protocol (interaction function) on unweighted edges. I will show the existence of bifurcations arising from the stated nonlinearity for strongly connected digraphs. I will also demonstrate how to combine the above result with techniques from fast-slow systems to get dynamic bifurcations, using the van-der-Pol-type nonlinearity as an example. If time permits, I will talk about a conjecture regarding a weaker criterion on digraphs that also show bifurcations as mentioned earlier. In the second part of the talk, I will explain a symmetrization algorithm that creates undirected graphs from digraphs. Finally, I will present a result on the topological equivalence of linear consensus dynamics on the input and output graphs for this algorithm. This constructs a bridge between dynamics on digraphs and signed undirected graphs.

08.11.2019 um 13:30 Uhr in Raum 69/E18

Timothy Nadhomi (University of Silesia in Katowice, Poland)

Properties of the Sugeno Integral

Fuzzy measure theory is a generalization of classical measure theory. This generalization is obtained by replacing the additivity axiom of classic measure with weaker axioms of monotonicity. The development of fuzzy measure theory has been motivated by the increasing apprehensiveness that the additivity property of classical measures is in some applications context too restrictive and consequently unrealistic. Jensen inequality is one of the most important tools in actuarial mathematics, and in the mathematics of finance in general. The Jensen inequality makes in particular any insurance policy possible. Recently there are many results extending the Jensen inequality for other aggregation operators and one of the is the so called Sugeno Integral. Sugeno integral is one of the most important fuzzy integrals, which has many applications in various fields. The thesis presents the notion of Sugeno integral. In particular we look at the Jensen type inequality for Sugeno integral, Conditions to the Jensen inequality for the generalized Sugeno integral, utility theory and Sugeno integral as an aggregation function

25.11.2019 um 13:30 Uhr in Raum 69/E23

Hendrik Pasing (Hochschule Ruhr West)

Some approaches on a posteriori error estimation in shape optimization

After a brief introduction to shape optimization we will discuss approaches on a posteriori error estimation, including a posteriori error estimation of the compliance and shape gradient approximation. In addition to the aforementioned approaches we will present associated open or pending questions. In general we will assume models of elastic structures if necessary.

26.11.2019 um 14:15 Uhr in Raum 69/E23

Marcin Wnuk (Universität Osnabrück)

Numerical Integration, Discrepancy and Negative Dependence

03.12.2019 um 14:15 Uhr in Raum 69/E23

Dominik Nagel (Universität Osnabrück)

Performance Analysis of the ESPRIT Algorithm

10.12.2019 um 14:15 Uhr in Raum 69/E23

Markus Wageringel (Universität Osnabrück)

Reconstructing measures under algebraic constraints

17.12.2019 um 14:15 Uhr in Raum 69/E23

Mathias Hockmann (Universität Osnabrück)

A mathematical perspective on structured illumination microscopy

Although physicists found a theoretic lower bound for the resolution of light microscopes in the 19th century, many researchers developed methods to overcome this diffraction limit in the last decades. Structured illumination microscopy is one of these inventions and it became a widely used tool in biological and medical applications. At the beginning of this talk, we will present the fundamental physical background of the method. Afterwards we will focus on the mathematics of the underlying imaging algorithm. In particular, the case of sparsely labeled samples will be addressed.

28.01.2020 um 14:15 Uhr in Raum 69/E23

Michael Schmischke (TU Chemnitz)

Learning high-dimensional functions on the torus