TUHH / Institut für Mathematik / Vorträge Englische Flagge

Vorträge

Suchen | Vortragsverwaltung

Vorträge 1 bis 10 von 684 | Gesamtansicht

Nächste Seite Letzte Seite
Datum Zeit Ort Vortrag
14.02.25 10:00 Am Schwarzenberg-Campus 3 (E), Raum 3.061 Ein Wachstumsmodell für zwei Infektionen in zufälligen Graphen [Bachelorarbeit]
Yannic Hillers

Symbol: Pfeil nach oben
12.02.25 13:30 Am Schwarzenberg-Campus 3 (E), Raum 3.074 Bachelorarbeit: Transformer Netzwerke als PDE-Lösungsoperatoren
Ali Mowafek Aouda

Symbol: Pfeil nach oben
10.02.25 13:00 Am Schwarzenberg-Campus 3 (E), Raum 3.074 Bachelorarbeit: Modellbasierte Positionsbestimmung autonomer Fahrzeuge
Ferdinand Grenzing

Symbol: Pfeil nach oben
05.02.25 12:00 Am Schwarzenberg-Campus 3 (E), Raum 3.074 und Zoom Mathematical Insights into Electrical Impedance Tomography for Chemical Reactors
Moritz Hollenberg

In this talk, I will explore the application of Electrical Impedance Tomography (EIT) in chemical reactors, presenting a theoretical deduction of the underlying mathematical problem from its real-world context. The discussion will focus on the ill-posed nature of the EIT inverse problem and demonstrate how additional modeling assumptions can stabilize the reconstruction process.

The presentation explores ideas for developing an objective framework to evaluate reconstruction performance and seeks input on how to effectively incorporate physical domain information to enhance physics-based reconstruction approaches.

Zoomlink:
https://tuhh.zoom.us/j/81920578609?pwd=TjBmYldRdXVDT1VkamZmc1BOajREZz09

Symbol: Pfeil nach oben
04.02.25 10:00 Am Schwarzenberg-Campus 3 (E), Raum 3.074 Bachelorarbeit: Verbesserung einer existierenden Lösung für 4D Gaussian Splatting mit Hilfe von Zeitinterpolation und LSTM Netzwerken
Anton Lausen

Symbol: Pfeil nach oben
29.01.25 12:00 Am Schwarzenberg-Campus 3 (E), Raum 3.074 und Zoom Energy consistent schemes for port-Hamiltonian systems
Jan Giesselmann, TU Darmstadt

Port-Hamiltonian systems are an energy based modelling paradigm that has received a lot of attention in recent years. It can can be applied to a wide variety of (physical) systems including PDE models in fluid and solid mechanics. We propose their structure preserving and arbitrary order discretisation via modified Petrov-Galerkin methods. These methods are provably energy consistent in the sense that they conserve or dissipate energy if the original system has this property. In numerical experiments we observe optimal convergence orders (depending on the polynomial degree) and nodal super convergence.

Zoomlink:
https://tuhh.zoom.us/j/81920578609?pwd=TjBmYldRdXVDT1VkamZmc1BOajREZz09

Symbol: Pfeil nach oben
22.01.25 12:00 Am Schwarzenberg-Campus 3 (E), Raum 3.074 und Zoom TorchBraid: High-Performance Layer-Parallel Training of Deep Neural Networks with MPI and GPU Acceleration
Jacob Schroder

Deep neural networks (DNNs) exhibit excellent performance for many machine learning tasks, e.g., image classification, natural language processing, and game playing. However, training DNNs remains challenging and computationally expensive, with much room for improvement, both in terms of new sources of parallelism and algorithmic speedup. One of the key bottlenecks is the serialization inherent in forward and backward propagation, which limits strong scaling in the limit. Recently, the parallel-in-time method, multigrid-reduction-in-time (MGRIT), has been applied to some DNNs to overcome this bottleneck by providing new parallelism in the layer dimension (layer-parallelism). This new parallelism is made possible by a connection between the layer-dimension and a hypothetical time-dimension. In this talk, we introduce layer-parallelism with MGRIT and then discuss TorchBraid, which is a high-performance implementation of this approach that supports MPI-based parallelism in combination with GPU acceleration. To achieve this, TorchBraid integrates the PyTorch neural network framework with the XBraid time-parallel library. We present results for Torchbraid with and without GPU acceleration, considering Tiny ImageNet and MNIST, as well as recurrent neural networks and transformers for language processing. We also present new results showing the computational advantage of combining layer-parallelism with data-parallelism and how to adapt standard deep learning techniques, like batch-normalization, to the layer-parallel setting. Lastly, we discuss TorchBraid's approach for overcoming the algorithmic challenges inherent in combining automatic differentiation with layer-parallel in a distributed MPI setting. Overall, TorchBraid enables fast training of DNNs, both in a strong and weak scaling context.

Zoomlink:
https://tuhh.zoom.us/j/81920578609?pwd=TjBmYldRdXVDT1VkamZmc1BOajREZz09

Symbol: Pfeil nach oben
16.01.25 16:45 Am Schwarzenberg-Campus 3 (E), Raum 3.074 A Palm approach to tail processes and tail measures from extreme value theory
Günter Last, Karlsruher Institut für Technologie, Institut für Stochastik

Tail processes and tail measures are important concepts in the theory of regularly varying (heavy tailed) time series. In this talk we will show that these concepts are intimately related to Palm theory of stationary random measures. To motivate the topic, we start with providing some background on regularly varying time series. Then we shall introduce tail fields in an intrinsic way, namely as spectrally decomposable random fields satisfying a certain space shift formula. The index set is allowed to be a general locally compact Hausdorff Abelian group. The field may take its values in an Euclidean space or even in an arbitrary measurable cone, equipped with a pseudo norm. We characterize mass-stationarity of the exceedance random measure in terms of a suitable version of the classical Mecke equation. As a rule, the associated stationary measure is not finite. We shall show that it is homogeneous, that is a tail measure. Finally we will establish a spectral representation of stationary tail measures.

Symbol: Pfeil nach oben
15.01.25 16:00 Am Schwarzenberg-Campus 3 (E), Raum 3.074 Polynomial stability of port-Hamiltonian systems
Sahiba Arora, Leibniz Universität Hannover

In this talk, we characterize quantitative semi-uniform stability for $C_0$-semigroups
arising from port-Hamiltonian systems, extending and complementing recent results
on exponential and strong stability. Using this characterization, we construct a sim-
ple, universal class of port-Hamiltonian systems that exhibit arbitrary decay rates
slower that $t^{1/2}$. This construction leverages results from the theory of Diophantine
approximation.
This is joint work with Felix Schwenninger (Twente), Ingrid Vukusic (Waterloo),
and Marcus Waurick (Freiberg).

Symbol: Pfeil nach oben
10.01.25 10:00 Am Schwarzenberg-Campus 3 (E), Raum 3.074 Development of mathematical algorithms for simulating camera raw data
Merlin Maximilian Arians

Symbol: Pfeil nach oben