Deutsch Intern
  • Students in the lecture hall
  • Nahaufnahme Notizen
  • Interior of the library of mathematics
Applied Stochastics

Applied Probability, Statistics and Learning Lab

The Applied Probability, Statistics and Learning Lab Lab, located in rooms 30.02.012 and 30.02.013, provides a dedicated space and infrastructure for applying stochastic models, statistics, and learning methods to real-world problems. In addition to technical equipment, collections of program code and teaching materials are collaboratively developed and made available for further use. As part of research projects and thesis work, the lab serves as a workspace and meeting point for both students and staff.

In our lab, we work with intra-daily high-frequency financial data from the electronic trading platform NASDAQ. We use limit order book data recorded at the highest possible frequency from the LOBSTER* database. For all NASDAQ-listed stocks, this database provides detailed records of limit order submissions and cancellations, including associated volumes and price levels. The data has at least millisecond precision, enabling analysis at the finest time resolution. The data is used for our research on statistical methods for the econometric analysis of high-frequency intra-daily prices, as well as for applied projects studying the economic implications of such analyses. Our research advances and integrates concepts such as volatility, jumps, and market microstructure in the modeling and analysis of high-frequency limit order book data.

* Link to LOBSTER

Project contact: Prof. Dr. Markus Bibinger, Adrian Grüber

This Wuedive project integrates customized Large Language Models (LLMs) into a joint course development initiative with the University of Bergen as part of the CHARM-EU framework. Its primary aim is to enhance learning processes, create interactive materials, and ensure the course’s scalability within the CHARM-EU Alliance. The course content focuses on network models, Markov processes, interacting particle systems, statistical physics of disordered systems, and neural networks and learning algorithms—topics of high scientific relevance, as underscored by recent prestigious awards (Nobel Prizes: Parisi 2021; Hinton and Hopfield 2024; Abel Prize: Talagrand 2024). Course content and materials will be developed using the computing resources of the Stochastics Lab, including local hardware and access to a high-performance cluster. Participants will also have access to the lab facilities.

The specific goals of the project include:

  • finely tuned LLM support for course content;

  • high content accuracy through domain-specific training;

  • the development of adaptable learning structures;

  • the creation of interactive materials for model and algorithm exploration;

  • differentiated learning paths with tailored exercises.

The course bridges mathematics, physics, and computer science with artificial intelligence, thus fostering interdisciplinarity. LLM integration supports teaching through self-directed and adaptive learning, as well as intelligent problem-solving. Continuous LLM availability increases scalability, enabling broader student participation. Lightweight LLMs and reusable content promote sustainability and establish expertise in AI-supported education and knowledge transfer.

Project contact: PD Dr. habil. Anton Klymovskiy, cooperation partner: Prof. Dr. Stein Andreas Bethuelsen, University of Bergen