Master Thesis Supervision

I enjoy supervising and supporting Bachelor’s and Master’s theses at the Max Planck Institute in Garching near Munich. To make it easier for students to get started and to help with finding a suitable topic, this page lists and briefly explains potential projects and thesis topics. If you are interested in one of the projects, please contact the respective contact person of my group indicated with the topic.

Please note that to ensure close guidance and a productive collaboration, supervision is only possible for students who are able to work on-site in the Munich area.

Students interested in writing their Bachelor’s or Master’s thesis under my supervision are kindly asked to contact Lena Tauber (Team Assistant).

Please include the following information in your message:

  • Short CV
  • Motivation and reason for requesting supervision
  • Proposed thesis topic (if defined)
  • Time frame required by your university
  • Transcript of records

All information will be forwarded to me for review, and you will be contacted either by the team assistant or by myself regarding possible next steps. If you already have an idea for a thesis topic that is not listed here, feel free to include a short description of your current concept as well.

Potential Master Thesis Topics

  • Spectral Bayesian Imaging of Galaxy Clusters in X-ray

    X-ray emission from galaxy clusters carries rich information about the hot, diffuse plasma encompassing the cluster, known as the intracluster medium (ICM). This project focuses on Bayesian imaging of galaxy cluster X-ray data from instruments such as Chandra and eROSITA. Building on an existing information field theory (IFT)–based framework, the work aims to reconstruct the ICM emission in spectral domain and additionally decompose it into continuum and line components. The project will also extend from purely spectral reconstructions to joint spatial-spectral reconstructions of the ICM emission.
    Contact: Mrinal Jetti

  • Strong and Weak Gravitational Lensing with Charm

    Gravitational lensing is a quirk of general relativity that enables measurements of dark matter by measuring the d istortion of background light caused by the space-time deflections of massive objects along the line of sight. Depending on the student’s interests, the project can focus either on applying LensCharm [1] to real data (e.g., strong-lens modeling and inference) or on methodological/code development. Possible directions include:

    • Modeling a strong lens in multiple frequency bands
    • Enhancing weak lensing in LensCharm
    • Multi-plane lensing
    • Adaptive grids for strong lensing
    • More ideas are welcome!
    Contact: Matteo Guardiani & Julian Rüstig

  • Theory of Mind in Opinion Formation: Does Distinguishing Honest Error from Deliberate Deception Reduce Misinformation?

    This master’s thesis investigates how agents can distinguish between honest but poorly informed communication and deliberate deception in opinion formation. Using an existing agent-based model including bayesian learning and theory-of-mind, it analyzes how confirmation-biased distrust of disconfirming information can be exploited by strategic communicators. The central question is whether explicitly modeling this distinction reduces agents’ susceptibility to different forms of strategic misinformation. The thesis contributes to a more nuanced understanding of communication and deception in social systems.
    Contact: Viktoria Kainz

  • Time variable gamma ray sources with the Universal Bayesian Imaging Kit

    The Universal Bayesian Imaging Kit (UBIK) is a framework for multi-instrument data analysis. UBIK reconstructs and separates point sources from diffuse emission exploiting spatial and spectral correlations. In this master project UBIK should be enabled to reconstruct the time variability of point sources from data of the Fermi gamma ray satellite.
    Requirements: Basic knowledge of information field theory, Python programming
    Contact: Torsten Enßlin

  • Artificial intelligence and information field theory

    Many, if not all artificial intelligence and machine learning methods seem to have an information theoretical foundation. In this master thesis it should be investigated how much methods like generative neural networks, variational autoencoder, and/or stable diffusion models can be understood in the mathematical framework of information field theory.
    Requirements: Advanced knowledge of information field theory
    Contact: Torsten Enßlin

  • Causal Analysis of Opinion Formation in Agent-Based Model

    This thesis applies causal inference methods to an agent-based opinion-formation model. While variables such as information exposure, opinion updates, communication strategies, network structures, etc. are highly correlated, the underlying causal mechanisms remain unclear in this complex system. The central goal is to evaluate causal analysis techniques developed within the group by applying them to simulation data and testing whether they recover and validate causal relationships that are intuitively expected from the model’s design or observable through exploratory analysis, thereby serving as a first methodological proof-of-concept linking causal inference and agent-based social simulations. Depending on the student’s interests, the thesis might be extended either toward (A) applying validated causal methods to substantive questions in computational social science or (B) toward further developing and evaluating causal inference techniques in more general, complex, non-linear systems.
    Contact: Viktoria Kainz