Theses

Getting started

We supervise Bachelor's and Master's theses for students enrolled at the Technical University of Munich (TUM Campus München and TUM Campus Heilbronn). Most of our students come from Informatics, Mathematics, Computational Science and Engineering, or the Information Engineering programs, but we welcome motivated students from related disciplines as well.

All open topics are listed on the TUM Thesis Portal. Please browse the portal and apply for a topic there — that is the single entry point for starting a thesis with our group. If you have your own topic proposal that fits our research directions, you can submit it through the portal as well.

Topic areas

Our theses typically sit at the intersection of numerical mathematics, high-performance computing, and scientific software engineering. The areas we most frequently supervise are:

• High-performance computing & parallel algorithms
• GPU programming
• Sparse numerical linear algebra
• Iterative solvers & preconditioning
• Mixed-precision and communication-avoiding algorithms
• AI for scientific computing

What we expect

A thesis with us is a supervised research project. We expect students to take ownership of the work, while we provide technical guidance, compute resources, and access to our collaborators and codebases.

Every thesis is managed through the TUM Thesis Portal. You either pick one of the open topics on the portal or submit your own topic there — we do not track theses outside of this system.

Submitting a proposal (as a LaTeX- or Typst-rendered PDF) is mandatory before any work begins. Your proposal should clearly describe:

• The research question you intend to investigate.
• Possible methodologies and the approach you plan to take.
• How you will tackle the problem, including tools, datasets, and evaluation strategy.
• A detailed weekly timeline covering the full duration of the thesis.

During the thesis itself we expect:

• Regular meetings with your primary supervisor — typically weekly or biweekly, depending on the phase of the project.
• A short kickoff presentation within the first few weeks to align on scope, methodology, and success criteria.
• Periodic written progress updates so that blockers are surfaced early.
• A final defense presentation to the group during one of our Thursday seminars, in person at TUM Campus Heilbronn.
• Thesis duration and formal steps follow the TUM examination regulations for your study program.

Writing your thesis

We expect theses to be written in English. Both LaTeX and Typst are welcome — pick whichever you are more comfortable with. Recommended templates:

• Exzellenz TUM Thesis — Typst template styled for TUM theses.
• TUM-Dev / tum-thesis-latex — LaTeX template for TUM theses.

A typical structure that works well for our theses:

• Abstract and introduction (motivation, research question, contributions)
• Background & related work
• Methodology — the algorithms, models, or systems you propose
• Implementation notes — enough detail to reproduce the work
• Experimental evaluation — setup, baselines, metrics, results
• Discussion, limitations, and future work
• Conclusion

Cite using BibTeX (or the equivalent bibliography mechanism in Typst), keep your bibliography file in the repository, and make sure every claim of prior work is backed by a citation. Figures should be generated from scripts rather than hand-drawn so that they can be regenerated from raw data.

Code and reproducibility

Your implementation work is part of the thesis. We expect the same engineering hygiene we apply to our own projects, so that your results can be reproduced and built upon.

• Keep all code in a git repository from day one — commit early and often.
• Provide a clear README with build instructions, dependencies, and how to reproduce the key experiments.
• Write tests for the core logic you introduce. Continuous integration is encouraged.
• Document datasets and benchmark problems you use, including their source and any preprocessing.
• When contributing to one of our group's codebases (e.g. Ginkgo), follow that project's contribution guidelines and coding style.
• Your final submission should include (or link to) an archived code snapshot matching the results reported in the thesis.

Learning about the group

To get a feel for what we work on before applying, have a look at our research areas, publications, and the group members. Each topic on the portal is owned by one of us, so knowing who works on what helps you pick a project that matches your interests.