Computational thermodynamics
From thermodynamics and statistical mechanics to predictive materials modelling.
Phase stability, free energies, defects and finite-temperature properties from first principles.
Janssen Lab · Materials Informatics
From studying to research.
Apply what you learn at Ruhr University Bochum to current research questions in computational materials science and materials informatics.
The Materials Informatics group at the Max Planck Institute for Sustainable Materials (MPI SusMat) develops computational methods that connect atomistic simulation, thermodynamics, machine learning, scientific workflows, high-performance computing and AI – to make the discovery of sustainable materials predictive, reproducible and scalable.
- Research
- Materials Informatics group, MPI SusMat, Düsseldorf
- Teaching
- Lectures at Ruhr University Bochum
- Group leader
- Dr. Jan Janssen
Learn → Apply → Research
How lectures connect to research
The methods taught in lectures are the same methods used to answer open questions in materials science. Each step builds on the previous one.
-
Step 1
Learn
Build foundations through lectures and open teaching material.
Programming · Thermodynamics · DFT · Atomistic simulation · Machine learning
-
Step 2
Apply
Use modern computational tools to turn theoretical knowledge into quantitative materials predictions.
Python · pyiron · Interatomic potentials · HPC · Scientific workflows
-
Step 3
Research
Use these methods to investigate open scientific questions.
Phase stability · Uncertainty · Materials discovery · Autonomous workflows
Research
Four connected research themes
Machine learning for materials
From machine-learning fundamentals to predictive atomistic models.
Machine-learned interatomic potentials, Bayesian methods and uncertainty quantification for materials discovery.
Scientific workflows & HPC
From Python functions to reproducible simulations on supercomputers.
pyiron, scalable execution, provenance and interoperable workflow definitions.
AI for science
From language models to autonomous scientific workflows.
Agents that combine large language models with validated simulation tools.
Your studies and our research
Where your knowledge fits
Many topics from Bachelor and Master courses lead directly into current research questions.
| If you enjoy or know … | … you could work on |
|---|---|
| Thermodynamics | Free energies, phase stability, defects |
| Atomistic simulation | Density functional theory, molecular dynamics, machine-learned interatomic potentials |
| Python programming | Scientific workflows, pyiron, research software |
| Machine learning | Interatomic potentials, Bayesian optimisation, uncertainty quantification |
| High-performance computing | Parallel execution and workflow infrastructure |
| AI and large language models | Agentic scientific workflows |
These are starting points, not prerequisites. Nobody is expected to know all of these topics – projects are matched to your background, your interests and the scientific question.
Recent research
Selected results
-
How precise is a DFT calculation – and which settings does a target precision require?
How precise is a DFT result? Error in the bulk modulus of nine fcc metals across energy cutoff and k-point sampling. Contour lines mark target precisions from 5 GPa down to 0.1 GPa. Figure 1 from J. Janssen et al., npj Comput. Mater. 10, 263 (2024), doi:10.1038/s41524-024-01388-2, licensed under CC BY 4.0. Instead of choosing convergence parameters by hand, the user specifies the precision they need. Uncertainty quantification then identifies the energy cutoff and k-point sampling that achieve it, tested on a large set of elements.
-
Can ordinary Python functions scale from a laptop to a supercomputer?
executorlib extends Python's standard Executor interface: a submitted function requests resources from the job scheduler, runs asynchronously and returns its result. From J. Janssen et al., J. Open Source Softw. 10, 7782 (2025), doi:10.21105/joss.07782, licensed under CC BY 4.0. executorlib extends the familiar Python Executor interface to HPC job schedulers such as SLURM and flux, so the same scientific code runs on a laptop and on a supercomputer.
-
Can a workflow written for one workflow manager run in another?
The Python Workflow Definition is an exchange format that allows the same computational workflow to be shared between the workflow management systems AiiDA, jobflow and pyiron – a step towards reproducible, interoperable simulation protocols.
-
How can large language models take part in scientific work?
A community review of 32 projects – including LangSim, which couples LLM agents to atomistic simulation codes – shows how language models are being used for property prediction, automation, data management and agentic workflows in materials science and chemistry.
Open-source software
Open-source research software
Software is a research output of the group. Students interested in scientific programming can contribute to these projects as part of a computational research project.
Take the next step
Interested in a Bachelor project, a Master thesis or another research project? You do not need a complete proposal – a short email is enough to start a conversation.
Or write directly to jan.janssen@rub.de