About potential thesis projects

Exciting thesis projects await you in the group of Prof. Erwin Laure. Our group performs research in computer science with special emphasis on high performance computing. Matching your personal interests, you will find thesis opportunities ranging from theoretically oriented subjects to very practical undertakings like code development or the analysis of performant software packages from all over the world.

We develop an adaptive runtime system for simulation applications that helps to improve load-balancing that is a very important factor to reach high computational performance. Some information about the runtime-system can be found on this page. Other interesting questions arise for example from the use of GPU devices in hybrid programming models or from new PGAS programming languages. You find several options for master thesis projects in the context of the runtime-system project.

Do you have interest in scientific computing and numerical mathematics? You could consider a thesis project dealing with the design and implementation of simulation algorithms. Simulations running on future parallel computer systems will reach the needed performance not only from new software development tools alone. Algorithmic approaches that have been used partly for quite a long time need a further development, which takes recent developments of system architectures into account. You could work on algorithms, design and implement them, use advanced numerical libraries and our runtime system as well as challenge Sweden's and even some of Europe's largest computer systems to prove the development.

Do you have more interest in seeing detailed how existing simulation applications tick? A project focused on application benchmarking and performance improvement could be something for you. We need to know in detail why parallel applications work very well as well as when and why they will fail to provide high performance in order to develop the right tools that help to overcome existing limitations. You could do such analyses of parallel programs, find their bottlenecks, and even work on improvements to get them fit for future challenges.

Finally follows one suggestion for enthusiasts of mathematical studies and theoretical computer science. There is the opportunity to have a thesis project focused on scheduling algorithms. The runtime-system makes use of scheduling algorithms to plan the distribution of computational tasks. Here you can work according to your interests on the analysis or on the implementation of scheduling algorithms.

Please contact me if you are interested in the mentioned subjects or if you would like to discuss your ideas of a thesis project.

Past projects

The list of previous thesis projects that I have supervised shows the wide range of possible projects. Many more interesting things are waiting.

Md Naim Implementation, performance analysis and optimization of a molecular dynamics tree algorithm for large-scale cluster systems. 2012.
now at University of Bergen.
Johannes Bulin Large-scale time parallelization for molecular dynamics problems. 2013.
now at University of Bonn and Fraunhofer SCAI.
Julian Hornich Hybrid Acceleration of a Molecular Dynamics Simulation Using Short-Ranged Potentials. 2013.
now at University of Erlangen-Nürnberg
Christoph Rachinger Scalable Computation of Long-Range Potentials for Molecular Dynamics. 2013.
now at University of Erlangen-Nürnberg
Patrik Seim Load-balancing of Molecular Dynamics by Subdivision And Distribution of Domains in Large Multi-Core Systems. 2013.
Maximilian Volkan Baloglu Parallelisation and Performance Analysis of a TreeSPH Code for Galaxy Simulations. 2014.
now at University of Erlangen-Nürnberg
Sebastian Tunstig System modeling for process mapping onto scattered computational nodes in high performance computing clusters. 2014.

Last change: 2014-12-23