Rijk Mercuur (Artificial Intelligence), “Interventions on Contextualized Decision Making: an Agent-Based Simulation Study”
For this Master’s thesis, Rijk Mercuur researched the effectiveness of interventions on routine behaviour, such as choosing vegetarian or non-vegetarian food. He combined a number of theories from sociology and agent technology into a model that can be checked with simulations. Mercuur paid close attention to investigating the validity of his model and conducted a large number of simulations that show that not only social practices, but also social relationships have a major impact on people's reactions to interventions on their behaviour.
Jan Willem Buurlage (Mathematical Sciences), “Self-improving sparse matrix partitioning and bulk-synchronous pseudo-streaming”
In his thesis, Jan Willem Buurlage describes an algorithm for the partitioning of sparse matrices for the purpose of parallel computing, starting at a simple cyclic partitioning and improving itself over time as it is being used in an iterative linear system solver. He presents an implementation of several parallel algorithms on the Parallella board, and a corresponding methodology for
pseudo-streaming algorithms, breaking up the input and output data of the algorithm
into streams of tokens.
Two other students were also nominated for the award: Jochem Wijten and Peter Boot.
Jochem Wijten (Nanomaterials Science), “Development of Metal-Organic Frameworks for Photocatalytic Applications”
Jochem Wijten researched catalysts that can be used in chemical storage methods for solar and wind energy, specifically a new class of materials called Metal-Organic Frameworks. In these materials, metal ions and organic molecules are combined in a porous network. Wijten studied the effect of different metals and organic molecules on the light-absorbing properties of the material itself, and also developed a new Metal-Organic Framework. His results will contribute to a faster and better development of renewable fuels. The thesis supervisor and the jury are impressed by the quality of the thesis and the quantity of the experiments described. The thesis of Jochem Wijten was also one of the three nominees for the university thesis award.
Peter Boot (Game and Media Technology), “Using Discovered and Annotated Patterns as Compression Method for determining Similarity between Folk Songs”
Peter Boot developed a new methodological framework for modeling the similarity between melodies, in which patterns are used as a means of compression. A large-scale retrieval experiment confirmed the intuition of musicologists that repeating patterns are crucial in the similarity between melodies. By bringing together various basic concepts from music research, Boot managed to identify shortcomings in existing state-of-the-art algorithms for pattern recognition. His results allow for fast searching in large data collections based on musical pattern information. The jury highly enjoyed reading the thesis which was written in an excellent style, understandable also to a non-expert reader.
Graduate School of Natural Sciences
The Graduate School of Natural Sciences organises all Utrecht Master’s and PhD programmes dealing with mathematics, computing and information sciences, physics and astronomy, and the interface of science and other disciplines.