Phd in the Spotlight: Joris Borgdorff
Joris Borgdorff (1984) will be awarded his doctorate degree at the UvA on 3 July. During his doctoral research at the Informatics Institute (IvI), he researched what is needed to enable supercomputers to carry out simulations efficiently.
What did you do?
'I researched the appropriate preconditions for distributed multiscale computing (DMC). DMC is a form of simulation using various scales, such as space or time. These scales are calculated on different computers and subsequently combined. This makes it easier to comprehend complex processes such as climate or the human body. I examined the requirements for efficient DMC. I did this using a simulation of a blood vessel on which PCI had been performed using a stent, a tiny metal "tube" which is inserted into a blood vessel to relieve an obstruction. This simulation incorporated several time scales, namely the blood flow and how the cells in the vascular wall respond to it. The latter process is slower than the blood flow.'
What did you do with the simulations?
'I linked two supercomputers together: the Cartesius supercomputer here at the Science Park and the SuperMUC supercomputer in Munich. I had these computers perform the blood vessel simulation. One computer calculated the blood flow, the other the blood vessel . To ensure the simulation proceeded as smoothly as possible, I developed the MUSCLE 2 software, in collaboration with the Poznan Supercomputing and Networking Center and other institutes, which is now used by several research groups. Combining the computers proved useful. Thanks to the additional calculation capacity, some DMC simulations were performed between two and a hundred times faster, and smart distribution of the calculation time made other simulations twice as efficient. This is not always the best approach, however. Not for simulations that require only a few processors, for instance.'
Why simulate a blood vessel?
'Because we do not yet know exactly what happens when a stent is inserted. Why does narrowing of the blood vessel occur in one patient but not in another? That is difficult to measure in living humans, so computer simulation could help increase our understanding. At conferences I often discussed the missing elements of a computer model with doctors; their suggestions helped me take the development of the model a step further. Sometimes they said that they could now view their subject from a different perspective. Ultimately, that's what it's all about.'
How did you come to study Computational Science?
'In Utrecht I studied Informatics and Mathematics. That was very theoretical; very different to Computational Science, the field in which I am about to obtain a doctorate. Computational Science is very much application oriented. It aims to express reality in computer models, and that's an approach that reflects my own interests. After this phase I'd like to work in an advisory capacity, helping researchers explore the possibilities of modelling and simulating systems. At University College London's Faculty of Chemistry they actually have three people working on facilitating simulations. This is the moment to make it clear to Dutch governing bodies that IT specialists and computational scientists are very much needed to help scientists. There is a vast amount of useful work they could do.'
Author: Carin Röst