Algorithms solve problems. But which algorithm is best suited to which problem? During his doctoral research, Diederick Vermetten developed methods to figure this out. And that is important for the entire field.
Vermetten investigated which type of algorithm works best in certain situations. From your Instagram timeline to online banking: algorithms are everywhere. And they are not always viewed in a positive light. But an algorithm is really nothing more than a set of instructions for solving a problem.
'At Leiden University, I was able to combine mathematics and computer science in my bachelor's degree.'
'We want to find methods to solve problems faster or better'
The researcher enjoyed his time in Leiden. He completed his bachelor's, master's, and PhD here. 'At Leiden University, I was able to combine mathematics and computer science in my bachelor's degree. That's when I discovered that computer science suited me better.' He found his place in Thomas Bäck's group, where he studied optimisation problems.
'Optimisation simply means that we want to find a method that can solve a problem better or faster,' explains Vermetten. He gives an example to illustrate this. 'Suppose you want to build a car. You can choose all kinds of materials in different thicknesses for the parts. Optimisation algorithms can search for the lightest possible combination of parts that is still safe. That is cheaper in production. At the same time, optimisation can also mean that something becomes more environmentally friendly or requires less computing power.'
When does an algorithm work well?
Vermetten's research goes one layer deeper than such applications. He wants to know how to choose the best algorithm for a given problem. That is why he created software to compare existing algorithms. This software collects data to see in which situations an algorithm works well. Thanks to this data, the PhD candidate can not only find the best algorithm, but also combinations of algorithms that can do the job even better.
'It can be quite difficult to link different algorithms together,' he admits. 'But we have already found situations in which this really pays off, which is great to see.'
Software that is useful for all computer scientists
The software is unique because it enables others to perform similar analyses. The PhD candidate shared his work open source, freely accessible. 'This allows us to share and apply knowledge more quickly.'
Vermetten started his PhD during Covid-19. That was a challenge, but fortunately you can do computer science anywhere, he puts it into perspective. 'It also gave me the opportunity to collaborate with many people internationally,' he adds. 'This allowed me to see what problems they were working on and how I could improve my software. My methods proved to be useful to others, which gave me a lot of satisfaction.'
From Leiden to Paris
'I was given the freedom to pursue what I found interesting,' Vermetten reflects on his PhD in Leiden. He is now continuing his research at the Sorbonne University in Paris. 'The beauty of research, in my opinion, is that you are guided by the wish to explore the unknown. How does something work exactly? There is no boss telling you what to do; you must ask your own questions.'
Vermetten defended his thesis 'From benchmarking optimization heuristics to dynamic algorithm configuration' successfully on 13 February 2025.
