TeXt: Marianne Fuglsang Welling Farsinsen, ITS og Niels Krogh Søndergaard, Department of Chemistry and Biosciences, Photo: AAU

 

Morten Smedskjær's research group deals with glass materials in different forms for different applications. This means that the mechanical properties of glass are the focal point, such as its ability and resistance to breakage. Typically, this requires a lot of experimentation in the physical laboratory, which is an expensive and cumbersome process. This is where supercomputer simulations make a valuable contribution early on in the process:

- We want to simulate before we go to the lab, so that we can separate out candidates for further design work. This requires computing power because we have to look at fractures and simulate the movements of individual atoms. Therefore, we need a lot of simulations and over long time scales to make it realistic and actually be used to guide the laboratory work," explains the professor.

The grant of 10 million CPU core hours on the EuroHPC VEGA supercomputer is record breaking at AAU and will accelerate the work of Morten Smedskjær and the research group:

- Our goal is to analyze data from the simulations so that we can extract knowledge. Instead of only being able to explain what mechanical properties a given material has, we want to turn it around and also be able to say: if I want this property, then this is the chemical composition of the material I need," explains the researcher from the Department of Chemistry and Biosciences.

Start small

Supercomputing and machine learning on this scale is not easy to jump straight into. It requires experience and knowledge to apply for, receive and manage the allocated resources. Fortunately, there is good help available at AAU, where the IT research support unit CLAAUDIA is ready to help. Morten explains the path to using supercomputers in research today:

- We started with resources that were available through AAU's CLAAUDIA. It was an important first step to have a playground where we could test, practice and gain experience with how simulations could be used in research. It was also valuable to share knowledge with other AAU colleagues who had been at it for longer than us. It is crucial to have local contacts, and here both colleagues and CLAAUDIA have formed the basis for the path to where we are today, says Morten Smedskjær.

Therefore, Morten's recommendation for how to get started is also to start small:

- I strongly believe that you need to have a concrete, closely related problem that, for example, simulations could help to solve within a relatively short time horizon of, say, six months or a year. And then start through CLAAUDIA with access to the local computer resources, where you can get access without having to write an application. This will provide a good start with learning and experience for several future research projects with calculation resources – also on a larger scale.

AAU's CLAAUDIA offers access to computational resources through among others DeiC, Danish e-infrastructure Cooperation, but also helps AAU's researchers in identifying needs, clarifying which computational resources match the data in the research, they provide guidance in application processes and in general are happy to meet AAU's researchers locally at the departments to discuss how the calculational power of supercomputers can create value in research.

All inquiries to CLAAUDIA are more than welcome at claaudia@aau.dk.