Here you can find examples of research projects at AU that have been granted resources on the supercomputer LUMI.
LUMI is a supercomputer in the highest international league. Denmark has a share of LUMI along with several other European countries, and all Danish researchers and PhD-students, regardless of field, can access LUMI, if their project is relevant and suitable to the machine.
Several AU researchers have already been given the opportunity to work with LUMI.
Get access to LUMI
There are two ways to access LUMI, depending on your research project's size.
In both cases, you can get help via AU's DeiC front office: [email protected]
In the Biomodelling-lab, we simualte complex biological cell-membranes and membrane proteins, that play a part in different diseases (e.g. Alzheimers and illnesses in the central nervous system).
I started work within the field having completed two Postdocs in the US, and I have since then applied for independent resources with a view to continuing my research into the brain-blood-barrier.
With LUMI we have reched the goals we defined for our simulation, and several of the lab's members, all at different stages of their careers (MA, PhD, and og Postdoc), have benefited from the powerful HPC-infrastructure.
A better understanding of the mechnism behind the formation and growth of molecular clusters in the atmosphere is important for our global climate. Quantum-chemical calculations provide direct insight into the thermo-chemistry of cluster formation, but such calculations demand a lot of computing power.
For this reason, exact benchmarks are necessary in order to identify effective methods for thorough examinations of new cluster systems' configurations. Via LUMI's architecture we were able to complete a groundbreaking benchmark-exam, both into the number of examined systems (11.749 acid-base-clusters) and the theoretical benchmark-level (PNO-CCSD(F12*)(T)/cc-pVDZ-F12).
We would not have been able to complete our benchmark exam without access to LUMI. The results show r2SCAN-3c to be a robust method that yields atmospheric cluster-binding energies that correspond quite closely with very exact methods. The results will shape the future of configurational sampling of atmospheric molecular clusters.
Published article: Massive Assessment of the Binding Energies of Atmospheric Molecular Clusters.
I have used the LUMI supercomputer, and it has been an amazing experience. Even though it resembles every other Linux-system at a glance, the resources it lets me access - e.g. large storage capacity and several CPUs - make it far easier to complete calculations that go beyond the capacity of my own system.
To have access to such a machine is really good for morale, because we don't have to worry about the resources, which can take up a lot of our time - time we could have spent better completing tasks.
Furthermore, it feels really good to know that I am one of very few people in the world with access to such a powerful computer.