Three Leiden Science projects receive computing time on national supercomputers
A night sky of more than 40 petabytes in size, simulating young star clusters and understanding how the body inhibits viruses: three Leiden projects have received computing time on one of the national computer systems.
‘The advanced national computer systems are used for technical-scientific research in which major computational problems need to be solved and where the computational facilities of the individual institutions are inadequate,’ it says on the website of the Dutch Research Council (NWO). Six times a year, NWO decides which applications are approved. In the second round of 2020, NWO honoured three Leiden Science projects:
Huub Rottgering & Timothy Shimwell (Leiden Observatory)
Project: ‘Realising the LOFAR surveys and opening up a new window on the Universe’
Computer system: Grid Data Processing (2.000.000 SBU + 300 TB/year disk storage + 600 TB/year tape storage)
The Low-Frequency Array (LOFAR) is a network of radio telescopes of more than 1000 kilometres in diameter, the heart of which lies in the Dutch province Drenthe. The project maps the radio-wavelength sky with unprecedented sensitivity and resolution at low frequencies. But this also involves a huge amount of data: the LOFAR archive contains more than 40 petabytes, which equals 40,000 terabytes.
‘Each LOFAR dataset is gigantic,’ says Timothy Shimwell, a researcher at ASTRON and a guest researcher at the Leiden Observatory. ‘An eight-hour observation quickly results in 32 terabytes of data and we have thousands of such observations. In order to process this vast quantity of data efficiently, it must be done on computing facilities that are local to the LOFAR archives so that researchers do not have to move the data. This grant and a similar one on the JUWELS computer in Jülich, Germany, allow for this and are thus critical to the success of our project.’
The LOFAR astronomers reveal the radio emission of relativistic particles—which move at almost the speed of light—in millions of galaxies, many of which have never been seen before. ‘Our aims are broad, with active research all the way from exoplanets to cosmology and at almost all scales in-between.’
Francisca Concha-Ramirez (Leiden Observatory)
Project: ‘External photoevaporation shapes the distributions of circumstellar disk sizes in young star clusters’
Computer system: Cartesius (4.000.000 SBU)
Planets form in circumstellar disks, which are reservoirs of gas and dust that surround stars. Francisca Concha-Ramirez: ‘These disks develop soon after stars form, and during their early stages they are immersed in an environment that can be very hostile for their survival.’ The surrounding gas and nearby stars can affect the disks in many ways and make them lose mass quickly, which limits the time and material available to form planets. ‘Understanding how these disks evolve can help us understand the formation of planets and our solar system.’
An important phenomenon in this process is external photoevaporation: bright stars nearby heat the disks enough to evaporate their material and, in some cases, completely destroy them. ‘Our previous simulations show that external photoevaporation is one of the most important processes in removing mass from the disks.’ Now, Concha-Ramirez aims to implement the star formation process itself in her simulations. ‘By simulating the collapse of a molecular cloud and subsequent star formation, we can obtain initial conditions that are more similar to real star-forming regions.
‘The simulations of photoevaporating disks are already very computationally expensive, taking several weeks to complete,’ Concha-Ramirez says. ‘Adding star formation will make the computing process even longer. Having access to supercomputer facilities allow us to run these intensive and detailed simulations in reasonable timescales, and also to perform more runs to obtain better statistics.’
Jelger Risselada, Niek van Hilten & Jeroen Methorst (Leiden Institute of Chemistry)
Project: ’Multiscale simulations of membrane processes’
Computer system: Cartesius supercomputer (2.000.000 SBU)
In response to a viral infection, our cells express so-called IFITM proteins to keep the viruses out. These proteins do this by inhibiting the fusion of the viral envelope and the cell membrane. However, the question is: how does such a protein distinguish between an incoming virus and the many innocent fusion processes that are vital to the cell? PhD candidates Van Hilten and Methorst: ‘With our research, we are trying to fathom the effect of viral fusion inhibitors through computer simulations. In other words, learning from nature!’
In order to investigate such mechanisms efficiently, the trio makes use of artificial intelligence. ‘We are working on a so-called genetic algorithm, a kind of simulated evolution. This allows us to optimise certain qualities that viral inhibitors should have. The molecular mechanisms generated by the computer will provide us with new insights. These ultimately help us to develop new and even better virus inhibitors.’