Supercomputer simulates the birth of stars

Drink your next cup of coffee or tea in a glass: If you put milk into it, the beverage does not immediately turn a uniformly light color, but bright vapor and small eddies are initially distributed in it. In space stars are formed in similar processes or in turbulence, but instead of coffee and milk interstellar gases, molecules and dust are interacting with each other at a wide variety of speeds. What exactly happens in these vortices has now been reconstructed by a research team with the help of supercomputers at the Leibniz Supercomputing Center (LRZ) in Garching: Under the direction of Prof. Dr. Ralf Klessen from the Center for Theoretical Astrophysics at Heidelberg University and Prof. Dr. Christoph Federrath from the Australian National University Canberra, the largest simulation of the interstellar turbulence was created on the SuperMUC. It can be used to demonstrate when and how stars form. A major role in this, according to a finding published in the journal Nature Astronomy, is played by the speed of the motions and the sonic scale, which could be localized and mapped in the model for the first time.

Speed forms clumps in clouds

Approximately only one star or sun is formed per year in the Milky Way. That's because the interstellar gas, which makes up about 10 to 15 percent of the matter visible there, doesn't distribute itself evenly between stars, but rather meanders like the milk in the coffee in swaths, rising, swirling, pulling apart, gathering. "This turbulent behavior appears to be key to how interstellar gas clouds fragment under their own gravitational weight and cluster together to form stars and star clusters," Klessen says.

In space there’s no spoon to drive it; only the speed at which the gas clouds move influences the behavior of the molecules the contain. In large-scale waths they travel rapidly. Towards little vortices energy decrease and qwith it motion and speed, matter comes together. So turbulence develops as a decrease of velocities, from large to short length or – better – from supersonic to subsonic. Transition is marked by the sonic scale. "There have been theoretical predictions of where this transition zone should be, but its exact location, shape and width were unknown until now," Klessen said. "The physical processes are so complex that they can only be explored with the help of computer simulations." But the sonic scale shapes the properties of dense cloud cores - rapid, turbulence-dominated motions in the supersonic range drive matter apart, while slow, gravity-dominated ones lead them toward each other.

Complex equations plus different scales

To model the turbulence, complex equations had to be developed for calculating different gas densities, and a wide variety of scales also had to be taken into account: "For our particular simulation, in which we want to resolve both the supersonic and the subsonic cascade of turbulence with the sonic scale in between, this requires at least 4 orders of magnitude in spatial scales to be resolved", Federrath reports. In total, more than a trillion resolution elements were calculated for the model. The simulation consists of more than 100 snapshots, each taking up about 23 terabytes of hard disk space. In total, more than 65,000 computational cores worked on the Garching supercomputer for this, and the simulation requires around 130 terabytes of working memory. A challenge for researchers and application specialists: "I see our mission as being the interface between the ever-increasing complexity of the HPC architectures, and the scientists, which don’t always have the right skill set for using HPC resource in the most effective way," Dr. Luigi Iapichino says, head of the AstroLab at the LRZ. "Collaborating with Christoph was quite simple because he is very skilled in programming for HPC performance. In this kind of collaborations, application specialists are often full-fledged partners of researchers, because it stresses the key role centres’ staffs play in the evolving HPC framework

The capabilities of supercomputing stimulate more research questions: Multiple award-winning visualizations have long since emerged from interstellar turbulence data. But Klessen’s and Federrath's team looks forward: "Next we'd like to add magnetic fields, chemistry and cooling to a simulation of this size, in order to learn more about the processes taking place when stars form," Federrath says. "This will be extremely challenging, as it would take even more memory, space, and computing power.“ (vs)

Photo Ch. Federrath, R. Klessen et al: Interstellar gases form a turbulence. The image from the simulation shows a section through the turbulent gas. Turbulence creates motion, and this creates density contrasts in gas clouds that play a key role in the formation of stars. (Nature Astronomy. DOI: 10.1038/s41550-020-01282-z )