The National Institute for Computational Sciences

Star Formation Exploration

Kraken helps researchers study the interaction of physical processes within star-forming clouds

by Caitlin Elizabeth Rockett


"We had the sky up there, all speckled with stars, and we used to lay on our backs and look up at them, and discuss about whether they was made or only just happened," wonders the title character in Mark Twain’s novel The Adventures of Huckleberry Finn. More than 100 years ago, Twain posed one of the greatest questions of modern astrophysics: How do celestial bodies form?

Figure 1: Movie: A simulated tour of a turbulent molecular cloud forming multiple protoplanetary disks. The Advanced Visualization Laboratory at the National Center for Supercomputing Applications rendered the data sets using in-house software. Credits: NCSA, NASA, A. Kritsuk, M. Norman.  

While mankind’s awe of the night sky remains undiminished since Twain’s time, contemporary astrophysicists have theorized that star formation occurs due to the interaction of turbulence, gravity and magnetism within the extremely dense, cold, gaseous environments of molecular clouds (MCs). However, the role these physical processes play to initiate star formation are poorly understood. But perhaps not for long—a team led by Alexei Kritsuk of the University of California at San Diego (UCSD) has taken this subject to task using Kraken, a Cray XT5 currently ranked as the 8th fastest computer in the world, to explore the interplay of turbulence, gravity and magnetism in star formation within MCs.

Kraken is located at the National Institute for Computational Sciences, a National Science Foundation-funded research facility managed by the University of Tennessee and located at Oak Ridge National Laboratory.

According to Kritsuk, the team is interested in the physical processes that determine molecular cloud structure and the formation rate of stars. Since 2008, Kritsuk and his team have used over 5 million CPU hours (the work done by a computer processing unit in one hour of wall clock time) on Kraken to conduct suites of simulations that elucidate how physical processes, both inside and outside MCs, interact to produce the wide array of stellar masses found in the cosmos.

Where stars are born

The matter that fills the space between stars is referred to as the interstellar medium (ISM). The ISM is a very low concentration of matter, consisting of approximately 99 percent gas and 1 percent dust.

The densest regions of the ISM are molecular clouds, and it is within these gaseous clouds that stars are born. Created by large-scale turbulence in the disk of a galaxy (like the Milky Way) and held together by the gravitational force of their own mass, molecular clouds are composed mostly of molecular hydrogen and helium, but also contain trace amounts of other heavier elements such as carbon, nitrogen and oxygen. Because of their dense concentration, molecules in MCs are shielded from destructive interstellar ultraviolet radiation, which, combined with an extremely cold environment (10 to 50 degrees Kelvin, or -440 to -370 degrees Fahrenheit), helps atoms to associate into molecules.

Molecular clouds form due to turbulence in the ISM. This turbulence originates from a number of sources including shockwaves from supernova explosions, large-scale shears resulting from differential rotation in the galactic disk, and gas accretion onto the galactic disc. MCs represent active regions of this turbulence that drain the kinetic energy created by these driving forces.

Interstellar turbulence creates an observable tiered structure of MCs—from giant molecular cloud complexes (containing tens of thousands to a million times more mass than the sun), to smaller MCs, to clumps, and finally dense molecular cores. It is these dense cores that are the immediate precursors of new stars. Turbulence continues to create and destroy these concentrations of mass, but some regions are sufficiently dense to undergo gravitational collapse. Eventually, these dense fragments condense into rotating balls of plasma that serve as stellar nuclei.

However, turbulence and gravity don’t act alone in the star formation process—magnetism also affects this phenomenon within MCs. Because the ISM is magnetized, MCs are threaded by magnetic fields. The magnetic field of a molecular cloud is directly tied to the ions and electrons within the cloud. As the ionized gas within the cloud moves, the magnetic field moves with it. This field undergoes stretching, twisting and folding, and as the gas gets sheared and compressed by the turbulence it causes the field to gain and lose strength. If the magnetic field grows strong enough, it will exert pressure on the ionized gas that can counteract self-gravity within the cloud, ultimately causing the dense cores within the cloud to collapse.

“It’s difficult to measure the strength of magnetic fields in molecular,” said Kritsuk. “Magnetic fields are thought to play an important role in star formation, but we don’t really understand the nature of magnetized supersonic turbulence in these clouds. Therefore, we needed to explore a wide range of different regimes in the turbulence to see which one better fits what we see in observations.”

But to determine the strength of MC magnetic fields or the roles that gravity and turbulence play in star formation, the team had to take a step back and begin by simulating the formation of MCs from first principles.

The science of star formation

To simulate the formation of molecular clouds in the ISM, Kritsuk and colleagues used high-resolution magnetohydrodynamic (MHD) models that describe the properties of electrically conducting fluids like plasma. These MHD simulations were intended to explore the nature of turbulence in newly forming MCs, and how this turbulence depends on the strength of the uniform magnetic field. Simulations showed that turbulence in molecular clouds is super-Alfvenic.

There is a magnetic field present in the ISM, and depending on the strength of the field, the ratio of kinetic energy to magnetic energy can be smaller or larger than one. In magnetohydrodynamics, this ratio is represented by the value of the Alfven Mach number.

“If this dimensionless number is larger than one, the kinetic energy dominates, and the magnetic field plays a secondary role and the turbulence is called super-Alfvenic,” Kritsuk explained.

MHD simulations of super-Alfvenic turbulence agree with noted observational data of MC fragmentation, signifying that turbulence dominates over a weak magnetic field. These simulations provided the team with important constraints on star formation scenarios.

However, the team’s most interesting finding, according to Kritsuk, is the effect that self-gravity has on the distribution of density in interstellar clouds. The findings of this research were published in the January 20, 2011 issue of Astrophysical Journal Letters.

Previous studies on supersonic turbulence in MCs showed a lognormal (bell-shaped curve) density distribution, but when self-gravity was introduced to the simulations there were hints at the development of a power-law tail at high-densities. A density distribution with a power-law tail displays an excess of dense and very dense gas in a cloud compared to the lognormal distribution.

While these power-law tails were known for nearly a decade, their origin remained obscure. Kritsuk and colleagues turned to high-resolution adaptive mesh refinement (AMR) simulations using the ENZO code—developed to simulate the formation of cosmological structures—in order to study the development of these power-law tails.

The high-resolution simulations show that the density distribution in star-forming MCs develops an extended power-law tail at high densities on top of the usual lognormal density, while clouds that produce no stars only demonstrate lognormal density distributions. These results allowed the team to make predictions about the distributions of projected density and magnetic field strengths of interstellar clouds, as well as provide new fundamental constraints for future star-formation simulations.

While the team has provided the astrophysics community with invaluable information on the formation of stars, their work on supersonic turbulence is also applicable to physics here on Earth, such as the fuel mixing and combustion in scramjets.

Now that the team has developed numerical methods that allow them to create realistic MCs, the next step is what Kritsuk refers to as “direct star formation simulations.”

“We will need large calculations on uniform grids combined with AMR. This way we will be able to reproduce turbulence and follow the collapse of prestellar clumps in molecular clouds in an accurate way.” This future work will no doubt require the computational might of Kraken, and provide even more insight into the nature of developing stars.