What Ignites a Neutron Star?

Supercomputers at the National Energy Research Scientific Computing Center (NERSC) helped set the stage for a team of astrophysicists from Stony Brook University, Los Alamos National Laboratory (LANL) and Lawrence Berkeley National Laboratory (Berkeley Lab) to perform the first detailed 3D simulation of an X-ray burst.

This ongoing research, which also involves supercomputing resources at the Oak Ridge Leadership Computing Facility (OLCF), is enhancing our understanding of the behavior of dense nuclear matter and neutron stars.

X-ray bursts are the thermonuclear “runaway” in a thin hydrogen/helium layer on the surface of a neutron star, a relatively small star (10 km or so) that is the densest stellar object in the universe. This fuel layer accretes from a binary companion star, and the immense gravitational acceleration compresses it, increasing the temperature and density to the point of explosion.

“The main utility of X-ray bursts these days is we want to understand the properties and behavior of dense nuclear matter in a neutron star, and X-ray bursts are good probes of that,” said Michael Zingale, associate professor of physics and astronomy at Stony Brook and lead author on an Astrophysical Journal paper outlining the group’s comparison of 2D and 3D simulations of Type I X-ray bursts. “We know a lot about these stars; there are maybe up to 100 sources that we can see in our own galaxy, and we can see multiple bursts from a single X-ray burst. If we can understand how to interpret those events, we can, with a lot of observations and analysis, try to infer what the mass is.”

One-dimensional hydrodynamic studies have been able to reproduce many of the observable features of x-ray bursts, such as burst energies and recurrence times. But multi-dimensional simulations are necessary to more fully understand the convection process in the extreme conditions found in a neutron star.

“There are lots of models with lots of different physics, but we don’t exactly understand how matter behaves when compressed to densities or more than 1014 grams/cm3,” Zingale said. “The idea is if we can simultaneously measure the mass and the radius of the neutron star—not the mass of one and the radius of another, but of the same star—this data point can tell us a lot of what the physics is inside of the neutron star and how the nuclear matter behaves.”

To model the convection in the burning layer on the neutron star surface, the team first turned to NERSC, using 5 million MPP hours on the center’s Edison system to explore parameter space.

“This is where the NERSC machine excels,” Zingale said. “It gives us fast turnaround and the ability to explore multiple models quickly.”

The researchers then used supercomputing resources at OLCF to run the “hero” calculation that resulted in the 3D simulation—the first reported 3D calculations of convective burning in a mixed hydrogen/helium X-ray burst, Zingale noted. In addition to Zingale, co-authors Chris Malone from LANL and Berkeley Lab’s Andy Nonaka, Ann Almgren and John Bell were instrumental in this research, which is supported by the Department of Energy’s Office of Nuclear Physics.

“We are the only group doing 3D studies that resolve the burning scales and capture the convective dynamics,” he said. “These studies will help us understand how to interpret observations of X-ray bursts.”

MAESTRO Code Pivotal

This latest research builds on the team’s previous 2D studies at NERSC using the publicly available MAESTRO code, developed at Berkeley Lab’s Center for Computational Sciences and Engineering (CCSE) in collaboration with Zingale. MAESTRO is a low-Mach-number hydrodynamics code that exploits the separation of scales between the fluid velocity and the speed of sound. It is ideally suited to modeling convective flows in astrophysics, supporting a stellar equation of state, nuclear reactions and diffusion, in addition to hydrodynamics. MAESTRO was written in the BoxLib software framework developed by CCSE and takes advantage of the multicore architectures used at NERSC through a hybrid MPI + OpenMP programming model.

“MAESTRO was originally developed for studying thermonuclear supernova, but it is also applicable to the convected stage in neutron stars,” Zingale said. In this case, the researchers used it to model a small box on the surface of a neutron star, not the entire star, he added. “We are modeling tens of meters on a side, 10 meters of depth on the surface of the neutron star, but that is enough to resolve features that are a few centimeters in length, which is enough to see the burning and to see it drive the convection.”

As a follow-on to this research, the team is running a new class of simple model problems at NERSC that uses a more realistic initial model and reaction network, Zingale noted.

“Ultimately we want to understand does convection bring any products of the burning up to the photosphere, to the place where you can seem them that can inform the observers to help interpret what they see,” he explained. “And if it is brought up to the surface, will it change how these things look? Will they be more opaque, less opaque? Is it going to change how easily the light escapes, which will have observational consequences? We are working our way toward that understanding.”

NERSC is a DOE Office of Science user facility.