When Stars Collide: 3D Computer Simulation Captures Cosmic Event
High-resolution Model Run at NERSC Reveals a Stellar Phenomenon
January 12, 2019
This article uses material from a news release originally published by the University of Alberta.
The aftermath of the collision of two neutron stars has been fully captured in a 3D computer model for the first time, thanks to research by University of Alberta astrophysicist Rodrigo Fernández and an international team. The achievement, which used supercomputing resources at Lawrence Berkeley National Laboratory’s National Energy Research Scientific Computing Center, has led to a better understanding of the cosmic collision, showing how heavy elements like lead and gold are created and accounting for a phenomenon missing in other models.
Their findings were published in the Monthly Notices of the Royal Astronomical Society.
Neutron stars are the smallest and densest stars, mostly made of elementary particles called neutrons. In August 2017, scientists detected the collision of two neutron stars for the first time by using the Laser Interferometer Gravitational-Wave Observatory. When two of these stars collide, they merge in a flash of light and debris known as a kilonova, as material explodes outward.
“The collision creates heavy elements, including gold and lead,” said Fernández. “In addition, we also saw for the first time a gamma ray burst from two neutron stars colliding. There’s a large amount of science coming out of that discovery.”
One of the important elements of studying the collision is the accretion disk—a collection of leftover debris that orbits the combined hyper-massive star—a cosmic footprint of the collision event. The material launched by the accretion disk should match up with the amount of matter that plays a part in the kilonova, helping scientists better understand the event.
The problem? The numbers haven’t added up—the kilonova was brighter than predicted by current models; meaning some of the material hadn’t been accounted for. That’s where Fernández’ research comes in, using computers to model the accretion disk to gain a better understanding of where everything ends up.
A Cosmic Conundrum
Modeling the event was no easy task. The first issue is that while the collision of the stars happens over the course of milliseconds, the disk can last for seconds, which means simulations take thousands of times longer. The second problem is that the physics of the accretion disk are complex and involve a large number of different physical processes, including nuclear reactions given the extremely high temperatures at play. But one process makes the model more challenging than any other.
“Among the processes at work, the main culprit is actually the magnetic field acting on the matter,” said Fernández. “We know the equations that describe that process, but the only way that we can properly describe them is in 3D. So not only do you have to run the simulation for a long time, you also have to model it in three dimensions, which is computationally very expensive.”
The computing was originally done in 2015 and 2016 when Fernández and his colleague and co-author Alexander Tchekhovskoy were post-docs at the University of California, Berkeley. They ran their simulations initially on NERSC’s Hopper system (decommissioned in 2015) and then on Edison. Other members of the team and co-authors on the Monthly Notices of the Royal Astronomical Society paper were Eliot Quataert, Francois Foucart and Daniel Kasen. Kasen is a scientist in Berkeley Lab's Nuclear Science Division, and Tchekhovskoy and Foucart have also served as researchers in the Lab's Nuclear Science Division.
The delay in getting their findings published actually worked to their advantage, Fernández noted, given that interest in visual counterparts to gravitational wave detection and multi-messenger astronomy has grown considerably in the last three years In addition, they used a customized version of HARM – a well-known astrophysics code – called HARMPI, a general relativistic 3D magnetohydrodynamic code upgraded to run in parallel, developed by Tchekhovskoy. Approximations to key physics (neutrino emission, nuclear recombination) were developed by Fernandez, which when added to HARMPI allowed for a more realistic simulation of the disk and the properties of its ejecta.
“Using NERSC resources and this code, we were able to compute at higher resolution and for a longer time than other groups who published before us on three-dimensional accretion around black holes,” Fernández said.
Leaving on a Jet
These resources also enabled the team to demonstrate that when simulations include magnetic fields, the disk ejects twice the amount of material and at higher speeds compared to when these fields are not included, Fernández noted. By modeling the aftermath of the collision in such detail, they have been able to account for both the accretion disk and another way matter is ejected from the collision: carried on an astrophysical jet, a narrow plume of particles and radiation shot out at nearly the speed of light as the stars collide, which is also thought to be the source of the gamma ray burst.
“The simulation’s technical aspects are impressive from a scientific standpoint because the interactions are so complex,” said Fernández. “It was expected that we could find jets, but this is the first time that we’ve been able to model this in enough detail to see this effect emerge.”
NERSC is a U.S. Department of Energy Office of Science User Facility.
About NERSC and Berkeley Lab
The National Energy Research Scientific Computing Center (NERSC) is a U.S. Department of Energy Office of Science User Facility that serves as the primary high performance computing center for scientific research sponsored by the Office of Science. Located at Lawrence Berkeley National Laboratory, NERSC serves almost 10,000 scientists at national laboratories and universities researching a wide range of problems in climate, fusion energy, materials science, physics, chemistry, computational biology, and other disciplines. Berkeley Lab is a DOE national laboratory located in Berkeley, California. It conducts unclassified scientific research and is managed by the University of California for the U.S. Department of Energy. »Learn more about computing sciences at Berkeley Lab.