New Astrophysics Code Models Stellar Collision Faster

A stellar collision event - where two stars usually circle each other before eventually merging - is a rare phenomenon of significant interest in astrophysics. Now, a new code is capable of generating models of these systems with greater speed and efficiency.

Called Octo-Tiger, the new astrophysics code models a stellar collision by simulating the progress of these self-gravitating systems through adaptive mesh refinement and a breakthrough method of parallelizing the program, resulting in greater expediency compared to existing codes for numerically simulating collision parameters.

The project is a collaborative effort including experimental computer scientists from the Louisiana State University Department of Physics & Astronomy and its Center for Computation & Technology, Indiana University Kokomo, and Macquarie University in Australia. Octo-Tiger is the culmination of more than a year conducting benchmark tests and running simulations.

ALSO READ: Exotic Peacocks of the Stellar World: A New Model for Colliding-Wind Binaries

A 'Reliable Computational Framework'

"Thanks to a significant effort across this collaboration, we now have a reliable computational framework to simulate stellar mergers," said Indiana University Kokomo physics professor Patrick Motl in a LSU press release. He explains that the reduction of the time it takes to run a simulation of a stellar collision allows them to stiart looking at questions that were not possible before, when the simulation for a single-merger event was "precious and very time consuming. The Indiana University professor adds that with Octo-Tiger, researchers can now explore greater parameter ranges or examine a stellar collision simulation with very high resolutions, at longer periods of time.

Researchers published their findings in the Monthly Notices of the Royal Astronomical Society in a report titled "Octo-Tiger: A New, 3D Hydrodynamic Code for Stellar Mergers That Uses HPX Parallelisation." The report details the researchers' investigation on the code performance and precision through simulations and testing. Researchers compared the stellar collision model generated by the code, comparing it to other analytic solutions from existing grid-based codes such as the popular FLASH code. Furthermore, researchers computed the interaction occurring between two white dwarfs from its early mass transfer stage up to the stars' merger, comparing the results with similar simulations.

Octo-Tiger's Promising Performance With Larger Models

Orsola De Marco, one of the authors of the study and a professor at Macquarie University, shared that Octo-Tiger runs on a core count of over 80,000. It was tested on Gadi, Australia's fastest supercomputer and ranks 25 in the World's Top 500, and displayed promising results for larger models of stellar collision.

According to the LSU release, the new astrophysics code is being optimized to simulate a stellar collision between well-resolved stars, like main sequence stars or white dwarfs, that can be approximated by barotropic structures. A correction algorithm in the program improves the precision of the machine and helps the gravity solver conserve angular momentum in the process. Octo-Tiger also utilizes HPX parallelization, which allows work and communication to run simultaneously and leads to excellent scaling properties.

"This paper demonstrates how an asynchronous task-based runtime system can be used as a practical alternative to Message Passing Interface to support an important astrophysical problem," said research scientist Patrick Diehl.