Big Red 200 supports research on deep space star cluster

Astrophysicists Enrico Vesperini from Indiana University and Francesco Calura from INAF-OAS in Bologna, Italy, collaborated using the Big Red 200 supercomputer to build better models of star clusters from early in the universe’s history.

Enrico Vesperini, professor of astrophysicsLooking into space, the farthest star clusters represent glimpses into the history and provide a material source of comparison to the largely unknown mechanics of star formation in the universe. Using data from new high-resolution images (from the Hubble Space Telescope and the Multi Unit Spectroscopic Explorer (MUSE) instrument on the Very Large Telescope (VLT) of the European Southern Observatory), the team employed IU Research Technologies’ High Performance Computing (HPC) system Big Red 200 to model how ancient stars originate within their clusters and to help shed light on the nature of the densest stellar aggregates. In the future, new observations from the James Webb Space Telescope will further constrain their models.

Ancient star clusters are made of ordinary, baryonic matter that composes less than five percent of all the mass in the universe. The rest is made up of dark matter and dark energy. By investigating the stars’ motions within their clusters using hydrodynamic models of gas, movements which adhere to the principles of classical physics, researchers hope to learn more about the formation of such systems. “By zooming into a cluster using the unprecedented resolution of a state-of-the-art HPC,” explained Calura at a seminar hosted by Indiana University’s Research Technologies, “the simulated star formation becomes less homogenous and more complex.”

Francesco Calura, an astrophysicist at INAF-OAS in Bologna, Italy

To model the formation of star clusters in the early universe, the international research team employed the RAMSES software to first algorithmically “discretize” the pressure and density of the gas within a cluster and account for the gravitational forces of the objects. This domain decomposition algorithm effectively allows one to map a large amount of information regarding the cells and stars and pass it to the multi-cores of the HPC system, to integrate the set of conservation equations. This is computed on a nested grid using the technique called adaptive mesh refinement to develop a model of the system. To accurately process this information, the physicists split the work between tens of Big Red 200 computing nodes.

Vesperini said the project benefited from the option of scaling up our computational node use as needed, without adding costs. “As we improve upon these simulations, we have the opportunity to increase scientific understanding of these far away stellar systems,” he said. ”The engineers at Research Technologies have been instrumental in troubleshooting our simulations’ load balancing issues, and we look forward to further refining our computational models in the future.”

Find out more about Big Red 200.