OSC helps researchers map the universe
The Ohio Supercomputer Center (OSC) played a critical role in helping researchers discover new insights in the growth of the universe from its infancy to present day.
New results released by the OSC confirm the theory that the present universe is composed of only 4 per cent ordinary matter, 26 per cent mysterious dark matter, and the remaining 70 per cent in the form of mysterious dark energy, which causes the accelerating expansion of the universe.
The findings from researchers at The Ohio State University and their colleagues from the Dark Energy Survey (DES) collaboration are based on data collected during the first year of the DES, which covers more than 1,300 square degrees of the sky or about the area of 6,000 full moons. DES uses the Dark Energy Camera mounted on the Blanco 4m telescope at the Cerro Tololo Inter-American Observatory high in the Chilean Andes.
According to Klaus Honscheid, PhD, professor of physics and leader of the Ohio State DES group, OSC was critical to getting the research done promptly. Honscheid’s colleagues, computational specialists – Michael Troxel and Niall MacCrann, postdoctoral fellows – used an estimated 300,000 core hours on OSC’s Ruby Cluster through a condo arrangement between OSC and Ohio State’s Center of Cosmology and Astro-Particle Physics (CCAPP).
The team took advantage of OSC’s Anaconda environment. Anaconda is an open-source package of the Python and R programming languages for large-scale data processing, predictive analytics, and scientific computing. The group then used its software to evaluate the multi-dimensional parameter space using Markov Chain Monte Carlo techniques, which is used to generate ‘fair’ samples using probability.
The team also ran validation code, or null tests, for object selection and fitting code to extract information about objects in the images obtained by simultaneously fitting the same object in all available exposures of the particular object.
The bulk of the team’s 4 million computational allocations is at the National Energy Research Scientific Computing Center (NERSC), a federal supercomputing facility in California. However, due to a backlog at NERSC, OSC’s role became to the project grew in importance.
According to Honscheid, for the next analysis round the team is considering increasing the amount of work done through OSC. The total survey will last five years, he said, meaning the need for high-performance computing will only increase.
‘Successful collaboration at this scale represents many years of deep commitment, collective vision, and sustained effort,’ said Ami Choi, CCAPP postdoctoral fellow who worked on the galaxy shape measurements.
Michael Troxel, CCAPP postdoctoral fellow, and leader of the weak gravitational lensing analysis, added, ‘These results are based on unprecedented statistical power and detailed understanding of the telescope and potential biases in the analysis. Crucially, we performed a 'blind' analysis, in which we finalised all aspects of the analysis before we knew the results, thereby avoiding confirmation biases.’
The DES measurements of the present universe agree with the results obtained by the Planck satellite that studied the cosmic microwave background radiation from a time when the universe was just 400,000 years old.
‘The moment we realized that our measurement matched the Planck result within 7 per cent was thrilling for the entire collaboration,’ said Honscheid. ‘And this is just the beginning for DES with more data already observed. With one more observing season to go, we expect to ultimately use five times more data to learn more about the enigmatic dark sector of the universe.’
The new results from the Dark Energy Survey will be presented by Kavli fellow Elisabeth Krause at the TeV Particle Astrophysics Conference in Columbus on Aug. 9, and by CCAPP’s Troxel at the International Symposium on Lepton Photon Interactions at High Energies in Guangzhou, China, on Aug. 10.
The publications can be accessed on the Dark Energy Survey website.
Read the rest of the story on the Ohio supercomputer Center website