NASA Supercomputer Assists the Hunt for Exomoons

A team of 21st-century explorers working for the Hunt for Exomoons with Kepler project, based at Harvard University, are searching for exomoons using data from NASA’s Kepler mission and the Pleiades supercomputer at the NASA Advanced Supercomputing facility at NASA’s Ames Research Center.

The discovery of exomoonsmoons situated beyond our own solar systemwould add to the growing list of celestial objects detected by the Kepler telescope that could potentially harbor life in some form. In the quest to find the first exomoon, HEK astronomers led by David Kipping at the Harvard-Smithsonian Center for Astrophysics have devised a unique, systematic computational approach that requires 5.2 million processor hours on Pleiades. Using their in-house LUNA light curve modeling algorithm and a massively parallel sampling algorithm called MultiNest, the project team simulates billions of possible star-planet-moon configurations and compares the results to the actual Kepler data to look for a good match.

So far, the team has surveyed 56 of about 400 identified Kepler planet candidates that could have a detectable exomoon. Surveying the remaining 340 planet candidates would require about 50,000 hours of processing time per object and would take nearly a decade to complete on smaller computers. Utilizing NASA’s powerful Pleiades systemwhich performs over 3 quadrillion calculations per secondwill speed up this computationally expensive process, reducing the processing time to 30,000 hours per object.

Over the next two years, the team will survey the remaining candidates for exomoons by performing photo-dynamical analysis of the public data from Kepler, consuming about 10 million processor hours on Pleiades. Their results will be used to determine the occurrence rate of Earth-like moons.

For more information about the HEK Project, visit: http://www.cfa.harvard.edu/HEK/index.html

For more information about NASA’s Kepler Mission, visit: http://www.nasa.gov/mission_pages/kepler/main/