Optimized BLENDER Code Speeds Up Kepler Planet Analyses
Visualization experts in the NASA Advanced Supercomputing (NAS) Division have improved the BLENDER software package used by the Kepler mission team to validate planet candidates, greatly shortening runtimes on the Pleiades supercomputer. The optimized BLENDER code allows for same-day turnaround of multiple analyses and helps the Kepler team to confirm smaller, near-Earth sized planets orbiting some of the 150,000 stars under constant observation by the Kepler spacecraft.
As more small planets are being identified and observed, the BLENDER code is being used more frequently to validate the data from Kepler. While large planets are more easily confirmed using spectroscopic methods, small planets in wider orbits around their host stars—like the Earth around the Sun—are more difficult to validate using visual data, due to longer durations between their transits and transit signals that are tiny and hard to identify. To identify smaller planets, the Kepler team uses statistical validation via BLENDER by comparing the light curve of the planet candidate with synthetic light curves that replicate other astronomical scenarios—called "blends"—that might mimic a transiting planet.
Inconsistent blend scenarios are rejected, and the remaining light curves are then compared with the statistical frequency of occurrence of planets that size—known as planet priors—to calculate the final probability that the candidate is a true planet. The light curves are also used to project the candidate's size, orbit, and distance from the host star.
Chris Henze, NAS data analysis and visualization lead, optimized BLENDER to make the code work more efficiently with the massively parallel Pleiades supercomputer, and optimized the code's performance on each individual Pleiades core. He also enhanced BLENDER's gridding system by enabling direct search optimization to hone in on more promising locations within the data to analyze. These improvements have resulted in much smoother light curve visualizations for the Kepler scientists to interpret, with much less work required from the supercomputer—shortening runtimes and allowing more planet candidate analyses to be performed.
BLENDER has been used by the Kepler team to identify candidate planets in several planetary systems, many of which are near-Earth sized or smaller. Most recently, they utilized the code to confirm the size and orbit of three super-Earth sized planets orbiting in the habitable zone around their host stars, where the temperature of an orbiting planet could be suitable for sustaining liquid water. Kepler-62e and Kepler-62f orbit a star that is smaller and cooler than our Sun. And at 60% and 40% larger than Earth, respectively, they are the smallest planets found orbiting in a star's habitable zone, to date. The third planet, Kepler-69c is approximately 70% larger than Earth and is the smallest planet found orbiting in the habitable zone of a Sun-like star.
Earlier this year, the Kepler team also used BLENDER to validate three small planets surrounding Kepler-37, a Sun-like host star in the constellation Lyra, approximately 215 light years away. The smallest of the three planets, Kepler-37b is 2,402 miles in diameter, making it only slightly larger than the Moon at about one-third Earth's size. To date, it is the smallest exoplanet discovered around a main sequence (dwarf) star.
Pleiades is an important component of the Kepler data pipeline, performing the most data-intensive computational tasks, including the pre-search data conditioning, transiting planet search, and data validation modules. The planet candidate validation typically utilizes 3,000 nodes and processes light curves in approximately 20 hours—a task that would take more than a month to complete on the computers at the Kepler Science Operations Center.
Additional optimization of BLENDER on Pleiades will further improve the turnaround time of the Kepler data—which will increase in size over the course of the mission—and help in the search for Earth-like planets able to support life.
