In 1605, Johannes Kepler announced his first law of planetary motion, essentially stating that planets move around the sun with an elliptical, rather than circular, orbit. Around the same time, the Age of Discovery witnessed countless European ocean voyages documenting an enormous – and often falsely frightening – wider world.
These forays into global exploration produced all sorts of skewed maps and mythical creatures, such as the Kraken, a giant sea monster – thought to have been inspired by real sightings of giant squid – and greatly feared among sailors of the day. Now, more than 400 years later, two magnificent machines bearing the namesakes of Kepler and Kraken are making new waves into the next great frontier: deep space.
Today’s Kraken is an XT5 high-performance computer. But, instead of devouring sailors, this monster favors numbers. It’s capable of more than a petaflop (a thousand trillion calculations per second) and managed by the University of Tennessee’s National Institute for Computational Sciences (NICS) for the National Science Foundation (NSF).
Kraken is used to measure the properties of the stars orbited by potential Earth-like planets, properties such as radius, mass, age, and bulk composition, or the proportions of individual gases throughout the star. This mountain of data comes from NASA’s Kepler space telescope, which is currently hunting for Earth-like planets throughout the Milky Way, surveying a multitude of stars to determine how many might support orbiting, Earth-like planets.
Two things are required for a planetary body to be labeled ‘Earth-like’: its orbit must reside within the habitable 'Goldilocks Zone' of the host star, a distance suitable for water, and possibly life, to exist; and it also must be roughly ‘Earth-sized,’ meaning no more than 25 percent larger than the radius of the Earth.
Kepler grabbed headlines with the discovery of Kepler 22b last year, the first planet discovered by the telescope that resides in the Goldilocks Zone. However, it failed the size test with a radius roughly 2.4 times that of Earth’s.
Judging a planet’s size can be tricky. The only method for doing so is by comparing it to the star it’s orbiting. This task is currently being taken up by a team led by Travis Metcalfe of the Space Science Institute in Boulder, Colorado, US. Metcalfe was on the team that catalogued both Kepler 22b and its host star.
An astronomer by training, Metcalfe was attending a conference on seismology five years ago when he heard about Science Gateways for the first time. Science Gateways is a network of high-performance computers, high-end visualization, and data-analysis resources open to researchers across the US. The concept was promoted by the NSF’s former TeraGrid program and is currently supported by the NSF’s Extreme Science and Engineering Discovery Environment (XSEDE). It provides easy-to-use, leading-edge tools to researchers from a range of scientific fields.
In Metcalfe’s case, his Asteroseismic Modeling Portal (AMP) features as the Science Gateway centerpiece: an easy-to-use interface coupled with a low-level artificial intelligence algorithm that allows users, such as those managing the Kepler program, to quickly attain much-needed stellar data.
For example, the data gathered by Kepler can be easily plugged into the AMP interface and the observed star modeled, thus giving researchers precious clues as to a star’s true identity, such as its radius, mass, bulk composition, or gas mixture, and age, which is especially important because quantifying a star’s age using traditional, observational methods presents unique difficulties.
This data gives astronomers a better idea of the qualities of any orbiting planets, a valuable tool in the quest to better understand the wider universe. It’s only in the last year that the team has been applying the AMP gateway to exo-planet host stars, or those that might support planets outside of our solar system. And Kepler keeps the work coming; the space telescope is sending data for hundreds of stars every month. “We’re analyzing five, ten stars at a time,” said Metcalfe, adding that his team is trying to keep up: “Kepler is specifically built for this purpose... we’ve never had this much data before.”
Metcalfe’s team has been using Kraken for three years now. So far, the team has classified around 100 stars including several with planets, said Metcalfe, and the team hopes to do 100 more this year alone. According to the team's findings, Kepler 22b, the latest planet to be classified, could even harbor water.
While the individual jobs themselves aren’t especially large, only requiring 512 of Kraken’s 100,000-plus cores, the team needs the somewhat small allocation for days at a time. And, when doing five to 10 stars at a time, the jobs can utilize more than 5,000 cores, a significant run by most measures.
Who knows what discoveries lay ahead? Kepler has only been in space for three years, but its mission has recently been extended by NASA for an additional four years. During the extended mission, all of the data becomes public, benefitting the entire scientific community and leaving little doubt that Kepler, AMP, and Kraken will continue to be instrumental in getting to know our galactic stellar and planetary neighbors.