Asian Scientist Magazine (Oct. 10, 2022) — From hologram Princess Leia in the internationally beloved Star Wars franchise to Bubs the robot in the popular Korean movie Space Sweepers, which space-themed science fiction story would be complete without a matrix of futuristic tech? Aboard Starfleet vessels, an entire arsenal of computers—even artificial intelligence units—and handheld Personal Access Display Devices were all the rave in Star Trek.
But these fancy technologies are not just some pipe dream set in an unreachable future. In reality, supercomputing is part and parcel of how modern astronomy operates. Cosmology, a branch of astronomy dedicated to unraveling the origins and evolution of the universe, is particularly data- intensive and requires sophisticated computing resources to piece together disparate clues from outer space.
“The integration of astronomy and supercomputing has accelerated the rate at which discoveries can be made. We can process data much faster, detect far more faint signals due to leaps forward in sensitivity and create images at a higher resolution than ever before,” said Dr. Sarah Pearce, deputy director for astronomy and space science at Australia’s Commonwealth Scientific and Industrial Research Organisation (CSIRO), in an interview with Supercomputing Asia.
Pearce is also leading the charge for the Australian arm of the Square Kilometre Array (SKA) project, an international effort to build the world’s largest radio telescope for collecting data over an area of a million square meters.
Technological juggernauts are critical to these astronomical missions, as one small step in high performance computing (HPC) can lead to one giant leap in understanding the cosmos and how our universe came to be.
Twinkling stars and other space objects like asteroids are not just fascinating features that dot the sky, but they also hold many secrets about the fundamental forces in our universe, its immense history and its dynamic evolution. Even the seeming emptiness of outer space should fool no one, as invisible gravitational wavelengths and radio emissions fill the void with a colossal mishmash of signals.
“The sensitivity and design of the SKA telescopes will enable the detection of extremely faint signals emitted shortly after the birth of the universe almost 14 billion years ago,” Pearce explained. “Like a time machine, such technologies allow us to look back to when and how the first stars and galaxies formed.”
Origins aside, the universe is ever boggling the minds of astronomers. For one, it is still expanding—and doing so faster than ever before. The gravitational attraction among galaxies should slow down this expansion, yet a perplexing component called dark energy may be counteracting this force.
To test whether such theories hold weight, astronomers are taking stock of the masses of numerous galaxies and their gravitational disturbances on the path of radio waves. These galactic surveying missions also involve searching for hydrogen gas emissions, believed to fuel star formation.
Sampling an enormous number of galaxies is key to revealing the subtle differences in emission wavelengths and distortions in the radio signals. Accordingly, scientists are tapping into supercomputers to calibrate, transform and analyze all that data as quickly as possible—performing trillions of calculations in the blink of an eye. These measurements can then be used to build models to simulate the cosmological past.
For example, researchers led by Dr. Masato Shirasaki at the National Astronomical Observatory of Japan have turned back the cosmic clock and reconstructed the early universe, running 4,000 simulated universes on the 3.087-petaFLOPS ATERUI II supercomputer.
During the Big Bang, the universe exploded from nothing to one trillion-trillion times its size in a fraction of a second. This cosmic inflation influenced how galaxies and other matter are distributed in space. To retrace this phenomenon, the team stripped the simulated galaxies of their gravitational effects to reduce interference and evolved them to see which one best mirrored the state of the early universe.
“This new method allows us to verify inflation theories using only one-tenth of the amount of data,” said Shirasaki to Supercomputing Asia. “Since less data is needed, it can also shorten the observing time required for future galaxy survey missions.”
Skavenging for signals
To discover the answers to the universe’s grand mysteries, scientists are devising machines that can match this galactic scale and decipher its cacophony of signals. Unlike their optical counterparts, radio telescopes like SKA can detect invisible waves and are not blocked by molecular dust, effectively peering into the “dark” regions where stars and planets are born.
The SKA low-frequency telescope in Western Australia is set to feature more than 130,000 antennas distributed across 512 stations, while the South African contingent will comprise 197 satellite dishes to cover the mid-frequency range.
“SKA will be receiving up to 10 billion data streams simultaneously,” Pearce highlighted. “The supercomputers at our Science Processing Facilities will be integral to keep up with the data pouring in from the receivers 24/7.”
Such extensive equipment can accelerate surveying missions by capturing several large parts of the sky in parallel and at unprecedented sensitivity. But to paint a picture out of the radio data, supercomputers need to correlate and synchronize the signals from the antennas, multiplying them together to generate data objects called visibilities.
“The difficulty is that inside these visibilities, the image of the sky is jumbled together with antenna responses and other radio signals such as from telecommunications devices,” noted Pearce.
From scopes to supercomputers
Supercomputers employ advanced data analytics to disentangle space signals from all the noise, including accounting for minor differences in the instruments used and any “spikes” that appear around bright stars. Through iterative loops of calculations, the machines can convert the radio waves into astronomical images with unrivaled quality and resolution.
Whether filtering out interfering signals or stitching together smaller images to create detailed representations, these complex computing tasks all unfold in real time and are conducted over thousands of radio frequencies. Such a feat, Pearce noted, is possible only because of the sheer power of HPC resources available today.
“Distant galaxies, only glimpsed by very long observations today, will routinely be observed in a fraction of the time. Astronomers using the SKA telescopes will encounter more data than has ever been available in the history of radio astronomy,” she added.
SKA also builds upon longstanding precursor projects from CSIRO, including the Australian Square Kilometre Array Pathfinder (ASKAP) and Murchison Widefield Array. The backbone of these space missions is Galaxy, a real-time supercomputing service for telescopes and astronomy research. Housed at the Pawsey Supercomputing Research Centre in Australia, this 200-teraFLOPS CRAY XC30 system is equipped with Nvidia K20X ‘Kepler’ graphical processing units and Intel Xeon E5-2690 host processors.
SKA’s HPC facilities will boast a collective computing capacity of 500 petaFLOPS and archive over 600 petabytes of data each year. Moreover, the alliance of SKA centers across the globe will be connected through a high-end fiber network that can transmit data at speeds of seven to eight terabits per second—about 100,000 times faster than current average broadband rates.
Collective ambitions, universal futures
By bridging localized efforts to global endeavors, Pearce envisions a more collaborative model for the future of astronomy.
“Deeply rooted in our ethos is the concept of open science,” said Pearce. “After a proprietary period, SKA’s enormous data sets will become accessible for anyone who wants to analyze them, which dramatically boosts the potential for further discoveries.”
Traditionally, a single astronomer or small team would request time to use a telescope for their individual research. Now, scientists and engineers from around 100 organizations across 20 countries are participating in the development of SKA—harnessing shared technological resources as the vehicle for driving advances in space science.
From revealing the secrets of dark matter to mapping the magnetic fields that permeate the universe, HPC systems are set to super-charge the next generation of astronomical observation. By capturing snapshots of space-time, these innovations can empower scientific teams to weave together riveting narratives that transform our understanding of the origin and fate of the universe.
This article was first published in the print version of Supercomputing Asia, July 2022.
Click here to subscribe to Asian Scientist Magazine in print.
Copyright: Asian Scientist Magazine. Image: Unsplash