An international team has found a pulsar in public data from NASA’s Fermi Gamma-ray Space Telescope using roughly 10 hours’ worth of computation on a standard processor, NASA reports. An earlier search using the previous best method missed the same object despite consuming the equivalent of 600 processor-years. Source
Those are equivalent computing costs, not elapsed calendar times. Nobody spent six centuries staring at a progress bar.
In its November 19 account, NASA describes the cryptography-inspired method as about a million times faster than previous searches. The agency says a paper describing the results appeared that day in Physical Review Letters. Source
The newly found object, PSR J1622−7204, is an isolated pulsar spinning 261 times per second. Researchers identified it in 15 years of Fermi observations. A pulsar is a rapidly rotating neutron star whose radiation beams sweep toward Earth like a lighthouse. Unfortunately for anyone trying to find one, a fast-spinning beacon doesn’t necessarily deliver an easy-to-read signal. Source
Fermi catches only a tiny number of photons from an individual pulsar, mixed with background gamma rays. Researchers must work out the star’s rotation rate, how quickly that rotation is slowing, and its precise position in the sky. The possible combinations create a crushing computational workload. Source
Borrowing the code-breakers’ toolkit
Study co-authors Dotan Gazith and Barak Zackay of Israel’s Weizmann Institute of Science recast the search as a geometry problem, according to NASA. Their approach involves finding the shortest vector, or step between points, in a high-dimensional lattice. That’s a regularly spaced grid with more dimensions than the familiar three. Source
The mathematical tools came from cryptographers trying to break encryption designed to withstand future quantum computers. That’s an unexpected career change for code-breaking math, but it doesn’t mean the astronomers used a quantum computer: NASA reports computation on a standard processor. Source
“We have nothing to do with cryptography in our analysis here, but those algorithms can help us solve the problem of searching for pulsars in our data,” Gazith said in NASA’s account. Source
After the new method identified a candidate, the researchers checked it using the full dataset to confirm the pulsar’s signal, NASA says. The public archive also mattered: researchers outside the Fermi mission could apply a different mathematical approach to observations already available to everyone. Source
Next up: stars with company
The team now wants to extend the method to pulsars orbiting companion stars. A companion’s gravity distorts the signal, making those searches harder. Finding this isolated pulsar does not mean that next problem is solved. Source
Co-author Colin Clark, an astrophysicist at Germany’s Max Planck Institute for Gravitational Physics, told NASA the researchers hope to combine the algorithm with volunteer computing through Einstein@Home to find more pulsars. Source
Comments