Jspace News

Space news: launches, missions and the sky above

Astronomy

A method borrowed from cryptography finds a gamma-ray pulsar the best previous search of Fermi data missed

PSR J1622-7204 spins 261 times a second. A November 19 paper reports finding it in about ten CPU hours; the best earlier search failed after some 600 CPU years.

By Astronomy Desk ·

Section
Astronomy
Report
News
Reading time
3 minutes
Sources
2, listed below
Checked

In brief

  • PSR J1622-7204, a lone pulsar spinning 261 times a second, was found in 15 years of data from the Fermi telescope's Large Area Telescope with a new lattice-based search.
  • The paper, published in Physical Review Letters on November 19, 2026, reports about ten CPU hours for the find, where the previous best search failed after about 600 CPU years.
  • The method treats the search as a shortest-vector problem in a high-dimensional lattice, with algorithms that come from cryptographers' work on encryption meant to resist quantum computers.
  • The authors write that the approach may open up pulsars in binary systems, which gamma-ray searches have not been able to find without detailed knowledge of their orbits.

A pulsar that the best existing search technique could not find has turned up in data from NASA's Fermi Gamma-ray Space Telescope, with the help of algorithms that cryptographers use to try to break codes. The object, PSR J1622-7204, is an isolated pulsar that turns 261 times a second. It was picked out of 15 years of observations by Fermi's Large Area Telescope. In a paper published in Physical Review Letters on Thursday, November 19, 2026, the six authors report that the search took about ten CPU hours. A semicoherent search, until now the most advanced approach, had run for about 600 CPU years without finding it.

Why gamma-ray pulsars are hard to find

A pulsar is a neutron star, the dense, fast-spinning core left behind when a star explodes, and its beams of radiation sweep past Earth the way a lighthouse beam sweeps past a ship. In gamma rays the signal is extremely sparse. The paper notes that typically only one photon is detected for every hundred million or so rotations, and the handful that do arrive are buried in background radiation from other sources.

To pull the pulse out of the noise, a search has to get several properties of the star right at once: its exact position on the sky, how fast it rotates and how quickly that rotation is winding down. No computer could test every possible set of values. The usual approach has been to split the work into small pieces for volunteers worldwide to run on their own computers; one discovery typically took hundreds of years' worth of processor time.

A geometry problem in many dimensions

Two of the co-authors, Dotan Gazith and Barak Zackay of Israel's Weizmann Institute of Science, developed the method. Their idea was to turn the pulsar hunt into geometry. In their version, the search becomes a matter of finding the shortest vector in a lattice, an orderly grid of points spread over many dimensions. The algorithms it relies on came out of cryptography, where researchers built them while testing whether codes meant to hold out against future quantum computers could be cracked. Satellite navigation is another field that has applied the technique, to make its systems more efficient.

Little hardware was needed. Each attempt took roughly five minutes of processing on a single core, well within what a home laptop can do. When a candidate emerged, the team checked it against the whole Fermi data set, and the pulsation stood out too clearly to be a fluke.

The search for PSR J1622-7204 in numbers
MeasureFigure
Data searched15 years from Fermi's Large Area Telescope
Spin rate261 rotations a second
Photons detectedTypically one per hundred million or so rotations
New lattice-based searchAbout 10 CPU hours
One trial runAbout 5 minutes on a single core
Previous best (semicoherent) searchFailed after about 600 CPU years

Who did it, and what comes next

The lead author is Daniel Pook-Kolb of the Max Planck Institute for Gravitational Physics (Albert Einstein Institute) and Leibniz Universität Hannover in Germany. His co-authors are Colin J. Clark and Bruce Allen of the same institutions, Gazith and Zackay of the Weizmann Institute, and Erik Agrell of Chalmers University of Technology. The paper appears in volume 137 of the journal as article 211201, under a Creative Commons Attribution license.

Next, the researchers want to go after pulsars locked in orbit with a companion star. In such pairs the companion's pull warps the pulsar's signal, and conventional gamma-ray searches have had little chance of finding them. In the paper's abstract, the authors write that their approach may lead to many new gamma-ray pulsar discoveries, binaries among them, which they call a previously intractable problem unless the orbit is already known in detail.

An open archive, searched again

Fermi has been in orbit since 2008 and covers the whole sky once every three hours in gamma rays, the most energetic form of light. Its data have been public since the end of a first year set aside for calibration, and new observations follow once processing is done, so groups with no role in the mission can test their own techniques on the archive. This team did so, preparing its data with Fermitools, the mission's openly available analysis software. The number of gamma-ray sources in Fermi's catalog, a few hundred at first, has since reached the thousands.

This item was prepared with the help of AI tools and draws on NASA material, listed under Sources. NASA has not reviewed or endorsed it and is not responsible for its accuracy.

Sources

  1. NASA Science Data Powers New Method to Discover Hard-to-Find Pulsars NASA Science, November 19, 2026
  2. Gamma-Ray Pulsar Discovery Using a Lattice-Based Method (Physical Review Letters 137, 211201; Crossref record, DOI 10.1103/vlt2-xzgk) American Physical Society, via Crossref, November 19, 2026

Researched and drafted with AI assistance, then checked against the 2 sources above in a separate fact-check on November 20, 2026. Spotted an error? Tell us.