An old lighthouse at the edge of the harbor blinks with the precision of a clockwork mechanism. Every few hours its lamp flares and dims, and sailors recognize that rhythm from afar, using the strength of the light to tell where the shore lies. Stars of type RR Lyrae behave in much the same way: they pulsate with periods ranging from a few hours to a day, and their brightness varies by tens of percent. Astronomers have long used such stars as "standard candles," meaning objects of known true brightness, to measure distances to far-off clusters. But this method has a weak spot: to calculate the true brightness, you need to know the star's chemical composition, and that can't always be pinned down precisely.
A new study proposes skipping that step. Instead of measuring chemical composition, researchers trained a neural network to look at the light curve itself: the graph showing how a star's brightness changes over time. The network works like a seasoned musician who, from the rhythm and timbre of pulsations, determines how loudly an instrument is playing, even without seeing it up close. For training, they took about 80,000 light curves from the Gaia DR3 catalog: 63,000 stars of one pulsation type and 16,000 of another. Each curve was carefully phase-folded, aligned, and normalized so that the network would see only the shape of the pulsations, not random differences in brightness.
The result proved more accurate than previous methods. For individual stars, the distance error was about 2.5% for one pulsation type and 1.7% for the other. When the method was applied to globular clusters, dense groups of hundreds of thousands of old stars, the accuracy rose to 1%. That is 1.8 times better than traditional calculations based on chemical composition. The neural network also determined the distance to the Large Magellanic Cloud, a neighboring dwarf galaxy, with an uncertainty of less than 0.02 magnitudes, which agrees with previous measurements.
Why does this matter? Distances to stars underpin the entire distance scale of the Universe. They determine the value of the Hubble constant, which describes the rate at which the cosmos expands. Today, local measurements of this constant disagree with calculations based on the cosmic microwave background, and this discrepancy is called the "Hubble tension." The new method offers an independent way to check the distance scale without relying on assumptions about stellar chemical composition. In the coming years, telescopes like LSST will discover millions of new variable stars, and a neural network like this one could quickly turn their light curves into a map of the Galaxy.
That very lighthouse at the edge of the harbor can now be "read" without a chemistry handbook for its lamp. All you have to do is listen to the rhythm of the flashes, and the neural network will tell you how far the shore is. That's the shift: astronomers used to measure composition to calculate brightness, but now the very shape of the pulsations reveals brightness directly. The error that used to accumulate at the intermediate step has vanished, and old stars speak more clearly.
RR Lyrae stars pulsate with periods from 0.2 to 1.2 days, and their brightness varies by 0.2–2 magnitudes. In globular clusters they are so numerous that they are called "beacons of poor old systems."
In Isaac Asimov's novel "The Gods Themselves," an alien civilization uses pulsating energy sources for interstellar communication. Our RR Lyrae are natural pulsars in the optical range, and their rhythm really can be turned into a distance code.