A translator in a meeting room hears the shift in tone before it is spoken aloud. They pick up subtle changes in breathing, pauses, and intonations, and already know that in a few minutes the conversation will turn heated. The X-ray binary 4U 1705-44 has been having the same kind of conversation for thirty years: its radiation flux alternately fades and flares, and so far no one could catch the moment when the calm phase gives way to a stormy one.
An X-ray binary is a pair consisting of a compact object (a neutron star or a black hole) and an ordinary star. Matter flows from the ordinary star onto the compact object, heats up, and shines in X-rays. The authors applied the Koopman operator, a mathematical tool that turns chaotic dynamics into linear ones. To do this, the system state is lifted into a space of observable functions, where evolution becomes a simple multiplication by a number. Each function changes at its own rate, and the slowest functions reveal how the system prepares for a transition.
The translator does not just hear the shift in tone; they can predict when it will happen and prepare the audience for it. Likewise, the new method allows telescopes to be pointed at the X-ray binary in advance, catching the transition across all wavelengths, from radio to gamma rays. This is the first test of Koopman theory on real X-ray data, and it paves the way for interpretable machine learning in high-energy astrophysics, where prediction and understanding emerge from the same decomposition.
4U 1705-44 has been observed almost continuously for thirty years, and during that time more than fifty transitions between high and low states have been recorded, a rare long record for a chaotic system. One of the transitions predicted by the method occurred four days after the eigenfunction changed sign.