Imagine the hum of a crowded square (that's what the gravitational wave background of the Universe sounds like, as detected by a network of pulsars) neutron stars that emit regular radio pulses with the precision of atomic clocks. Hidden within this cosmic din are the 'fingerprints' of individual pairs of black holes orbiting each other. Each such pair, as it emits gravitational waves, leaves a unique spatial pattern in the pulsar network — a kind of gravitational 'fingerprint.' The geometry of this pattern depends solely on the source's position on the celestial sphere and is unrelated to the signal's intensity. By comparing data collected from dozens of pulsars, scientists can pick out the sought-after pattern even when the overall gravitational background is hundreds of times louder than the signal of interest.

How exactly does this work? The method relies on analyzing cross-correlations in the time series of pulsars. As a gravitational wave passes through space, it causes tiny but coherent shifts in the rotation periods of different pulsars. For a pair of black holes, these shifts form a characteristic 'map' on the sky — a spatial correlation function. By calculating the degree of consistency of fluctuations across many directions, researchers reconstruct this map, and from its shape they determine the source's position.

Tests on numerical models have shown that accounting for these correlations boosts detection confidence by hundreds of times compared to methods that ignore the spatial structure. Moreover, the source's position can be determined 11 times more precisely, like narrowing the search area from half the sky down to the size of a single constellation. Importantly, the method relies not on the detailed shape of the gravitational wave but on its geometric imprint — a stable and reliable indicator that is largely insensitive to the specifics of the emitting system.