Recent data from pulsar timing arrays point to a nanohertz stochastic gravitational-wave background, motivating the search for waves from localized sources. Most methods rely on specific waveform templates, which is computationally expensive and may miss unusual signals. A template-free method is proposed: pulsar time delays are modeled as a Fourier series, with a Lorentzian hyperprior placed on the variance of the coefficients, providing a flexible spectral envelope that captures the signal's dominant frequency and bandwidth. Analytical integration over the Fourier coefficients leads to a Bayesian hierarchical model that jointly determines the source coordinates, its frequency composition, and the stochastic background parameters. To account for unmodeled pulsar noise, the model includes additional flat-spectrum components for each pulsar. Tests on synthetic data demonstrate the method’s robustness and suitability for future PTA surveys, capable of detecting any gravitational-wave events.
Astronomers listen to the cosmos like an old radio: they twist the knob not to find a particular station, but to catch any unusual noise. This approach breaks the mold: instead of hunting for gravitational waves using memorized patterns, they catch everything — without bias.
Pulsars are ultra-dense remnants (neutron stars) that spin and beam narrow radio waves into space, like cosmic lighthouses. When a gravitational wave ripples through the universe, it subtly wobbles the very fabric of spacetime (an effect predicted by Einstein), knocking the pulsar’s steady ticking off beat.
This allows them to spot even unexpected waves: the collision of invisible black holes, or ripples left over from the birth of the universe.
🎯 Some pulsars spin hundreds of times per second, and their pulses are more precise than the best atomic clocks.