Continuous gravitational waves (CWs) from non-axisymmetric spinning neutron stars are among the most interesting, though still undetected, targets of the Advanced LIGO-Virgo and KAGRA detectors. The search for this class of signals is difficult due to their expected weakness and potentially large source parameter space. Stochastic gravitational-wave background (SGWB) searches, by contrast, use cross-correlation techniques to run unmodeled searches for gravitational waves, either stochastic background or persistent signals, including CWs. 

In particular, the All-Sky, All-Frequency radiometer search has been used to efficiently identify outliers in the sky–frequency plane, requiring negligible computational resources compared to traditional CW searches, albeit at reduced sensitivity. CW pipelines have been then tested and used on O3 and O4a data to more thoroughly inspect the interesting portions of the parameter space provided.

I show in particular how the adaptability of the CW implementations of the Hough transform algorithm allows to leverage the high parallelism and computational power of the modern Graphics Processing Units (GPUs), yelding speed-ups of orders of magnitude. The flexibility and computational efficiency of GPU-enabled CW-Hough algorithms make them one of the best candidates to produce, together with SGWB searches, prompt results for the current and future LVK runs.