Gravitational waves have been detected by the LIGO (Laser Interferometer Gravitational-wave Observatory) and Virgo interferometric detectors. The gravitational wave events detected thus far have been from the collision and merger of two black holes (e.g. GW150914), or two neutron stars (e.g. GW170817). The detectable signatures of events of this type are typically very short, lasting from much less than a second to a few minutes.

Continuous gravitational waves are a different, hypothesised class of gravitational waves which have so far not been detected. Unlike the short-lived chirp events from binary black holes and binary neutron stars, continuous gravitational waves are expected to be always present, like a background humming sound, in the gravitational wave detectors. See here for illustrations and an example sound bite of such signals.

The most likely sources of continuous gravitational waves are spinning neutron stars in our Galaxy. A small bump on the star's surface a few millimetres in height, or a slight wobble in its rotation, could generate gravitational disturbances that would be detectable as gravitational waves here on Earth. These waves would, however, be much weaker than those generated by binary black hole or binary neutron star collisions, which makes them much more challenging to detect.

The youngest neutron stars which could be radiating continuous gravitational waves are likely to be hundreds of years old, if not much older. However, it is also possible that a very young neutron star — born in the collision and merger of two neutron stars — may also radiate continuous gravitational waves for a short period of time.

A detailed review article about sources, search methods and results is arXiv:2206.06447 (Riles 2022).

In searching for continuous gravitational wave sources, we must make corrections for the intrinsic change in the frequency of the waves over time, the Earth's rotation and orbital motion around the Sun, and in some cases the motion of the source itself if it is in orbit around a companion star. As the timespan of the data being analysed increases, these corrections become very computationally challenging to perform when all or some of the parameters of the source (e.g. its sky position and intrinsic frequency) are not known a priori. For example, an all-sky survey for sources spinning at detectable frequencies, if carried out using the most sensitive search algorithm, would easily exhaust the capability of all computers in existence! For these kinds of searches, we must therefore make use of different algorithms which are not the most sensitive, but are computationally cheap enough to run on modern supercomputers.