The Nobel Prize winning direct detection of gravitational waves (GWs) by LIGO in 2015 ushered in a new era of physics and astronomy. This discovery provides us with unprecedented opportunities to explore some of the deepest mysteries of the Universe: from the nature of dark matter to measuring the expansion of the universe, understanding the formation history of black hole populations, testing extreme gravity and exploring the equation of state of matter at super nuclear densities. These investigations critically rely on the availability of accurate theoretical predictions for GW signals to be observed; just like the police needs a finger-print data base for uncovering the origin of finger prints taken at a crime scene. Our project is dedicated to the numerical generation of such templates for a dark-matter candidate named Boson Stars. 

The vast majority of the nearly 100 GW events detected by the LIGO-Virgo-KAGRA network of GW detectors has been in agreement with predictions for vacuum black holes in Einstein’s theory of general relativity. This analysis may, however, be subject to some bias since black-hole binaries are also the best-modelled sources of GWs and therefore dominate the available GW catalogs. Put simply, our main tool is a hammer, so everything we see may look like a nail. Among the classes of GW sources alternative to black holes, boson stars play a special role for two reasons. First, they are among the most popular candidates for dark matter. Second, they are exceptionally amenable to numerical modeling and yet offer a remarkably rich phenomenology. Thanks to this rich phenomenology, boson stars are also an ideal proxy for generic compact objects. 

The key goal of our current work is two-fold. First, we have demonstrated that numerical relativity is capable of generating GW templates from boson stars with as high accuracy as achieved for black holes. Second, we have explored the capacity of the current analysis pipelines of GW observatories to detect boson-star mergers (should these occur in the Universe) and whether we could distinguish them from black-hole binaries. To this end, we have simulated 20 orbit boson-star inspirals, verified convergence and accuracy of the numerical data, and injected the resulting GW signals into the LIGO analysis pipeline Bilby. An example is shown in the figure for the case of two boson stars of 40 solar masses at a distance of 500 Mpc. This signal is very well recovered using a data bank consisting only of black-hole templates. LIGO would comfortably detect this event, although it would not distinguish it from a black-hole source and would also report incorrect masses. For other types of BS binaries, instead, we observe the opposite; their signals are not convincingly recovered, but are markedly different from black-hole signals. Our results demonstrate the importance of generating more comprehensive GW template banks for boson-star binaries. These will allow us to distinguish different types of compact objects and estimate their parameters.