Enhancing dark siren cosmology via Gaussian process reconstruction of incomplete galaxy catalogs

The universe is expanding, and the speed of that expansion is a number scientists call the Hubble constant. Knowing this number precisely is crucial because it tells us how old the universe is and how it has evolved over time. However, measuring it has become a source of great tension in modern physics. When scientists look back at the early universe using the cosmic microwave background, they get one value. When they look at nearby stars and supernovae, they get a different, faster value. This disagreement is so large that it cannot be explained by simple measurement errors; it suggests our understanding of the universe might be incomplete. To solve this puzzle, astronomers are turning to a new tool: gravitational waves. These are ripples in spacetime caused by violent cosmic events, like the collision of two black holes. Unlike light, these waves carry a direct measurement of how far away the event happened, acting as a "standard siren" that does not need to be calibrated against other objects. The missing piece of the puzzle, however, is the distance's partner: the redshift, or how fast the source is moving away from us. Since the waves themselves do not reveal this speed, scientists must find the host galaxy of the collision and measure its redshift. But finding the right galaxy is difficult because the signals from these collisions are often fuzzy, pointing to a large area of the sky filled with thousands of potential hosts, many of which are too faint to see.

In a new study, researchers have developed a clever way to fill in the gaps of these incomplete galaxy maps to get a better handle on the expansion rate of the universe. The problem they tackled is that current methods for dealing with missing galaxies are too simple. When a telescope cannot see every galaxy in a specific direction, the standard approach assumes the missing ones are spread out evenly, like sand scattered uniformly on a beach. This assumption ignores the fact that galaxies are actually clumped together in groups and filaments, with empty voids in between. These clumps and gaps contain vital information that helps pinpoint the true location of a gravitational wave source. By treating the missing data as if it were perfectly uniform, scientists were effectively blurring out the very features that make the measurement precise. The team, led by Matteo Tagliazucchi and colleagues, proposed a different strategy. Instead of guessing where the missing galaxies are, they used a statistical tool called a Gaussian process. Think of this tool as a smart interpolator that learns the pattern of the galaxies it can see and uses that pattern to guess the shape of the distribution where the data is missing, preserving the natural clumps and voids rather than smoothing them over.

The researchers tested their idea using a massive collection of simulated data. They created a virtual universe containing millions of galaxies and then simulated the gravitational waves that would be detected by future observatories. They deliberately made their galaxy catalogs incomplete, mimicking the limitations of real telescopes that miss faint objects. In these simulations, they compared the standard "uniform" method against their new Gaussian process method. The results were striking. When the galaxy catalog was missing about three-quarters of its galaxies, the new method produced constraints on the expansion rate that were 37 percent more precise than the standard method. In the most challenging scenarios, where the data was very sparse, the improvement reached 66 percent. The new method worked by recognizing that the galaxies the telescope did see were part of a larger, structured pattern. It reconstructed the hidden parts of the map by following the density of the visible ones, effectively restoring the "clumpiness" that the standard method had erased. This allowed the researchers to narrow down the possible locations of the gravitational wave sources much more effectively.

The study also explored how different levels of error in measuring galaxy distances affected the outcome. They found that even when the measurements of the galaxies themselves were quite uncertain, the new method still outperformed the old one. This is because the Gaussian process is flexible; it learns the correlation scale of the galaxy distribution directly from the data, rather than relying on rigid assumptions about how galaxies should behave. It does not need to know the complex physics of how galaxies form or how they are distributed in dark matter halos. It simply learns the shape of the distribution from the observed points. The researchers noted that the method is particularly powerful when looking at very distant events, where galaxy catalogs are most incomplete. In these deep reaches of the universe, the standard method fails to capture the structure of the cosmos, while the new approach can still recover the underlying density features.

While these results are promising, they come from simulations, not yet from real observations. The authors acknowledge that applying this to real data will require further development, such as combining the reconstruction of the galaxy map with the measurement of the expansion rate in a single, unified step. They also plan to test the method on actual data from current and future surveys. For now, the work demonstrates that by treating the missing pieces of the cosmic puzzle with more statistical sophistication, scientists can extract significantly more information from the same amount of data. This approach offers a path forward to resolving the tension in the Hubble constant, potentially revealing whether the universe is expanding faster than our current models predict or if there is a deeper mystery waiting to be uncovered. The key takeaway is that the universe is not a smooth, featureless expanse, and by respecting its natural structure even in the data we cannot see, we can learn more about its fundamental nature.