Cosmology from one galaxy in a void?

Understanding galaxy properties may be the key to unlocking some of the most intriguing mysteries of modern cosmology. Recent work relied on machine learning to extract cosmological constraints on $\Omega_\mathrm{m}$ using only one galaxy. But if this is true, how should we select the galaxy to use for cosmology inference? In this paper, we consider selecting a galaxy that lies in cosmic voids, the underdense regions of the cosmic web, and compare the constraints obtained with the ones obtained when randomly selecting a galaxy in the whole sample. We use the IllustrisTNG galaxy catalog from the CAMELS project and the VIDE void finder to identify galaxies inside voids. We show that void galaxies provide stronger constraints on $\Omega_\mathrm{m}$ compared to randomly selected galaxies. This result suggests that the distinctive characteristics of void galaxies may provide a cleaner and more effective environment for extracting cosmological information.