A Study of cosmic void properties within redshift z ≤ 0.114 through analytical shape descriptors and deep latent representations

Abstract

In this paper we examined the distribution and morphology of 518 cosmic voids from the SDSS DR7 REVOLVER catalog using a combination of deep learning and conventional statistics. Using ellipticity, prolateness, and radial alignment metrics, we describe the shapes of voids and find that larger voids have increasing anisotropy but decreasing ellipticity (r=-0.526). There is no discernible change in void size or alignment with redshift according to statistical tests. We processed six void properties by a deep autoencoder, which retained 86.62% of the variance. Our combined method offers fresh insights into the dynamics of the cosmic web by confirming that void shapes are shaped by boundary confinement and tidal forces.