Cosmic Web Classification through Stochastic Topological Ranking

This paper introduces ASTRA (Algorithm for Stochastic Topological RAnking), a new method for classifying galaxies into cosmic web structures -- voids, sheets, filaments, and knots -- specifically designed for large spectroscopic surveys. ASTRA operates on observed galaxy positions and a corresponding random catalog, generating probabilistic cosmic web classifications for both datasets. The method's key innovation lies in using random points to trace underdense regions, enabling robust identification of cosmic voids that are poorly sampled by galaxies. We evaluate ASTRA using N-body simulations (dark matter-only and hydrodynamical) and SDSS observational data, performing both visual inspections and quantitative analyses of mass and volume distributions. The algorithm successfully produces void catalogs with size functions following theoretical expectations and demonstrates consistent environmental statistics across diverse datasets. Comparative analysis against established cosmic web classifiers confirms ASTRA's effectiveness, particularly for filament identification. By incorporating both observed and random points in its classification scheme, ASTRA provides a full cosmic web characterization without requiring density field interpolation or fixed geometric assumptions. The method's ability to quantify spatial correlations among different cosmic web components offers promising avenues for enhancing cosmological parameter constraints through non-standard clustering statistics.