The Cosmic Web in the DESI Early Data Release: A Probabilistic Environment Catalog

TL;DR

This paper introduces a probabilistic catalog classifying cosmic web environments in DESI data, validated against existing surveys, and provides a new baseline for galaxy evolution research.

Contribution

It presents the first public environment catalog for DESI data using a novel probabilistic classification method without density field reconstruction.

Findings

Classified DESI objects into voids, sheets, filaments, and knots with probabilistic measures.

Validated web classifications against GAMA, COSMOS, and SDSS, showing consistency.

Revealed environmental dependencies of galaxy properties like color and star formation.

Abstract

We present the first public cosmic-web environment catalog built on any DESI data release. Using ASTRA (Algorithm for Stochastic Topological RAnking), we classify each object in the DESI Early Data Release into void, sheet, filament, or knot by combining observed positions with matched random catalogs, without reconstructing a continuous density field. We apply this method to four DESI extragalactic tracers (BGS, LRG, ELG, and QSO) across the 20 EDR rosettes ($\sim 175$ deg$^2$ total), running 100 realizations per tracer-zone pair to derive per-object membership probabilities and classification entropies. We calibrate the classification thresholds using BGS as an anchor to match the volume-filling fractions reported for GAMA, and recover a physically consistent web morphology across all tracers. For BGS, the resulting web-type fractions and the environmental dependence of star formation are consistent with GAMA, COSMOS, and SDSS-based references, validating the method against established benchmarks. A normalized mutual information analysis on BGS reveals a clear dependence of the statistical associations between galaxy color, stellar mass, and specific star formation rate across environments. These results provide a new observational baseline for galaxy evolution studies with DESI. All data products and the open-source pipeline are publicly available.