Cosmic Web Classification in the DESI Survey
We are using the ASTRA algorithm to assign each DESI object a cosmic–web environment (void, sheet, filament, or knot) and provide probabilistic labels suitable for downstream science.
4 environmentsVoid · Sheet · Filament · Knot
EDR total657,306 objects
DR1 total497,948 objects
DR2processing in progress!
ASTRA algorithm Per-object probabilities FoF groups for web classes Random catalogs for Poisson noise DESI EDR/DR1/DR2
Why it matters
ASTRA operates directly on DESI data, without reconstructing continuous density fields or introducing smoothing scales, using matched random catalogs to account for survey geometry, masks, and sampling variations.
How it works: ASTRA builds stochastic graphs from the data and matched random catalogs,
compares connectivity patterns, and derives web-type probabilities. See the method paper:
ASTRA (arXiv).
What we release
- Environment classifications: (Void/Sheet/Filament/Knot) for each object.
- Classification probabilities: enables uncertainty-aware analyses.
- Friends-of-Friends (FoF) groups: for selected web classes (e.g., filaments, voids).
- Real vs. random comparisons: to analyze the incidence of Poisson noise at scale.
Catalogs
Collaborators
- Diana C. Zapata-Zuluaga Universidad de Antioquia (Medellín, Colombia)
- Sofía Guevara-Montoya Universidad Nacional de Colombia (Bogotá, Colombia)
- Juliana Hernandez-Hernandez Universidad de los Andes (Bogotá, Colombia)
- Jaime E. Forero-Romero Universidad de los Andes (Bogotá, Colombia)