About

How this atlas is built, and what it can and cannot tell you.

The catalog

The map is built from the DESI DR1 large-scale-structure clustering catalogs (iron, v1.5), which are already filtered to reliable galaxy redshifts. The current build holds 3,957,865 galaxies of tracer BGS_BRIGHT, spanning redshift 0.01 to 0.50 with a median of 0.20.

Downloading, ingesting, and exporting are three separate, re-runnable stages, and the database is rebuilt from scratch each time, so the catalog is a deterministic function of the snapshot and the code that reads it.

Coordinates

Each galaxy's sky position and redshift become a point in three dimensions. Redshift is converted to comoving distance under Planck18, then combined with right ascension and declination to give Cartesian coordinates in megaparsecs, in the equatorial frame: x points at RA 0°, Dec 0°, and z at the north celestial pole. The Milky Way sits at the origin, which is why every view is a view outward from where we happen to be.

Giving the points names

A redshift catalog is a list of coordinates. It does not know that one of its entries is a well-studied barred spiral with a name people have used for two centuries. To recover that, every DESI galaxy is cross-matched on the sky against HyperLEDA, the standard compilation of galaxy designations and properties — 2,211,337 of its galaxies fall in the right patch of sky and the right velocity range to be candidates.

The match is nearest-neighbour within 3 arcseconds, done locally so the result is reproducible rather than dependent on a remote service. It gives 486,546 DESI galaxies an identity — 12% of the catalog — of which 134,843 carry a morphological type and 2,225 carry a designation a reader might actually search for, from UGC (1,189), NGC (662) and IC (374).

There are no Messier galaxies in that list, and the reason is worth knowing: DESI's Bright Galaxy Survey does not observe the very brightest, largest galaxies. They saturate the instrument and their light spills across many fibres, so the famous nearby showpieces — Virgo's giants, Andromeda — are excluded by design. The named galaxies here are the second tier: catalogued, studied, but not household names.

Position alone is not proof, so the match is checked against a second, independent quantity. Where HyperLEDA has its own velocity, it is compared with the DESI redshift: 99.3% agree to better than 1000 km/s, and the median disagreement is a few tens of km/s — the scale of measurement error, not of two different galaxies. The median separation between the two catalogues' positions is 0.20 arcseconds. Matches that fail the velocity test are kept and flagged rather than quietly deleted, so the failure rate stays countable.

Exported views

The viewer never loads the whole database. Each dataset below is a sampled, optionally cut slice written as a compact binary file, so a browser can hold it in memory and draw it every frame.

Honest limits

Three things this map does not show, and cannot be made to show without more work than a rendering change:

• Redshift-space distortion. Positions along the line of sight are contaminated by galaxies' own motions. Clusters appear stretched; the effect is strongest exactly where the structure is most interesting.
• Selection. A brightness-limited survey sees fewer galaxies further away, so the thinning of points with distance is mostly the survey running out of reach, not the universe emptying out. The density field corrects for this; the raw point cloud does not.
• Footprint. DESI covers a specific patch of sky. Sharp edges in the map are the edges of the survey.

Credit

Galaxy catalogs are from the Dark Energy Spectroscopic Instrument. See DESI's data license and acknowledgments. This site is an independent project and is not affiliated with the DESI collaboration.