NYU Value-Added Galaxy Catalog
Fibers on the same tile cannot be placed closer than 55''. Only about 30% of such pairs are in the overlaps of multiple tiles and therefore we can observe both galaxies. Thus, there is a selection function which correlates with the density field that must be accounted for in large-scale structure statistics.
A simple method of doing so which works surprisingly well on large scales is to simply assign a galaxy which was unobserved due to collisions the redshift of the galaxy with which it collided. We assign such a "fixed" redshift for about 6% of all galaxies. Because the correlation function of galaxies is strong on these scales, it turns out that this is a reasonable procedure for about 60% of galaxies, in the sense that the actual redshift of the galaxy is about 60% likely to be within 10 Mpc (as determined by looking at the statistics of close pairs in overlaps of tiles).
You can imagine going one step further and requiring that the assigned redshift be within 0.05 of the photometric redshift of the fixed galaxy. This results in a smaller number (3%) of more reliable fixes (70% within 10 Mpc). The improvement is not very great, so it is probably not necessary to use this more complex procedure.
The results of the collision corrections are in files with the naming convention:
collisions.[collision_type].fitswhere collision_type can be
The FITS file contains 8 columns:
An ASCII version also exists:
collisions.[collision_type].dat
with columns:
[ra] [dec] [z] [z_spectro] [z_fixed] [neighbor_used] [fixed] [got]