General description: Redshift probability distributions derived using the "weighting" technique of Lima et al. 2008 and Cunha et al. 2009. The method derives weights for a training set of spectroscopically confirmed galaxies such that the distribution of relevant quantities, such as magnitudes or colors, matches that of a set of galaxies without known redshifts (the photometric sample). Assuming these quantities correlate with redshift, and are the only relevant quantities for redshift determination, the resulting weighted redshift histogram is proportional to the redshift distribution N(z) of the photometric sample. This technique is also used to derive the redshift probability distribution for individual galaxies P(z). The files described below contain these P(z) for a magnitude limited sample (r=21.8). These objects are chosen from the "galaxy" sample; in other words they have objc_type == 3, although there are certainly stars in the sample at this magnitude limit. For more details please see the associated paper Sheldon et al. 2011.
For the P(z) files:
For the zbin files:
We release both gzipped FITS files and gzipped ASCII files. The FITS files and ASCII files contain the same data. POFZRUN is a "run" identifier, representing a run of the P(z) code. We also release a small file containing the redshift bin edges for the P(z). The P(z) in the above files are evaluated in these bins.
The data in the FITS files are in a binary table at the second HDU (first extension). Each row of the table represents an object.
Notes: The p(z) should be regarded as un-normalized. The ASCII files have the same data model as the FITS files.
The FITS file data is contained in the second HDU in a binary table. Each row of the table represents a single redshift bin.