This pull request includes new files with the kernels and dNdz for the secondary evaluations of how challenge entries scale with number of spectra and width of bins.
Following cases:
lens and source z bins both half and a quarter the width of the fiducial case (same number of bins as fiducial case, same underlying dNdz). The half tne quarter refer to the width of the top hat selection function on the true dNdz prior to applying photo-z scatter for the lenses, and the fraction of the galaxies used to create this top-hat selection function for the sources. In both cases, the sigz is scaled down relative to fiducial to avoid the photo-z uncertainty totally smearing out the reduced bin width (sigz=0.006 for lens, 0.01 for sources).
starting from the fiducial set up, dropping either 0, 1 or 2 redshift bins starting at higher redshifts from sources and / or 0, 2 or 4 from lenses. redshift distribution of each bin is unchanged but total dNdz is different as missing higher z bins. 8 different combinations between lens and source configuration possible other than the fiducial case. sigz fiducial in this case.
This also modifies the get_tracer_dNdzs and get_tracer_kernels functions in the base class to allow an optional filename argument so we can run more easily with these new files.
This PR also includes the machinery used to generate the dNdz's from the 'true' dNdz from the SRD.
This pull request includes new files with the kernels and dNdz for the secondary evaluations of how challenge entries scale with number of spectra and width of bins.
Following cases:
This also modifies the get_tracer_dNdzs and get_tracer_kernels functions in the base class to allow an optional filename argument so we can run more easily with these new files.
This PR also includes the machinery used to generate the dNdz's from the 'true' dNdz from the SRD.