New issue 1258: potential unit error in cosmology.py https://bitbucket.org/yt_analysis/yt/issues/1258/potential-unit-error-in-cos... Pengfei Chen: I was trying to generate light rays using yt 3.4-dev, but found that the light rays have wrong redshift intervals. The error happens when I make light rays from one single dataset. I reproduced the error with the dataset under src/yt-hg/tests. The code I used is here: http://paste.yt-project.org/show/6751/ After running the code, I did h5ls -d lightray.h5/grid/redshift and found the last redshift was 9.98999107782409. However, the correct value should be 9.97888260334447. When I call LightRay this way, it will call the _deltaz_forward function in cosmology_splice.py, which will call the comoving_radial_distance in cosmology.py: distance2 = self.cosmology.comoving_radial_distance(z2, z) , where distance2 is not in comoving units. Then it will calculate z2: z2 = ((target_distance - distance2) / m.in_units("Mpccm / h")) + z2 . I think the subtraction here is the problem, since it tries to convert distance2 to comoving units, even though the numerical value of distance2 is already comoving. So a wrong factor of (1+z) is multiplied.