search

aimalz / prob-z

working with probabilistic redshifts
2 stars 2 forks source link

show me the p(z) for a galaxy that has z_obs outside the z_true range #36

Closed davidwhogg closed 8 years ago

aimalz commented 8 years ago

This is an example from the multimodal test case. The dashed lines are the true redshifts and the dotted lines are {z_obs} (where there are multiple z_obs for some galaxies). samplepzs

davidwhogg commented 8 years ago

None of these is problematic, is it? There are no bad edge effects visible there, are there?

Do those sigma-z values make sense? These are pretty bad photo-zs! Much worse than we expect, no?

aimalz commented 8 years ago

Yes, these are definitely worse than what would be expected for real data. I've experimented with other formulations but never settled on a physically motivated way to modify the standard deviations of the Gaussian components.

davidwhogg commented 8 years ago

Let's verify all code and then return to that issue.