search

LSSTDESC / CCL

DESC Core Cosmology Library: cosmology routines with validated numerical accuracy
BSD 3-Clause "New" or "Revised" License
141 stars 64 forks source link

Interpolation problem at high redshift bins #1035

Open DavidNavarroG opened 1 year ago

DavidNavarroG commented 1 year ago

There seems to be an interpolation issue when computing the power spectrum for Intrinsic Alignments. The problem might come from the nl_pt.PTIntrinsicAlignmentTracer or nl_pt.get_pt_pk2d functions. When defining the IA bias parameters as a function of redshift, there is a bump in the power spectrum at the last bins of the redshift grid.

The next plot shows the problem for a redshift grid from 0 to 5: image

This other plot is for a redshift grid from 0 to 3, the issue moves to the other redshift values, so it is not a physical problem:

image
damonge commented 1 year ago

@DavidNavarroG do you have a piece of code reproducing this?

DavidNavarroG commented 1 year ago

Hi @damonge,

Yes, I have a Jupyter notebook where you can create those plots, I attach it. I have also added a flag called 'z_fixed' that, when set to True, allows to select the IA bias parameters at a fixed redshift. If that's the case, the problem does not appear.

pyccl_interpolation.ipynb.zip