Closed alexander-mead closed 11 months ago
@alexander-mead This is a really nice fix! @tilmantroester It seems having this included in pyhmcode would also be useful for the nonlinear task force in Euclid.
I'm happy to accept pull requests ;)
Ha! I can probably do this next week.
This should now be covered with #9
There was a small problem with the predictions from the HMcode-2020 feedback model at very low k. These are described here: https://github.com/cmbant/CAMB/issues/135. HMcode-2020 with feedback would erroneously predict ~10% level effects on the power at k ~ 1e-4 h/Mpc.
I was able to fix these by changing the form of the one-halo term at large scales, which goes into the HMcode prediction for feedback, which is modelled as a response. The change can be seen here:
Before:
After:
You will need to pull the latest version of my Fortran library to have access to this fix, and also change the halo model corresponding to HMcode-2020 feedback from
HMcode2020_feedback
(number 103 in my numbering convention) toHMcode2020_feedback_lowkfix
(number 141).