Background cosmology neutrinos

Implements neutrinos at the background level. All cosmological quantities are unchanged unless Cosmology:N_ur or Cosmology:N_nu is specified in the parameter file. See theory/Cosmology for details of the calculation.

Only changes outside cosmology.c are

• Replace occurrences of (Omega_m - Omega_b) with Omega_cdm, computed in cosmology.c
• Add a flag space.with_neutrinos to indicate presence of neutrino parts

Thanks. I had a brief initial look. Nice work and documentation.

I'll come back to it in details later but one fundamental that caught my eye is the change to the background density in the FOF calculation. That will change the halo masses and HMF. Is that something we actually want?

In the first instance, it is best if the halo finder is only applied to the cdm particles. My understanding is that mean_matter_density sets the linking length for dark matter particles. If neutrinos are treated separately (with_neutrinos = true), then we should just use O_cdm for that purpose. However, if cdm + nu is modelled as a single fluid (with_neutrinos = false), then it makes sense to use O_cdm + O_nu = O_m - O_b, in the same way that O_cdm + O_b is used in DMO simulations. Do you agree with that?

At this stage, I don't know. :) I think there is no uniquely correct choice here. Ignoring baryons for now and only considering the case where we have nus without particles, we already have a difference in the FOF masses generated when running with O_nu=0 or O_nu > 0. Maybe that is what we want but that means it may make things more complicated to compare.

I guess this boils down to "what do people mean when they say mean inter-particle separation"? That question has a single answer when running plain-old CDM (DMO) runs but that has changed. For instance, EAGLE and OWLS or BAHAMAS don't make the same choice here.

Do you have a small example that I can use to run everything? Say a 64^3 DMO box with nus?

Is there a reason to use a different table length and bounds for the nu-related integrals?

Good stuff overall, thank. Well documented and coded. I have left you some small comments throughout.

added 1 commit

• 8a075047 - Replace pow with cbrt & readability

Compare with previous version

added 1 commit

• 518e4553 - All interp tables same length & substituted equivalent form of O_m - O_nu_0

Compare with previous version

added 1 commit

• ea609f1d - Allow integrating beyond a=1

Compare with previous version

Thanks for the update.

Regarding the last sticking point, I am now convinced we don't want to change the inter-particle separation and hence halo mass definition in the way done here as this will prove problematic for comparisons down the line.

Do you have an alternative way of computing this in the FOF code and VELOCIraptor interface?

added 3 commits

• e69a050c - Changed definitions such that Omega_m = Omega_cdm + Omega_b no longer includes neutrinos
• 80aa2eca - Simplified further
• 25756a69 - Documented new convention to not include neutrinos in Omega_m

Compare with previous version

added 1 commit

• 84413aa8 - Removed unnecessary edits

Compare with previous version

added 1 commit

• 6bb230f6 - Rolled back further changes

Compare with previous version