CAMB: high accuracy matter power spectrum

Use of Cobaya. camb, CLASS, cosmomc, compilers, etc.
Post Reply
Ariel Sanchez
Posts: 4
Joined: September 28 2004
Affiliation: Max Planck Institute for Extraterrestrial Physics
Contact:

CAMB: high accuracy matter power spectrum

Post by Ariel Sanchez » November 24 2006

Hi,

I have a question related to the matter power spectra computed using CAMB. In the plot below I show the output from CAMB for the linear and non-linear P(k) (upper panel) and their ratio with the power spectra for the same cosmology but computed with the Eisenstein and Hu fitting formula without acoustic oscillations (lower panel).
In the lower panel is possible to see a small step in the power spectra close to the turn over. Why is it there? How can I compute a P(k) corrected from this?
Cheers,

Ariel

Note: when computing the non-linear power spectrum I used Min_kh_nonlinear = 0.001 instead of 0.005 in halofit.f90. Without this change the non linear P(k) shows another small step at this scale.

http://star-www.dur.ac.uk/~cmb/camb.ps

Antony Lewis
Posts: 1941
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: CAMB: high accuracy matter power spectrum

Post by Antony Lewis » November 24 2006

The bump in the linear spectrum (at 0.3% level) in Sept 2006 version is due to CAMB not propagating many radiation multipoles for low k values. This is good for the C_l, but does give this small error for P_k.

You can fix it by editing equations.f90 from

Code: Select all

         if (CP%Transfer%high_precision) then
           EV%lmaxnr=max(nint(25*lAccuracyBoost),3) 
         end if
to

Code: Select all

        if (CP%Transfer%high_precision) then
           EV%lmaxnr=max(nint(25*lAccuracyBoost),3) 
           EV%lmaxg=max(EV%lmaxg,nint(min(8,nint(sqrt(scal) *600 * EV%q))*lAccuracyBoost)) 
         end if
I'll include this in the next update so high_precision calculations should be accurate to ~0.1%.

Post Reply