CAMB: changing the primordial power spectrum

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Zhiqi Huang
Posts: 12
Joined: April 22 2006
Affiliation: Sun Yat-Sen University

CAMB: changing the primordial power spectrum

Post by Zhiqi Huang » January 30 2008

Hi Antony,

If I want to change the primordial power spectrum, is it correct that I just change the file power_tilt.f90? I have to make sure that the functional form of primordial power spectrum (i.e. the standard A_s,n_s, n_run, A_T, n_T parametrization) is not assumed anywhere else.

Of course to modify the parametrization, I have to add new parameters to cosmomc. But that's another issue, which I clearly know how to do.

Thanks a lot!


Antony Lewis
Posts: 1594
Joined: September 23 2004
Affiliation: University of Sussex

Re: CAMB: changing the primordial power spectrum

Post by Antony Lewis » January 31 2008

Yes, CAMB's InitialPower module determines the power spectrum, by default implemented in power_tilt.f90. Only smoothness properties are assumed elsewhere.

Pascal Vaudrevange
Posts: 50
Joined: March 26 2006
Affiliation: DESY

CAMB: changing the primordial power spectrum

Post by Pascal Vaudrevange » January 31 2008

If I remember correctly, at some point the Ly-alpha module assumes that there is A_s, n_s only. But you should ask Pat McDonald about this..

Patrick McDonald
Posts: 16
Joined: November 06 2004
Affiliation: CITA

CAMB: changing the primordial power spectrum

Post by Patrick McDonald » February 01 2008

The LyaF constraint in the standard version of CosmoMC assumes that the deviations of the power spectrum from LambdaCDM can be described by a change in the amplitude and effective slope of the power spectrum on scales within about an order of magnitude of k=1 h/Mpc.

My original measurement, which Anze Slosar has a CosmoMC patch implementing, included (relatively weak) constraints on the effective running of the slope on this scale. The standard CosmoMC version just drops this dimension (presumably for ease of implementation of a fortran version), which isn't as accurate, but shouldn't be too bad.

So the assumption isn't that the primordial power spectrum is described by A_s, n_s, (+possibly alpha_s), it is just that it is relatively smooth near k=1 h/Mpc. For example, you can *not* use the standard version for Warm Dark Matter, where the power spectrum is sharply truncated at some scale to which the LyaF is sensitive (we have separate sims for that).


Post Reply