CosmoMC: Matter power spectrum and mpk.f90

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Davide Pietrobon
Posts: 7
Joined: January 11 2007
Affiliation: University of Rome, Torvergata

CosmoMC: Matter power spectrum and mpk.f90

Post by Davide Pietrobon » March 13 2008

Hi Antony,

I'm studying dark energy equation of state using matter power spectrum.
First, I modified the dark energy equations, both background and perturbations: the cls power spectrum is fine, and the matter power spectrum seems good as well. Is the latter computed taking the square of the total density perturbation? so that it is enough to modify equation.f90?

If this is right, when I run cosmomc with use_LSS =T I get a negative value for LnLike in mpk.f90, that is pretty weird. Actually, matter power spectrum is very sensitve to cdm/de EoS, expecially if it is negative the matter power spectrum explodes. Could this cause some problem in the LSS_mpklike routine?
Sometimes the output of the same routine is NaN, that actually seems to be translated by calclike.f90 into LogZero. Can I trust the translation? :)

Any suggestion?

Related to that, I've seen that in equations.f90 in the subroutine outtrans when you compute \delta_C in grho you put only the matter. In the latest version of camb inside cosmomc, you added a line, that is commented actually, where in the presence of DE you add its contribution. This actually affects sigma8 only, not the shape of the matter power spectrum.
Which is the correct choice for that?

Thank you very much.

Davide

Many thanks

Davide

Antony Lewis
Posts: 1501
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: CosmoMC: Matter power spectrum and mpk.f90

Post by Antony Lewis » March 14 2008

The matter power spectrum is the total, yes (baryons + CDM + neutrinos).

I don't know about the errors. The normalization of the log likelihood is arbitrary so could be negative and still correct. NaN sounds bad.

What exactly you want to compute probably on how you are defining things (e.g. depending on what you want to compare it with - what you are observing). The power spectrum is usually defined in terms of matter densities only, not \delta_\rho/\rho_total. If you change the definition there will just a different biasing function relating it to the observation.

Davide Pietrobon
Posts: 7
Joined: January 11 2007
Affiliation: University of Rome, Torvergata

CosmoMC: Matter power spectrum and mpk.f90

Post by Davide Pietrobon » March 14 2008

Ok. Great!

Thank you very much, Antony

One more question: the data of SDSS and 2dF are expressed In terms of \delta\rho_c/\rho_c? So in order to compute a power spectrum useful for cosmomc i need to be consistent with that, don't I?

Thanks

Davide

Post Reply