Cls numerical accuracy in CosmoMC

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Alex Zucca
Posts: 4
Joined: June 24 2015
Affiliation: Simon Fraser University

Cls numerical accuracy in CosmoMC

Post by Alex Zucca » March 17 2016

Hi Antony,

I'm working on primordial magnetic fields and I have a CAMB code that works fine: compensated and passive modes are computed correctly. However as soon as I try to compute them in cosmomc, I face some issues. In particular the scalar compensated power spectrum appears to be full of spikes (I used the same parameters in CAMB and it produces a smooth power spectrum). I was wondering if cosmomc passes a different numerical accuracy to the camb subroutines (either CAMB_GetTransfers or CAMB_TransfersToPowers). If so, where are these parameters set?

Many thanks,
Alex Zucca

Antony Lewis
Posts: 1547
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: Cls numerical accuracy in CosmoMC

Post by Antony Lewis » March 17 2016

Accuracy settings will vary depending on what data and settings you are using. I would write out the relevant things in CAMB and see how they differ when run with cosmomc.

Alex Zucca
Posts: 4
Joined: June 24 2015
Affiliation: Simon Fraser University

Cls numerical accuracy in CosmoMC

Post by Alex Zucca » March 21 2016

Hi Antony,
thanks for your help. I managed to check all the parameters and the accuracy settings. In particular I checked the computation until the iCl_scalar array (before interpolation) and that is correct. In some way the interpolation process when called in CosmoMC is having some problems, but I cannot understand why.

Alex

Post Reply