View previous topic :: View next topic 
Author 
Message 
Giacomo Baso
Joined: 26 May 2016 Posts: 3 Affiliation: University of Padua  Department of Physics and Astronomy

Posted: October 06 2016 


I'm a PhD student in computational cosmology but I'm new to CAMB, so forgive me if the question is trivial.
I have a need to compute the growth factor with CAMB, in preparation for a work with massive neutrinos, for which I don't have a analytical approximation in the transition region ~ k_{nr}.
From the README the output file <root>_transfer_out.dat contains the values of Delta_X ≡ (delta rho_X)/rho_X at the desired redshift, so it should be possible to run CAMB at a set of different redshifts and just take the ratio of Delta_tot(z=0) and Delta_tot(z) to obtain the growth factor at redshift z.
In a pure ΛCDM universe this ratio should not depend on the scale k. However, I find something like this:
The red horizontal line is the prediction from the linear theory, using the approximation from Lahav & Suto (2004): D(a) ~ a g(a)
Highk cutoff notwithstanding, there are two different values at different scales, neither of which is equal to the linear prediction. The difference gets bigger at higher redshifts, i.e. Δ D ~ 0.2 at z=20 and Δ D ~ 6 at z=99. Am I missing something? Or is there some physical effects at play here, or maybe a numerical instability?
I should also mention that CAMB gives me the following error every time I run it:
Code:  Note: The following floatingpoint exceptions are signalling: IEEE_UNDERFLOW_FLAG 
The code was compiled with gfortran 5.4.0
I have the last version of CAMB, with the following parameter file:
Code:  DEFAULT(params.ini) # Version 0.1.2
#output_root is prefixed to output file names
output_root = planck/05.00
#What to do
get_scalar_cls = T
get_vector_cls = F
get_tensor_cls = F
get_transfer = T
do_lensing = F
#Main cosmological parameters, neutrino masses are assumed degenerate
use_physical = F
omega_baryon = 0.04825
omega_cdm = 0.25886
omega_lambda = 0.69289
omega_neutrino = 0
hubble = 67.77
#Transfer function settings, transfer_kmax=0.5 is enough for sigma_8
#transfer_k_per_logint=5 samples fixed spacing in logk
transfer_high_precision = T
transfer_kmax = 100
transfer_k_per_logint = 5
transfer_num_redshifts = 1
transfer_redshift(1) = 05.00
#which variable to use for defining the matter power spectrum and sigma8
#main choices are 2: CDM, 7: CDM+baryon+neutrino, 8: CDM+baryon, 9: CDM+baryon+neutrino+de perts
transfer_power_var = 8
#Increase accuracy_boost to decrease time steps, use more k values, etc.
#Decrease to speed up at cost of worse accuracy. Suggest 0.8 to 3.
accuracy_boost = 2

Thanks! 

Back to top 


Antony Lewis
Joined: 23 Sep 2004 Posts: 1183 Affiliation: University of Sussex


Back to top 


Giacomo Baso
Joined: 26 May 2016 Posts: 3 Affiliation: University of Padua  Department of Physics and Astronomy

Posted: October 12 2016 


Thanks for the reference, I'm not an expert on cosmological gauges but I see that's a physical effect that I should include (although it's probably outside the scale range we are interested with).
I'm still a little worried by the underflow exception, however. What's the cause of it? 

Back to top 


Antony Lewis
Joined: 23 Sep 2004 Posts: 1183 Affiliation: University of Sussex

Posted: October 12 2016 


That's harmless and fairly normal (if something is exponentially small, it's fine to underflow it to zero). 

Back to top 


Giacomo Baso
Joined: 26 May 2016 Posts: 3 Affiliation: University of Padua  Department of Physics and Astronomy

Posted: October 12 2016 


Got it, thank you. 

Back to top 


