likelihoods from camb and cosmomc

Use of Cobaya. camb, CLASS, cosmomc, compilers, etc.
Post Reply
Deepak Vaid
Posts: 7
Joined: May 24 2007
Affiliation: Penn State

likelihoods from camb and cosmomc

Post by Deepak Vaid » July 20 2007

Hello,

I have asked my share of stupid questions on this forum but don't let that discourage you from tackling the latest one.

Now cosmomc outputs -2 Ln(Like). For eg for lcdm I get 1769.36 (using only the wmap data, not CBI or ACBAR). But this number doesn't correspond to anything in the wmap3 paper. For eg. If I take the parameters I get from cosmomc for lcdm, generate the spectrum with camb and then use the wmap likelihood code directly using the 'test' program that's provided with the code I get the following:
------------------------------------------------------------------
Breakdown of -2ln(L)
------------------------------------------------------------------
MASTER TTTT = 1037.663329 for 970 ls

Beam/ptsrc TT correction = 1.888136

low l TTTT chi2 = 907.978652 for 957 pixels

low l TTTT det = -14.151686

MASTER TETE chi2 = 417.430196 for 427 ls

MASTER TETE det = 1.694096 TT/TE/EE/BB

lowl chi2 = 1183.418457 for 1172 pixels

TT/TE/EE/BB lowl det = 2.851461
------------------------------------------------------------------
TOTAL -2ln(L) = 3538.772642
------------------------------------------------------------------
Expected -2ln(L) = 3540.792945
Difference = -2.020303
The numbers here are close to the wmap3 paper, but clearly nothing like 1769 ! How does cosmomc compute this number? How many degrees of freedom does it use? Can someone please clarify this for me. For reference below are the lcdm parameters that gave the above likelihood.
Omega_b h^2 0.0223456
Omega_{DM} h^2 0.101281
Omega_{Lambda} 0.778772
z_{re} 12.0669
n_s 0.962249
10^{10} A_s 20.7481
H_0 74.7543
Age/GYr 13.6973
Omega_m 0.221228
tau 0.100327

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

likelihoods from camb and cosmomc

Post by Pascal Vaudrevange » July 23 2007

Hi,

only marginally related: as far as I understand, starting with v2.0 of the wmap3 likelihood code, there is some constant number being subtracted from the computed likelihood.

more on topic: which combination of TT, EE, TE, BB do you use in cosmomc? Is this the same combination that's being used when you run the wmap3 code standalone? Does the fake spectrum from Camb contain tensors?

Anze Slosar
Posts: 183
Joined: September 24 2004
Affiliation: Brookhaven National Laboratory
Contact:

likelihoods from camb and cosmomc

Post by Anze Slosar » July 23 2007

1768*2 = 3536 which is about right.
cosmomc ouputs -log L which is chi^2/2

a

Post Reply