CosmoMC: action = 2 vs action = 4

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Erick Almaraz
Posts: 25
Joined: June 10 2012
Affiliation: University of Mexico

CosmoMC: action = 2 vs action = 4

Post by Erick Almaraz » October 28 2016

Hi everybody. I have run CosmoMC setting action = 2 to find the best-fit point. The minimum output file lists, besides the best fit values of all the parameters, the information about the total chi-sq and the chi-sq for every data set I used:

Code: Select all

 -log(Like)   chi-sq    data
      7.527    15.053   CMB: lowl = commander_rc2_v1.1_l2_29_B
    380.793    761.585   CMB: plik = plik_dx11dr2_HM_v18_TT
    347.808    695.616   SN: JLA December_2013
      0.008      0.016   BAO: 6DF
      0.704      1.407   BAO: MGS
      1.102      2.204   BAO: DR11CMASS
      0.155      0.310   BAO: DR11LOWZ
Then I fixed the central values of the parameters in the ini file to the best fit values quoted in the minimum file and now use action = 4 to calculate the likelihood for this particular choice. I suppose the results of the chi-sq should be very similar. However, I get:

Code: Select all

    loglike     chi-sq
      7.550     15.100   CMB: lowl = commander_rc2_v1.1_l2_29_B
    451.099    902.198   CMB: plik = plik_dx11dr2_HM_v18_TT
    347.818    695.636   SN: JLA December_2013
      0.008      0.016   BAO: 6DF
      0.704      1.407   BAO: MGS
      1.102      2.204   BAO: DR11CMASS
      0.155      0.310   BAO: DR11LOWZ
As can be seen, the results for CMB-lowl, SNe and BAO are very similar, but the results for CMB-highl are very different. Does anybody knows what's going on?

Regards

Antony Lewis
Posts: 1363
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: CosmoMC: action = 2 vs action = 4

Post by Antony Lewis » October 28 2016

Look like a small shift in amplitude (or possibly theta) - check your output parameters are the same to enough decimal places?

Erick Almaraz
Posts: 25
Joined: June 10 2012
Affiliation: University of Mexico

CosmoMC: action = 2 vs action = 4

Post by Erick Almaraz » October 28 2016

Dear Antony. Thanks for the quick reply. I noticed that I only fixed the usual cosmological parameters (eg, omegabh2, omegach2, etc), but I omitted to fix also the Planck's hard parameters (eg, calPlanck, kgal100, etc). After doing that, the results from action = 2 and action = 4 are the same!

Regards.

Post Reply