I'm trying to use CosmoMC-Nov2016 with only simulated Planck data generated using the makePerfectForecastDataset.py script but the parameters recovered through MCMC are very different from the ones used to generate the data. I use CAMB (the version included into CosmoMC-Nov2016) to produce the scalar Cls, modifying only the following parameters in the params.ini file (all the other parameters are the default ones with which CosmoMC is dowloaded)
Code: Select all
l_max_scalar = 2000
do_lensing = F
use_physical = T
ombh2 = 0.022
omch2 = 0.11
omnuh2 = 0.0
omk = 0
hubble = 70
re_optical_depth = 0.09
scalar_spectral_index(1) = 0.96
scalar_amp(1) = 2.4e-9
l_sample_boost = 50
Code: Select all
dataCols = 'TT EE TE'
fwhm_arcmin = 7.0
NoiseVar = 3e-4
Code: Select all
cmb_dataset[MyForecast]=data/MyForecast/simulatedCls_exactsim.dataset
Code: Select all
DEFAULT(batch2/MyForecast.ini)
DEFAULT(batch2/common.ini)
MPI_Max_R_ProposeUpdate = 30
propose_matrix= planck_covmats/base_TT_lowTEB_plik.covmat
action = 0
start_at_bestfit =F
feedback=1
use_fast_slow = T
checkpoint = T
sampling_method = 7
dragging_steps = 3
propose_scale = 2
indep_sample=10
Code: Select all
use_nonlinear_lensing = F
block_semi_fast = T
CMB_lensing = F
stop_on_error= F
Code: Select all
param[omegabh2] = 0.0221 0.005 0.1 0.0001 0.0001
param[omegach2] = 0.12 0.01 0.2 0.001 0.0005
param[theta] = 1.03 0.5 1.2 0.0004 0.0002
param[tau] = 0.089 0.01 0.3 0.01 0.005
neutrino_hierarchy = degenerate
num_massive_neutrinos=1
param[mnu] = 0.0
param[meffsterile] = 0
param[omegak] = 0
param[w] = -1
param[nrun] = 0
param[nrunrun] = 0
param[r] = 0
param[wa] = 0
param[nnu] = 0.0
param[yhe] = 0.24
param[alpha1] = 0
param[deltazrei] = 0.5
param[Alens] = 1
param[Alensf]=-1
param[fdm] = 0
param[ns] = 0.96 0.8 1.2 0.004 0.002
param[logA] = 3.1 2 4 0.001 0.001
param[Aphiphi] = 1
omegabh2 ~ 0.019 insted of 0.022
omegach2 ~ 0.069 instead of 0.11
theta ~ 1.04
tau~0.0757 instead of 0.089
logA~2.94 instead of 3.13
ns~0.801 instead of 0.96
H0*~46.7 instead of 70
omegal*~0.592 instead of 0.73
GetDist gives
Code: Select all
Number of chains used = 4
var(mean)/mean(var), remaining chains, worst e-value: R-1 = 0.00241
RL: Thin for Markov: 27
RL: Thin for indep samples: 31
RL: Estimated burn in steps: 130 ( 52 rows)
using 55299 rows, 47 parameters; mean weight 2.52382502396, tot weight 139565.0
Approx indep samples (N/corr length): 4362.0
Equiv number of single samples (sum w)/max(w): 5169.0
Effective number of weighted samples (sum w)^2/sum(w^2): 32924
Best fit sample -log(Like) = 268.629400
Ln(mean 1/like) = 276.122745
mean(-Ln(like)) = 271.872252
-Ln(mean like) = 270.858158
I tried also increasing the temperature to 20 but I obtained the same wrong parameters. Should I add a prior on H0 maybe? Any help is greatly appreciated, many thanks for your time and patience.