CMB S4 likelihood for LCDM+massive neutrinos

[ptaule]

Hello,

I'm trying to use the CMB-S4 mock likelihood in combination with fake planck likelihoods as in 1808.05955, for LCDM with massive neutrinos, but I repeatedly obtain chains with very low acceptance rate of 0.1 - 1%, even after tens of thousands of steps. I have tried with/without --superupdate 20 and also turning on/off the ncdm fluid approximation in CLASS, with no luck. Do you have an idea what could be the issue?

This is the input file I use, from which also the fiducial files are created. I give also the input covmat -c covmat/base2018TTTEEE.covmat when running the chains.

data.experiments=['fake_planck_realistic_lowl', 'fake_planck_realistic_highl_fsky0p17', 'cmb_s4_highl']

data.parameters['omega_b']              = [ 2.2387   , None  , None , 0.015   , 0.01 , 'cosmo']
data.parameters['omega_cdm']            = [ 0.1198   , None  , None , 0.0013  , 1    , 'cosmo']
data.parameters['100*theta_s']          = [ 1.042    , None  , None , 0.00030 , 1    , 'cosmo']
data.parameters['ln10^{10}A_s']         = [ 3.052    , None  , None , 0.015   , 1    , 'cosmo']
data.parameters['n_s']                  = [ 0.9659   , None  , None , 0.0042  , 1    , 'cosmo']
data.parameters['tau_reio']             = [ 0.05736  , 0.004 , None , 0.008   , 1    , 'cosmo']

data.parameters['m_ncdm']               = [ 0.02        , 0     , 1    , 0.1     , 1    , 'cosmo']

data.parameters['z_reio']       = [1, None, None, 0,     1,   'derived']
data.parameters['Omega_Lambda'] = [1, None, None, 0,     1,   'derived']
data.parameters['YHe']          = [1, None, None, 0,     1,   'derived']
data.parameters['H0']           = [0, None, None, 0,     1,   'derived']
data.parameters['A_s']          = [0, None, None, 0,  1e-9,   'derived']

# Other cosmo parameters (fixed parameters, precision parameters, etc.)

data.cosmo_arguments['sBBN file'] = data.path['cosmo']+'/bbn/sBBN.dat'
data.cosmo_arguments['k_pivot'] = 0.05

data.cosmo_arguments['N_ur'] = 0.00641
data.cosmo_arguments['N_ncdm'] = 1
data.cosmo_arguments['deg_ncdm'] = 3

data.N=10
data.write_step=5

Best, Petter Taule

[brinckmann]

Hi Petter,

Most likely the Planck 2018 covmat isn't sensitive enough, so you're exploring too large a portion of parameter space leading to a poor acceptance rate. Your best bet would be to take the 1-sigma values from 1808.05955 for Planck+S4 and add those as the starting guess sigmas in your param file (fourth entry in the list) without passing an input covmat (which would override those values).

Best, Thejs

[ptaule]

Hi Thejs,

thanks for the reply! It still did not work with the 1-sigma from 1808.05955, but I got it working using the Fisher method + superupdate. Closing