Monte Carlo Markov Chains for mock Planck data (with and without lensing extraction)

We provide here the chains from the paper "Probing cosmological parameters with the CMB: Forecasts from full Monte Carlo simulations" by Laurence Perotto, Julien Lesgourgues, Steen Hannestad, Huitzu Tu, Yvonne Y.Y. Wong, JCAP 0610 (2006) 013 [astro-ph/0606227]. These chains could be useful for people willing to estimate future errors for a combination of Planck data with another mock data set: it is then possible to perform some importance sampling analysis starting from our chains.

These chains cover four situations: eight or eleven free cosmological parameters, each with or without lensing extraction.

The model with eight free parameters is the minimal model to consider in a Planck analysis: its parameters are the six parameters of the "vanilla model" considered by e.g. WMAP, plus the primordial helium fraction (should not be kept fixed at the level of precision of Planck) and the fraction of dark matter in the form of massive neutrinos (cannot be neglected for Planck, given lower bounds on the neutrino mass from neutrino oscillation experiments). The model with eleven parameters is an extended model with extra parameters alpha (running of the scalar tilt), r (tensor-to-scalar ratio), and w (Dark Energy pressure over density).

In the chains where lensing extraction is included, it was assumed that the deflection spectrum dd and the cross-spectrum Td are being measured using the quadratic estimator technique of Hu and Okamoto. These spectra are then fitted together with the TT/TE/EE spectra, including an estimate of the noise for the dd spectrum which is derived from Hu and Okamoto's analytic formulas.

Moreover, in each case, the parameter reconstruction can be performed by fitting directly the theoretical spectra instead of some random realization. In that case, the best-fit parameter values coincide with the fiducial values and the minimum log likelihood is zero; despite these unrealistic features, this approach is sufficient for estimating the error bar on each parameter. So, in each case, we provide chains fitted directly to the fiducial model spectrum. In the case of eight parameters / no lensing extraction, we also provide chains fitting five different random realizations of the fiducial model.

Name of the chain roots:

The file chains.tar contains all these chains (30 chains in each case). They are raw chains, the burn-in phase still needs to be remove.