## CosmoMC June 2016 release: Issue regarding converged statistics

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Luke Hart
Posts: 39
Joined: July 13 2015
Affiliation: University of Manchester

### CosmoMC June 2016 release: Issue regarding converged statist

All,

I am currently undergoing an analysis of new parameters which have been added to CosmoRec, I have configured these parameters to successfully work in CosmoRec's batch mode and also communicate successfully to CAMB. From CAMB, these parameters are then communicated to CosmoMC and everything seemed to be working.

However, I have been trying to recreate the values in a published paper on the arxiv. The results of all the parameters are consistent however my new parameter $x$ has an error which is almost exactly 10 times smaller than the literature. This is the only discrepancy from the results and I've checked that the likelihoods are similar. I could understand slight changes in parameter values due to nuisance parameters or even the updates in CosmoRec since then, however as I have mentioned,

$\sigma_x \sim 0.1\sigma_x^{\rm paper}$

I have tried recreating PLANCK 2015's results for the two photon decay rate $A_{2s1s}$ as well and unfortunately that is very consistent as well; yet the error is ten times lower again. In this analysis, I basically replaced $x$ by $A_{2s1s}$. The communication through CosmoRec->CAMB->CosmoMC is identical.

Any ideas?

Update: This issue has been solved but has sparked a new question which I want to discuss in http://cosmocoffee.info/viewtopic.php?t=2820