[CosmoMC] Posterior distribution with weird secondary peak contours

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Luke Hart
Posts: 57
Joined: July 13 2015
Affiliation: University of Manchester

[CosmoMC] Posterior distribution with weird secondary peak contours

Post by Luke Hart » September 24 2018

Dear all,

I have been attempting to recreate some results from a previous paper with the Planck data (TTTEEE+lowTEB) however I have just ran the CosmoMC code with a small perturbation to the Cls added in and then sampled over.

This has been done before and the contours have come out very smoothly. I have redone the perturbations and the differences are very small (.1% at l=2000 and smaller at lower l's) however; the contours are vastly different, actually giving some bi-modal almost behaviour in the posterior distribution. Rather than two solitary peaks, it looks more like there is a second peak superposed close to the 68% limit of the first peak and this is very irregular compared to the previous plots. It's even weirder given that the bimodality is clearly coming from the amplitude of this Cl perturbation (we shall call A_1) being heavily degenerate with theta, leading to bimodality in theta as well.

If anyone could help me shed some light on this that would be awesome. Thank you

Luke

Post Reply