Custom Covariance Matrix Question

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Lawrence Kazantzidis
Posts: 1
Joined: August 28 2019
Affiliation: University of Ioannina

Custom Covariance Matrix Question

Post by Lawrence Kazantzidis » August 29 2019

Hi everyone,

I am currently working on a project using cosmomc and I want to separate the data in two different categories. The first one consists of BAORSD+DES+lowl data and the other one of highl+BAO+Pantheon+HST data. I have successfully obtained the following contours for the first set of data but as you can see clearly from the following picture they are not at all smooth.

https://imgur.com/a/zeoXTd9

where [math] is an extra parameter of my model. In my analysis in the test.ini file I have set the propose matrix to be

Code: Select all

propose_matrix= planck_covmats/base_BAO_HST_JLA_TT_lensing_lowTEB_plik.covmat
Could it be because the covariance matrix does not correspond to the specific combination of data? I have searched the folder planck_covmats but I can not find the specific combination that I want. I am not sure what changes should I make in order to obtain an accurate covariance matrix corresponding to the exact combination of data. Is there an easy way to do that?


Thank you in advance,
Lawrence.

Antony Lewis
Posts: 1501
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: Custom Covariance Matrix Question

Post by Antony Lewis » September 01 2019

Cosmomc will learn the covariance, so it's not critical as long as it's not much too big. The contours don't look too bad (DES, esp. lensing, can give two regions of parameters space leaving odd looking marginalized contours, e.g. see the Planck 2018 lensing paper).

Post Reply