problem about getdist margestats

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Duke Deng
Posts: 1
Joined: January 27 2019
Affiliation: Beijing Institute of Technology

problem about getdist margestats

Post by Duke Deng » January 28 2019

Hi,
I have a problem when using getdist to get .margestats file of my chain. The samples of one my parameter gather at one side in the 1D marginalized density plot. However, when I get the .margestats file, the mean(4.09) of this parameter is larger than its upper1(3.97). I wonder if it is normal in getdist? Under what circumstances will this happen? Thanks a lot !
Attachments
9f5cb1c60e757f0396267371227bf78.png
9f5cb1c60e757f0396267371227bf78.png (6.57 KiB) Viewed 120 times

Antony Lewis
Posts: 1456
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: problem about getdist margestats

Post by Antony Lewis » January 28 2019

Anything sufficiently skewed can have a mean larger than a 1 sigma upper confidence limit. Even a half Gaussian has mean at the 60% limit.

Post Reply