Add nested samples to multinest result
This MR adds the nested samples to the result of a pymultinest
analysis. This is pulled from the files written by multinest
.
I basically figured out which data to use by trial and error, but the columns in the ev.dat
and phys_live.points
files are SEARCH_PARAMETERS ln_liklelihood ln_prior_volume mode
where
-
ln_prior_volume
is the prior volume associated with the current sample, i.e.,ln(w_i)
in (8) of Skilling. -
mode
is the number of the ellipse that point was drawn from.
The prior volume associated with the current live points just takes the naive remaining prior volume / N described in Skilling. I don't know if this will exactly match the method used internally in multinest, but it gives results that look good to me.