Skip to main content

Come analyze HEASARC, IRSA, and MAST data in the cloud! The Fornax Initiative is now welcoming all interested beta users.

Xspec Home Page

next up previous contents
Next: Comparing the results Up: Running each sampler Previous: Hamiltonian Monte Carlo.   Contents

Nested sampling.

This is the only sampler that does not warm-start from the MLE; it samples from the prior and uses the likelihood to compress towards the posterior. Two of our parameters (nH and the apec norm) have hard-coded upper bounds that span far more than four decades, so we narrow their search ranges and switch to a Jeffreys (log-uniform) prior before invoking nest. 

newpar 1 0.05 0.001 1e-6 1e-6 100000 1e6
newpar 5 0.04 0.01  1e-30 1e-30 1e20 1e24
bayes on
bayes 1 JEFFREYS
bayes 5 JEFFREYS

nest live 400
parallel nest 4
nest run nest.fits

Observed: 155 s elapsed time, evidence $\log Z = -236.1 \pm 0.2$.