Astrophysics Source Code Library

Making codes discoverable since 1999

ASCL Code Record

[ascl:2401.009] Harmonic: Learnt harmonic mean estimator

harmonic learns an approximate harmonic mean estimator (referred to as a "learnt harmonic mean estimator") from posterior distribution samples to compute the marginal likelihood required for Bayesian model selection. Using a large number of independent Markov chain Monte Carlo (MCMC) chains from another package such as emcee (ascl:1303.002), harmonic uses importance sampling to learn a new target distribution in order to optimize an approximate harmonic estimator while minimizing its variance.

Code site:
Used in:
Described in:
Preferred citation method: ; please see additional citation information here:

Views: 134

Add this shield to your page
Copy the above HTML to add this shield to your code's website.