A statistical test for Nested Sampling algorithms
read the original abstract
Nested sampling is an iterative integration procedure that shrinks the prior volume towards higher likelihoods by removing a "live" point at a time. A replacement point is drawn uniformly from the prior above an ever-increasing likelihood threshold. Thus, the problem of drawing from a space above a certain likelihood value arises naturally in nested sampling, making algorithms that solve this problem a key ingredient to the nested sampling framework. If the drawn points are distributed uniformly, the removal of a point shrinks the volume in a well-understood way, and the integration of nested sampling is unbiased. In this work, I develop a statistical test to check whether this is the case. This "Shrinkage Test" is useful to verify nested sampling algorithms in a controlled environment. I apply the shrinkage test to a test-problem, and show that some existing algorithms fail to pass it due to over-optimisation. I then demonstrate that a simple algorithm can be constructed which is robust against this type of problem. This RADFRIENDS algorithm is, however, inefficient in comparison to MULTINEST.
This paper has not been read by Pith yet.
Forward citations
Cited by 4 Pith papers
-
First analytical coverage bounds of a fully specified nested sampling algorithm
Derives heuristic coverage bounds for MLFriends nested sampling under a Binomial point process model, claiming the bias is negligible compared to statistical variance.
-
X-ray luminosity function of Compton-thick AGN in the early Universe (z > 3). Robustness and biases of the CTK population
After applying IR-informed priors to X-ray NH posteriors in a sample of 811 high-z AGN, CTK sources comprise 17% of the population at 3<z<6 with no significant evolution.
-
Counting voids and filaments: Betti Curves as a Powerful Probe for Cosmology
Betti curves from persistent homology of large-scale structure provide complementary cosmological constraints on ns, sigma8, and Om, with tighter bounds when analyzed jointly with the power spectrum.
-
The ESO SupJup Survey XI. Atmospheric properties of six isolated M- and L-type dwarfs with CRIRES+
Retrievals on six isolated brown dwarfs yield near-solar C/O (0.51-0.63), metallicities, and 12C/13C ratios (91-155) supporting molecular cloud fragmentation origin.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.