Removing the fat from your posterior samples with margarine

  title={Removing the fat from your posterior samples with margarine},
  author={H. T. J. Bevins and Will Handley and Pablo Lemos and Peter H Sims and Eloy de Lera Acedo and Anastasia Fialkov and Justin Alsing},
Bayesian workflows often require the introduction of nuisance parameters, yet for core science modelling one needs access to a marginal posterior density. In this work we use masked autoregressive flows and kernel density estimators to encapsulate the marginal posterior, allowing us to compute marginal Kullback– Leibler divergences and marginal Bayesian model dimensionalities in addition to generating samples and computing marginal log probabilities. We demonstrate this in application to topical… 

Figures from this paper

Marginal Bayesian Statistics Using Masked Autoregressive Flows and Kernel Density Estimators with Examples in Cosmology
This work presents a means by which to combine constraints from different data sets in a computationally efficient manner by generating rapid, reusable and reliable marginal probability density estimators, giving access to nuisance-free likelihoods.


A general Bayesian framework for foreground modelling and chromaticity correction for global 21 cm experiments
A new physics-motivated method of modelling the foregrounds of 21 cm experiments in order to fit the chromatic distortions as part of the foregrounding and it is demonstrated that fitting this model for varying N using a Bayesian nested sampling algorithm allows the 21 cm signal to be reliably detected in data of a relatively smooth conical log spiral antenna.
Anesthetic: Nested Sampling Visualisation
This work unifies many existing tools and techniques in an extensible framework that is intuitive for users familiar with the standard Python packages, namely NumPy, SciPy, Matplotlib and pandas.
Full-Shape Galaxy Power Spectra and the Curvature Tension
With recent evidence for a possible “curvature tension" among early and late universe cosmological probes, Effective Field Theories of Large Scale Structure (EFTofLSS) have emerged as a promising new
Bayesian evidence-driven likelihood selection for sky-averaged 21-cm signal extraction
The generalised normal likelihood function is proposed as a good approximation of the true likelihood function of a dataset if the noise structure is a priori unknown.
Bayesian evidence-driven diagnosis of instrumental systematics for sky-averaged 21-cm cosmology experiments
A Bayesian evidence-based model comparison is capable of determining whether or not such a systematic model is needed as the true underlying generative model of an experimental dataset is in principle unknown.
Radio antenna design for sky-averaged 21 cm cosmology experiments: the REACH case
J. Cumner1†, E. De Lera Acedo, D.I.L. de Villiers, D. Anstey, C. I. Kolitsidas, B. Gurdon, N. Fagnoni, P. Alexander, G. Bernardi, H.T.J. Bevins, S. Carey, J. Cavillot, R. Chiello, C. Craeye, W.
HERA Phase I Limits on the Cosmic 21 cm Signal: Constraints on Astrophysics and Cosmology during the Epoch of Reionization
Recently, the Hydrogen Epoch of Reionization Array (HERA) has produced the experiment’s first upper limits on the power spectrum of 21 cm fluctuations at z ∼ 8 and 10. Here, we use several
Informing antenna design for sky-averaged 21-cm experiments using a simulated Bayesian data analysis pipeline
A method for guiding the antenna design of a global experiment through data analysis simulations is demonstrated, finding that the log spiral performs significantly better than the other antennae tested, able to correctly and confidently identify every inserted 21-cm signal.
Constraining the state of the intergalactic medium during the Epoch of Reionization using MWA 21-cm signal observations
The Murchison Widefield Array (MWA) team has derived new upper limits on the spherically averaged power spectrum of the 21-cm signal at six redshifts in the range z ≈ 6.5−8.7. We use these upper
Nested sampling with any prior you like
Nested sampling is an important tool for conducting Bayesian analysis in Astronomy and other fields, both for sampling complicated posterior distributions for parameter inference, and for computing