We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

astro-ph.CO

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Astrophysics > Cosmology and Nongalactic Astrophysics

Title: Accelerating Approximate Bayesian Computation with Quantile Regression: Application to Cosmological Redshift Distributions

Abstract: Approximate Bayesian Computation (ABC) is a method to obtain a posterior distribution without a likelihood function, using simulations and a set of distance metrics. For that reason, it has recently been gaining popularity as an analysis tool in cosmology and astrophysics. Its drawback, however, is a slow convergence rate. We propose a novel method, which we call qABC, to accelerate ABC with Quantile Regression. In this method, we create a model of quantiles of distance measure as a function of input parameters. This model is trained on a small number of simulations and estimates which regions of the prior space are likely to be accepted into the posterior. Other regions are then immediately rejected. This procedure is then repeated as more simulations are available. We apply it to the practical problem of estimation of redshift distribution of cosmological samples, using forward modelling developed in previous work. The qABC method converges to nearly same posterior as the basic ABC. It uses, however, only 20\% of the number of simulations compared to basic ABC, achieving a fivefold gain in execution time for our problem. For other problems the acceleration rate may vary; it depends on how close the prior is to the final posterior. We discuss possible improvements and extensions to this method.
Comments: 10 pages, 5 figures, prepared for submission to JCAP
Subjects: Cosmology and Nongalactic Astrophysics (astro-ph.CO); Machine Learning (stat.ML)
DOI: 10.1088/1475-7516/2018/02/042
Cite as: arXiv:1707.07498 [astro-ph.CO]
  (or arXiv:1707.07498v2 [astro-ph.CO] for this version)

Submission history

From: Tomasz Kacprzak [view email]
[v1] Mon, 24 Jul 2017 11:47:56 GMT (1714kb,D)
[v2] Tue, 25 Jul 2017 16:55:23 GMT (1714kb,D)

Link back to: arXiv, form interface, contact.