Journal Article

Fast optimal CMB power spectrum estimation with Hamiltonian sampling

J. F. Taylor, M. A. J. Ashdown and M. P. Hobson

in Monthly Notices of the Royal Astronomical Society

Published on behalf of The Royal Astronomical Society

Volume 389, issue 3, pages 1284-1292
Published in print September 2008 | ISSN: 0035-8711
Published online September 2008 | e-ISSN: 1365-2966 | DOI: http://dx.doi.org/10.1111/j.1365-2966.2008.13630.x
Fast optimal CMB power spectrum estimation with Hamiltonian sampling

More Like This

Show all results sharing this subject:

  • Astronomy and Astrophysics

GO

Show Summary Details

Preview

We present a method for fast optimal estimation of the temperature angular power spectrum from observations of the cosmic microwave background. We employ a Hamiltonian Monte Carlo (HMC) sampler to obtain samples from the posterior probability distribution of all the power spectrum coefficients given a set of observations. We compare the properties of the HMC and the related Gibbs sampling approach on low-resolution simulations. The correlation lengths of the samples produced by Gibbs and HMC are generally comparable. At very high signal-to-noise (≳10) the Gibbs outperforms the HMC while for intermediate signal-to-noise (∼0.1–10) the HMC results in shorter correlation lengths than the Gibbs. At low signal-to-noise (≲0.1) both samplers appear to perform similarly. We also demonstrate the method on high-resolution data by applying it to simulated Wilkinson Microwave Anisotropy Probe (WMAP) data. Analysis of a single map from a WMAP-resolution data set is possible in around 80 h on a high-end desktop computer. HMC uses around a factor of 3–4 fewer spherical harmonic transforms than is reported for a Gibbs sampler (with pre-conditioning) and therefore produces noticeable performance gains. HMC imposes few conditions on the distribution to be sampled and obviates the requirement of having to be able to sample efficiently from the conditional densities, thus providing us with an extremely flexible approach upon which to build.

Keywords: methods: data analysis; methods: statistical; cosmic microwave background

Journal Article.  5682 words.  Illustrated.

Subjects: Astronomy and Astrophysics

Full text: subscription required

How to subscribe Recommend to my Librarian

Users without a subscription are not able to see the full content. Please, subscribe or login to access all content.