期刊
MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY
卷 481, 期 1, 页码 494-508出版社
OXFORD UNIV PRESS
DOI: 10.1093/mnras/sty2118
关键词
methods: statistical; dust, extinction
资金
- United Kingdom Science Technology and Facilities Council (STFC) [ST/K00106X/1, ST/M00127X/1]
- European Research Council under the European Union's Seventh Framework Programme (FP7/2007-2013)/ERC grant [321067]
- STFC [ST/N000919/1, ST/K00106X/1, ST/M00127X/1, ST/I001077/1] Funding Source: UKRI
Gaussian processes (GPs) are the ideal tool for modelling the Galactic interstellar medium, combining statistical flexibility with a good match to the underlying physics. In an earlier paper, we outlined how they can be employed to construct three-dimensional maps of dust extinction from stellar surveys. GPs scale poorly to large data sets though, which put the analysis of realistic catalogues out of reach. Here, we show how a novel combination of the expectation propagation method and certain sparse matrix approximations can be used to accelerate the dust mapping problem. We demonstrate, using simulated Gaia data, that the resultant algorithm is fast, accurate, and precise. Critically, it can be scaled up to map the Gaia catalogue.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据