Imperial College London
Alternate

Professor Andrew H Jaffe

Faculty of Natural Sciences, Department of Physics

Professor of Astrophysics and Cosmology
 
 
 
//

Contact

 

+44 (0)20 7594 7526a.jaffe Website

 
 
//

Assistant

 

Miss Louise Hayward +44 (0)20 7594 7679

 
//

Location

 

1018BBlackett LaboratorySouth Kensington Campus

//

Summary

 

Publications

Citation

BibTex format

@article{Heavens:2020:mnras/staa2589,
author = {Heavens, A and Sellentin, E and Jaffe, A},
doi = {mnras/staa2589},
journal = {Monthly Notices of the Royal Astronomical Society},
pages = {3440--3451},
title = {Extreme data compression while searching for new physics},
url = {http://dx.doi.org/10.1093/mnras/staa2589},
volume = {498},
year = {2020}
}
Download

RIS format (EndNote, RefMan)

TY  - JOUR
AB  - Bringing a high-dimensional dataset into science-ready shape is a formidablechallenge that often necessitates data compression. Compression has accordinglybecome a key consideration for contemporary cosmology, affecting public datareleases, and reanalyses searching for new physics. However, data compressionoptimized for a particular model can suppress signs of new physics, or evenremove them altogether. We therefore provide a solution for exploring newphysics \emph{during} data compression. In particular, we store additionalagnostic compressed data points, selected to enable precise constraints ofnon-standard physics at a later date. Our procedure is based on the maximalcompression of the MOPED algorithm, which optimally filters the data withrespect to a baseline model. We select additional filters, based on ageneralised principal component analysis, which are carefully constructed toscout for new physics at high precision and speed. We refer to the augmentedset of filters as MOPED-PC. They enable an analytic computation of Bayesianevidences that may indicate the presence of new physics, and fast analyticestimates of best-fitting parameters when adopting a specific non-standardtheory, without further expensive MCMC analysis. As there may be large numbersof non-standard theories, the speed of the method becomes essential. Should nonew physics be found, then our approach preserves the precision of the standardparameters. As a result, we achieve very rapid and maximally preciseconstraints of standard and non-standard physics, with a technique that scaleswell to large dimensional datasets.
AU  - Heavens,A
AU  - Sellentin,E
AU  - Jaffe,A
DO  - mnras/staa2589
EP  - 3451
PY  - 2020///
SN  - 0035-8711
SP  - 3440
TI  - Extreme data compression while searching for new physics
T2  - Monthly Notices of the Royal Astronomical Society
UR  - http://dx.doi.org/10.1093/mnras/staa2589
UR  - http://arxiv.org/abs/2006.06706v1
UR  - https://academic.oup.com/mnras/advance-article/doi/10.1093/mnras/staa2589/5897375
UR  - http://hdl.handle.net/10044/1/81975
VL  - 498
ER  -
Download