Constraining the polarisation flux density and angle of point sources by training a convolutional neural network

Abstract

Constraining the polarisation properties of extragalactic point sources is a relevant task not only because they are one of the main contaminants for primordial cosmic microwave background B-mode detection if the tensor-to-scalar ratio is lower than r = 0.001, but also for a better understanding of the properties of radio-loud active galactic nuclei. We develop and train a machine learning model based on a convolutional neural network to learn how to estimate the polarisation flux density and angle of point sources embedded in cosmic microwave background images knowing only their positions. To train the neural network, we use realistic simulations of patches of area 32x32 pixels at the 217 GHz Planck channel with injected point sources at their centres. The patches also contain a realistic background composed by dust, the CMB and instrumental noise. Firstly, we study the comparison between true and estimated polarisation flux densities for P, Q and U. Secondly, we analyse the comparison between true and estimated polarisation angles. Finally, we study the performance of our model with real data and we compare our results against the PCCS2. We obtain that our model is reliable to constrain the polarisation flux above 80 mJy. For this limit, we obtain errors lower than 30%. Training the same network with Q and U, the reliability limit is above +-250 mJy for determining the polarisation angle of both Q and U sources with a 1sigma uncertainty of +-29deg and +-32deg for Q and U sources respectively. We obtain similar results to the PCCS2 for some sources, although we also find discrepancies in the 300-400 mJy flux density range with respect to the Planck catalogue. Based on these results, our model seems to be a promising tool to give estimations of the polarisation flux densities and angles of point sources above 80 mJy in any catalogue with practically null computational time.

arXiv comment9 pages, 9 Figures. Forthcoming article Astronomy & Astrophysics journal

References in corpus (28)

• HEALPix -- a Framework for High Resolution Discretization, and Fast Analysis of Data Distributed on the Sphere
• Planck 2018 results. I. Overview and the cosmological legacy of Planck
• Radio and Millimeter Continuum Surveys and their Astrophysical Implications
• The pre-launch Planck Sky Model: a model of sky emission at submillimetre to centimetre wavelengths
• Planck 2013 results. XXVIII. The Planck Catalogue of Compact Sources
• High--frequency predictions for number counts and spectral properties of extragalactic radio sources. New evidences of a break at mm wavelengths in spectra of bright blazar sources
• The Mexican Hat Wavelet Family. Application to point source detection in CMB maps
• Predictions on the angular power spectrum of clustered extragalactic point sources at CMB frequencies from flat and all--sky 2D-simulations
• Exploring Cosmic Origins with CORE: B-mode Component Separation
• CMB power spectrum estimation and map reconstruction with the Expectation - Maximization algorithm
• The impact of polarized extragalactic radio sources on the detection of CMB anisotropies in polarization
• The Atacama Cosmology Telescope: Two-season ACTPol Extragalactic Point Sources and their Polarization properties
• Convolutional Neural Networks on the HEALPix sphere: a pixel-based algorithm and its application to CMB data analysis
• Polarization of the WMAP Point Sources
• Forecasting Polarized Radio Sources for CMB observations
• Fractional Polarisation of Extragalactic Sources in the 500-square-degree SPTpol Survey
• ForSE: a GAN based algorithm for extending CMB foreground models to sub-degree angular scales
• A novel multi-frequency technique for the detection of point sources in Cosmic Microwave Background maps
• Statistics of the fractional polarisation of compact radio sources in Planck maps
• Inpainting Galactic Foreground Intensity and Polarization maps using Convolutional Neural Network
• Detection/estimation of the modulus of a vector. Application to point source detection in polarization data
• Statistics of the fractional polarisation of extragalactic dusty sources in Planck HFI maps
• Average fractional polarization of extragalactic sources at Planck frequencies
• CENN: A fully convolutional neural network for CMB recovery in realistic microwave sky simulations
• Characterization of polarimetric and total intensity behaviour of a complete sample of PACO radio sources in the radio bands
• Compact source detection in multi-channel microwave surveys: from SZ clusters to polarized sources
• Characterization of Extragalactic Point-Sources on E- and B-mode Maps of the CMB Polarization
• A Bayesian method for point source polarization estimation