arXiv subjects

J. D. Santos

Publications and source records attributed to J. D. Santos.

7 recordsLinked to original sources

arXiv2023

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

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.

astro-ph.CO↗

arXiv2022

CENN: A fully convolutional neural network for CMB recovery in realistic microwave sky simulations

Component separation is the process with which emission sources in astrophysical maps are generally extracted by taking multi-frequency information into account. It is crucial to develop more reliable methods for component separation for future CMB experiments. We aim to develop a new method based on fully convolutional neural networks called the Cosmic microwave background Extraction Neural Network (CENN) in order to extract the CMB signal in total intensity. The frequencies used are the Planck channels 143, 217 and 353 GHz. We validate the network at all sky, and at three latitude intervals: lat1=0^{\circ}<b<5^{\circ}, lat2=5^{\circ}<b<30^{\circ} and lat3=30^{\circ}<b<90^{\circ}, without using any Galactic or point source masks. For training, we make realistic simulations in the form of patches of area 256 pixels, which contain the CMB, Dust, CIB and PS emissions, Sunyaev-Zel'dovich effect and the instrumental noise. After validate the network, we compare the power spectrum from input and output maps. We analyse the power spectrum from the residuals at each latitude interval and at all sky and we study the performance of our model dealing with high contamination at small scales. We obtain a power spectrum with an error of 13{\pm}113 μK^2 for multipoles up to above 4000. For residuals, we obtain 700{\pm}60 μK^2 for lat1, 80{\pm}30 μK^2 for lat2 and 30{\pm}20 μK^2 for lat3. For all sky, we obtain 20{\pm}10 μK^2. We validate the network in a patch with strong contamination at small scales, obtaining an error of 50{\pm}120 μK^2 and residuals of 40{\pm}10 μK^2. Therefore, fully convolutional neural networks are promising methods to perform component separation in future CMB experiments. Particularly, CENN is reliable against different levels of contamination from Galactic and point source foregrounds at both large and small scales.

astro-ph.CO↗

arXiv2021

Multi-frequency point source detection with fully convolutional networks: Performance in realistic microwave sky simulations

Point Source (PS) detection is an important issue for future Cosmic Microwave Background (CMB) experiments since they are one of the main contaminants to the recovery of CMB signal at small scales. Improving its multifrequency detection would allow to take into account valuable information otherwise neglected when extracting PS using a channel-by-channel approach. We develop a method based on Neural Networks (NNs) to detect PS in multifrequency realistic simulations and compare its performance against one of the most popular methods, the matrix filters. The frequencies used are 143, 217 and 353 GHz and we impose a Galactic cut of 30 degrees. We produce simulations by adding contaminating signals to the PS maps as the CMB, the Cosmic Infrared Background, the Galactic thermal emission, the thermal Sunyaev-Zel'dovich effect and the instrumental noise. These simulations are used to train two NNs called Flat and Spectral MultiPoSeIDoN. The first one considers PS with a flat spectrum and the second one is more realistic because it takes into account the spectral behavior of the PS. Using a detection limit of 60 mJy, Flat MultiPoSeIDoN reachs the 90% of completeness level at 58 mJy and at 79, 71 and 60 for the spectral case at 143, 217 and 353 GHz respectively, while the matrix filters reach it at 84, 79 and 123 mJy. Using safer 4σ detection limit does not help to improve these results. In all cases, MultiPoSeIDoN obtain a much lower number of spurious sources than the filter. The NNs recover the flux density of the detections with a relative error of 10% above 100 mJy, while the filter above 150 mJy. Based on the results, NNs are the perfect candidates to substitute filters to detect multifrequency PS in future CMB experiments. Moreover, we have shown that a multifrequency approach can detect sources with higher accuracy than single-frequency approaches also based on NNs.

astro-ph.IM↗

arXiv2021

Point Source Detection with Fully-Convolutional Networks: Performance in Realistic Simulations

Point sources (PS) are one of the main contaminants to the recovery of the cosmic microwave background (CMB) signal at small scales, and their detection is important for the next generation of CMB experiments. We develop a method (PoSeIDoN) based on fully convolutional networks to detect PS in realistic simulations, and we compare its performance against one of the most used PS detection method, the Mexican hat wavelet 2 (MHW2). We produce realistic simulations of PS taking into account contaminating signals as the CMB, the cosmic infrared background, the Galactic thermal emission, the thermal Sunyaev-Zel'dovich effect, and the instrumental and PS shot noises. We first produce a set of training simulations at 217 GHz to train the network. Then we apply both PoSeIDoN and the MHW2 to recover the PS in the validating simulations at all 143, 217, and 353 GHz, comparing the results by estimating the reliability, completeness, and flux density accuracy and by computing the receiver operating characteristic curves. In the extra-galactic region with a 30° galactic cut, the network successfully recovers PS at 90% completeness corresponding to 253, 126, and 250 mJy for 143, 217, and 353 GHz respectively. The MHW2 with a 3$σ$ flux density detection limit recovers PS up to 181, 102, and 153 mJy at 90% completeness. In all cases PoSeIDoN produces a much lower number of spurious sources with respect to MHW2. The results on spurious sources for both techniques worsen when reducing the galactic cut to 10°. Our results suggest that using neural networks is a very promising approach for detecting PS, providing overall better results in dealing with spurious sources with respect to usual filtering approaches. Moreover, PoSeIDoN gives competitive results even at nearby frequencies where the network was not trained.

astro-ph.GA↗

arXiv2019

QSOs sigposting cluster size halos as gravitational lenses: halo mass, projected mass density profile and concentration at $z\sim0.7$

Magnification bias is a gravitational lensing effect that is normally overlooked because it is considered sub-optimal in comparison with the lensing shear. Thanks to the demonstrated optimal characteristics of the sub-millimetre galaxies (SMGs) for lensing analysis, in this work we were able to measure the magnification bias produced by a sample of QSOs acting as lenses, $0.2 13.6_{-0.4}^{+0.9}$, also confirmed by the mass density profile analysis ($M_{200c}\sim 10^{14} M_\odot$). These mass values indicate that we are observing the lensing effect of a cluster size halo signposted by the QSOs, as in previous studies of the magnification bias. Moreover, we were able to estimate the lensing convergence, $κ(θ)$, for our magnification bias measurements down to a few kpcs. The derived mass density profile is in good agreement with a Navarro-Frank-White (NFW) profile. We also attempt an estimation of the halo mass and the concentration parameters, obtaining $M_{NFW}=1.0^{+0.4}_{-0.2}\times10^{14} M_\odot$ and $C=3.5_{-0.3}^{+0.5}$. This concentration value is rather low and it would indicate that the cluster halos around these QSOs are unrelaxed. However, higher concentration values still provides a compatible fit to the data.

astro-ph.CO↗

arXiv2016

Experience with wavefront sensor and deformable mirror interfaces for wide-field adaptive optics systems

Recent advances in adaptive optics (AO) have led to the implementation of wide field-of-view AO systems. A number of wide-field AO systems are also planned for the forthcoming Extremely Large Telescopes. Such systems have multiple wavefront sensors of different types, and usually multiple deformable mirrors (DMs). Here, we report on our experience integrating cameras and DMs with the real-time control systems of two wide-field AO systems. These are CANARY, which has been operating on-sky since 2010, and DRAGON, which is a laboratory adaptive optics real-time demonstrator instrument. We detail the issues and difficulties that arose, along with the solutions we developed. We also provide recommendations for consideration when developing future wide-field AO systems.

astro-ph.IM↗

arXiv2008

Kinetics of crystallization of FeB-based amorphous alloys studied by neutron thermo-diffractometry

Kinetics of crystallization of two amorphous alloys, Fe70Cr10B20 and Fe80Zr10B10, have been followed up by neutron thermodiffractometry experiments performed in the two axis diffractometer D20 (ILL, Grenoble). The structural changes are directly correlated with the temperature dependence of the magnetization. Fe70Cr10B20 crystallizes following a two-step process: an eutectic crystallization of alfa-Fe (bcc) and the metastable tetragonal phase (Fe0.8Cr0.2)3B followed by another eutectic transformation to the stable phase (Fe0.75Cr0.25)2B and more segregation of alfa-Fe. These tetragonal phases are magnetically anisotropic, giving rise to a large increase of the coercivity. This behaviour is similar to that of Fe80B20 alloys, with Cr atoms replacing the Fe positions in both crystalline phases. Fe80Zr10B10 shows also a two-step process in which two polymorphic transformations take place.

cond-mat.mtrl-sci↗