Search

Future astrophysical surveys will provide unprecedented datasets in both scale and precision for testing our most concordant theories of dark matter and dark energy. For the first time, the confidence of our understanding of the components of the Universe will be limited by systematic uncertainties rather than the statistical precision we can achieve. Improved modeling of measurement biases in survey data is necessary to achieve the theory-constraining power of ambitious future astrophysical surveys. Recent in-lab, on-sky, and simulated experiments offer opportunities to investigate coherence between systematic and phenomenological observables and invent new techniques to differentiate them. This thesis details my contribution to measuring the Dark Universe, and it is structured in two parts. The first part of this thesis improves template modeling of the foreground gamma-ray emission in the Galactic plane in 12 years of Fermi-LAT data. This has implications for the observed excess \gray{} emission near the Galactic center, which was previously associated with dark matter. We compare the likelihood of observing gamma-ray emission from cold hydrogen gas clouds using models derived from spectral line emission maps of abundant ($^{12}\mathrm{CO}$) and rare ($^{13}\mathrm{CO}$) tracer molecules. Using this approach, the rare tracer predicts previously unmodeled gamma-ray emission at $47\sigma$ significance, localized to the pixels around the densest regions, where emission is undercounted due to reabsorption along the line-of-sight. We also identify unassociated point-like sources in the Galactic plane which were previously smoothed out by re-absorption consistent with peaked gas structure as well as point-like gas structure which could have been mistaken for millisecond pulsars by earlier works. We further utilize a convolutional neural network (CNN) to try to predict the $^{13}\mathrm{CO}$ distribution in the same region, which could prove useful to predict the $^{13}\mathrm{CO}$ distribution in data-poor regions such as the Galactic center. The second part of this thesis develops a correction for the observed size-magnitude relation of stars, which are used calibrate observables for the Legacy Survey of Space and Time (LSST). The brighter-fatter effect results in distorted images of bright calibration stars, creating a flux-dependent point spread function (PSF) that if left unmitigated, could make up a large fraction of the error budget in Stage IV weak-lensing (WL) surveys such as the Legacy Survey of Space and Time (LSST). Using data from the LSST camera, we compare the BFE in flat field and star images and test the ability of scalar correction methods to mitigate it. We observe that the BFE evolves anisotropically as a function of flux due to higher-order BFEs which were previously assumed to be negligible by a commonly used correction technique. We also find that poor assumptions of the correction result in unphysical violations of flux conservation in the images at the O(0.01\%) level. Based on these analyses, we then propose a new, unbiased correction methodology to account for these effects. With these improvements we can correct on average 95\% of the total effect in the sensors we tested. We use these results to predict the impact of these previously unmodeled effects on sensitive observables of inferred cosmological outcomes and advise data processing strategies to limit the impact of the effects on LSST science pathways.

The Legacy Survey of Space and Time (LSST) carried out by the Vera C. Rubin Observatory is one of the most ambitious optical surveys planned in the near future. It will generate $\sim 20$~TB of data every night during its 10 year survey duration. Simulated data sets (``Data Challenges'') have been generated by the LSST Dark Energy Science Collaboration (DESC) to allow development of efficient analysis algorithms, evaluation of their performance and scaling capabilities, and to study the impact of systematic uncertainties. In this work we focus on the analysis of Large Scale Structure using galaxies selected from the DESC Data Challenge 2 (DC2) corresponding to a survey area of 300 square degrees. We develop and perform an end-to-end analysis of the galaxy Large Scale Structure with the DC2 data, validating the extraction of galaxy power spectra and galaxy bias in the presence of complex systematics. We provide a range of strong consistency checks of both the simulated data and the analysis techniques.Based on matching selected galaxies with the DC2 truth catalog, we optimize selection to obtain a clean sample of galaxies. After generating and applying survey masks, we produce galaxy density maps binned in redshift, with six bins from $z = 0.2 - 1.4$. We generate the power spectra measurements from the processed data and fit the data using theoretical predictions to estimate the galaxy bias as a function of redshift. Finally, we study the effect of systematic uncertainties on estimation of the galaxy bias.We find that the relative uncertainty $\sigma_\ell / C_\ell$ in the measured angular power spectra is smaller than 10\% in most cases, and smaller than 4\% at multipoles above 1500.%Changes in the angular power spectra due to systematic uncertainties are found to be on average of order 0.5\%, i.e. $\Delta C_\ell / C_\ell \approx 0.5\%$, with a maximum deviation of 1.5\%.%Systematic effects are small compared to the statistical uncertainty of the measured angular power spectra: $\Delta C_\ell \sim 0.1 \sigma_\ell$ on average with a maximum of $0.25 \sigma_\ell$.

Rubin Observatory has made open science a core value of its mission since the beginning of construction and will continue to do so throughout the full 10 years of the Legacy Survey of Space and Time (LSST). We believe that open science is only meaningful if it is supported by well-documented open-source reusable software; our more than 1.5 million lines of code are all publicly accessible on github. But open source is not sufficient; we need to lower barriers to doing science especially for under-resourced and under-represented communities. The Rubin Science Platform (RSP) (LSE-319) democratises access to LSST data products by providing the same level of ease and power to any user anywhere with an internet connection. Rubin Data Previews (RTN-011), early shared-risk previews of the LSST data products and services, are hosted on Google and supported by the Community Science Team’s tutorials and hack sessions. Data Preview 0 has already demonstrated the potential of the RSP for rapidly performing scientific analyses for anyone in the community. The algorithms within our processing software are also discussed with and in some cases contributed to by community members. In addition to open source software and a compute platform, the third leg of our "open science support tripod" is documentation. We consider accessible and complete documentation for code and data products as essential for lowering barriers to participation, and continue to push towards this goal. There are many further challenges to open science such as interoperability. We continue to work with IVOA on standardisation and are committed to our “VO First” approach. We build on the existing standards that enable interoperability; however, we acknowledge that they may not scale to Rubin data volumes and are discussing what to do about that. Our pioneering cloud based solution may also help with interoperability - having all datasets in one place is hard, having them all on one cloud may be possible. We have identified and tackled some of the open science issues and are sure to encounter more on our ten year journey making the Legacy Survey of Space and Time.&nbsp

Overview of Rubin Observatory at AAS 241

Data processing pipelines need to be executed at scales ranging from small runs up through large production data release runs resulting in millions of data products. As part of the Rubin Observatory's pipeline execution system, BPS is the abstraction layer that provides an interface to different Workflow Management Systems (WMS) such as HTCondor and PanDA. During the submission process, the pipeline execution system interacts with the Data Butler to produce a science-oriented execution graph from algorithmic tasks. BPS converts this execution graph to a workflow graph and then uses a WMS-specific plugin to submit and manage the workflow. Here we will discuss the architectural design of this interface and report briefly on the recent production of the Data Preview 0.2 release and how the system is used by pipeline developers., 4 pages, submitted to Astronomical Data Analysis Software and Systems XXXII, October 2022

Rubin Observatory is committed to enabling high-impact science during the first year of operation of the Legacy Survey of Space and Time (LSST). This document provides the plans for a dedicated Early Science Program designed to realize that goal. It outlines the strategy for obtaining observations during commissioning to enhance opportunities for Early Science and presents plans to implement incremental template generation to augment alert production in the early phases of the survey. The Rubin Operations team will work closely with the science community to design the Early Science Program so as to maximize the time-domain and solar system science achievable in the first year of operations. This is a living document; both it and the Early Science Program will continue to evolve over the course of commissioning and pre-operations in response to the state of the as-built system and to community guidance.

Next generation experiments such as the Vera Rubin Observatory Legacy Survey of Space and Time (LSST) will provide an unprecedented volume of time-domain data opening a new era of big data in astronomy. To fully harness the power of these surveys, we require analysis methods capable of dealing with large data volumes that can identify promising transients within minutes for follow-up coordination. In this talk I will present Fink, a broker developed to face these challenges. Fink is based on high-end technology and designed for fast and efficient analysis of big data streams. It has been chosen as one of the official LSST brokers and will receive the full data stream. I will highlight the state-of-the-art machine learning techniques used to generate early classification scores for a variety of time-domain phenomena including kilonovae and supernovae, as well as for artifacts, like satellites glitches. Such methods include Deep Learning advances and Active Learning approaches to coherently incorporate available information, delivering increasingly more accurate added values throughout the duration of the survey. I will also highlight the potential for discovery that can be unveiled by adding domain knowledge as part of the learning process. This will require an adaptation of learning system as well as as an update on how domain knowledge experts interact with the data taking process., {"references":["Moller et al, 2021, DOI:10.1093/mnras/staa3602","Leoni et al, 2021, aXiv: https://arxiv.org/abs/2111.11438"]}

This report is the result of a joint discussion between the Rubin andEuclid scientific communities.The workpresented in this report was focused on designing and recommending an initial set of Derived Data products (DDPs) that could realize the science goals enabled by joint processing. All interested Rubin and Euclid data rights holders were invited to contribute via an online discussion forum and a series of virtual meetings. Strong interest in enhancing science with joint DDPs emerged from across a wide range of astrophysical domains: Solar System, the Galaxy, the Local Volume, from the nearby to the primaeval Universe, and cosmology.&nbsp

Rubin Observatory is committed to enabling high-impact science during the first year of operation of the Legacy Survey of Space and Time (LSST). This document provides the plans for a dedicated Early Science Program designed to realize that goal. It outlines the strategy for obtaining observations during commissioning to enhance opportunities for Early Science and presents plans to implement incremental template generation to augment alert production in the early phases of the survey. The Rubin Operations team will work closely with the science community to design the Early Science Program so as to maximize the time-domain and solar system science achievable in the first year of operations. This is a living document; both it and the Early Science Program will continue to evolve over the course of commissioning and pre-operations in response to the state of the as-built system and to community guidance.

Pocino, A., et al. (Euclid Collaboration), Photometric redshifts (photo-zs) are one of the main ingredients in the analysis of cosmological probes. Their accuracy particularly affects the results of the analyses of galaxy clustering with photometrically selected galaxies (GCph) and weak lensing. In the next decade, space missions such as Euclid will collect precise and accurate photometric measurements for millions of galaxies. These data should be complemented with upcoming ground-based observations to derive precise and accurate photo-zs. In this article we explore how the tomographic redshift binning and depth of ground-based observations will affect the cosmological constraints expected from the Euclid mission. We focus on GCph and extend the study to include galaxy-galaxy lensing (GGL). We add a layer of complexity to the analysis by simulating several realistic photo-z distributions based on the Euclid Consortium Flagship simulation and using a machine learning photo-z algorithm. We then use the Fisher matrix formalism together with these galaxy samples to study the cosmological constraining power as a function of redshift binning, survey depth, and photo-z accuracy. We find that bins with an equal width in redshift provide a higher figure of merit (FoM) than equipopulated bins and that increasing the number of redshift bins from ten to 13 improves the FoM by 35% and 15% for GCph and its combination with GGL, respectively. For GCph, an increase in the survey depth provides a higher FoM. However, when we include faint galaxies beyond the limit of the spectroscopic training data, the resulting FoM decreases because of the spurious photo-zs. When combining GCph and GGL, the number density of the sample, which is set by the survey depth, is the main factor driving the variations in the FoM. Adding galaxies at faint magnitudes and high redshift increases the FoM, even when they are beyond the spectroscopic limit, since the number density increase compensates for the photo-z degradation in this case. We conclude that there is more information that can be extracted beyond the nominal ten tomographic redshift bins of Euclid and that we should be cautious when adding faint galaxies into our sample since they can degrade the cosmological constraints.

Weak gravitational lensing is one of the most promising probes to constrain dark energy parameters. It corresponds to the distortion of a source image, induced by the bending of space-time, thus of the light path, generated by the presence of mass along the line of sight. This effect is small and can only be detected measuring the correlation of the shapes of a group of background galaxies. The value of this correlation yields the value of the cosmic shear. Several surveys dedicated to its study are going to start soon, such as the Vera Rubin Legacy Surveys of Space and Time (LSST), and Euclid. These surveys have been built with particular requirements, especially about the treatment of systematic errors, in order to reach an extreme precision on the dark energy equation of state parameters. This thesis happened in the context of the Dark Energy Science Collaboration (DESC), inside the LSST experiment. It focuses on the issue of blending, the overlap of astronomical sources. The blending of galaxies introduces a dominant systematic uncertainty on the cosmic shear measurement. This systematic effect impacts the shape and the redshift measurements, both necessary for weak gravitational lensing analysis. We propose two avenues based on Bayesian deep learning methods. The first one is a deblending algorithm which uses a deep generative network called variational autoencoder. This neural network allows to learn a prior for the generation of isolated galaxy images. The latter is used in a second network to perform the deblending of the centred galaxy on images of simulated galaxies. We show that the pixel joint analysis of LSST and Euclid data decreases the median error on galaxy shape reconstruction from 8 to 47%. Blending of sources being closely linked to detection, we demonstrate that our method is robust to small decentring. Also we test our method on images of real galaxies artificially blended showing the interest of transfer learning from a neural network trained on simulated galaxy images. An iterative process is then designed in order to separate all the galaxies in an image going through detection, classification and deblending of sources. Finally this deblending algorithm is tested on images extracted from the DC2 simulation, generated within the DESC to prepare for the analysis of futures images taken by LSST. Our results show an improvement of 70 to 120% on the median error on galaxy shape reconstruction compared to the generic method used in the current LSST pipeline. Then, inspired by this first method, we propose a neural network allowing for the direct estimation of galaxy shape and redshift parameters from DC2 images, without going through the deblending. We show that this neural network allows for a precise measurement of these parameters, even when sources are blended, and we compare these results to the ones obtained with deblending. Then we present the first application of a Bayesian neural network to galaxy shapes estimation. This kind of network provides the estimation of the epistemic uncertainty which characterizes the uncertainty coming from the training sample. It can be viewed as a confidence level in the measurement performed by the network, a primordial information for the application of neural networks in science and in cosmology. It can be used to decrease the importance (or reject) of an incorrect measurement in the analysis.; Le lentillage gravitationnel faible est l'une des sondes cosmologiques les plus prometteuses pour contraindre les paramètres de l'énergie noire. Il correspond à la modification de l'image d'une source, induite par la courbure de l'espace-temps, et donc du passage des rayons lumineux, générée par la présence de masse le long de la ligne de visée. Cet effet est très faible et ne peut être détecté qu'en mesurant la corrélation des formes d'un ensemble de sources en arrière-plan. La valeur de cette corrélation donne la mesure du cisaillement gravitationnel. Plusieurs relevés dédiés à son étude vont bientôt démarrer, tels que le Vera Rubin Legacy Survey of Space and Time (LSST), et Euclid. Ces relevés ont été construits avec des exigences particulières, notamment sur le traitement des erreurs systématiques, afin d'atteindre une précision extrême sur les paramètres de l'équation d'état de l'énergie noire. Cette thèse s'est déroulée dans le contexte de la Dark Energy Science Collaboration (DESC), au sein de l'expérience LSST. Elle se focalise sur la problématique de la superposition de sources. Cet effet génère une incertitude systématique dominante dans la mesure du cisaillement gravitationnel. Cette systématique impacte à la fois les mesures de forme et celles du décalage vers le rouge des galaxies, deux mesures nécessaires à l'étude du lentillage gravitationnel faible. Nous proposons deux solutions basées sur les méthodes bayésiennes d'apprentissage profond. La première est un algorithme de séparation de galaxies qui utilise un réseau de neurones génératif appelé autoencodeur variationnel. Ce réseau de neurones permet l'apprentissage d'un prior pour la génération d'images de galaxie isolée. Celui-ci est utilisé dans un second réseau qui fait la séparation de la galaxie centrale sur des images de galaxies simulées. Nous montrons que l'analyse jointe des pixels des données LSST et Euclid permet une diminution de l'erreur médiane de reconstruction des formes des galaxies de 8 à 47%. La superposition de sources étant étroitement liée à l'étape de détection, nous démontrons la robustesse de notre méthode à de petits décentrages. Nous avons aussi testé notre méthode sur des images de vraies galaxies artificiellement superposées, montrant l'intérêt du transfert d'apprentissage à partir d'un réseau de neurones entraîné sur des images de galaxies simulées. Une processus itératif est ensuite mis en place afin de pouvoir séparer toutes les galaxies d'une image en passant par des étapes de détection, classification, et séparation des sources. Enfin, cet algorithme de séparation de galaxies est testé sur des images extraites de la simulation DC2, générée au sein de DESC dans le but de préparer à l'analyse des futures images prises par LSST. Nos résultats montrent une amélioration de 70 à 120% sur l'erreur médiane de reconstruction des formes comparée à la méthode générique utilisée dans la pipeline LSST actuellement. Ensuite, inspirés de cette première méthode, nous proposons un réseau de neurones permettant de faire directement l'estimation des paramètres de forme et de décalage vers le rouge d'une galaxie à partir des images DC2, sans appliquer de méthode de séparation de sources. Nous montrons que ce réseau permet des mesures précises de ces paramètres, même lorsque les sources sont superposées, et comparons ces résultats à ceux obtenus avec séparation de source. Par la suite, nous présentons une première application d'un réseau de neurones bayésien à l'estimation des formes des galaxies. Ce type de réseau permet l'estimation de l'incertitude épistémique qui représente l'incertitude liée à l'échantillon d'entraînement. Celle-ci peut être vue comme un niveau de confiance dans la mesure réalisée par le réseau, une information cruciale à l'utilisation des réseaux de neurones en science et en cosmologie. Elle permet de diminuer l'importance (voir de rejeter) d'une mesure qui a de fortes probabilités d'être erronée dans l'analyse.

We present the data used in "DeepAdversaries: Examining the Robustness of Deep Learning Models for Galaxy Morphology Classification". Data processing and analysis pipelines in cosmological survey experiments introduce data perturbations that can significantly degrade the performance of deep learning-based models.We consider perturbations associated with two primary sources: 1) increased observational noise as represented by higher levels of Poisson noise; 2) data processing noise incurred by steps such as image compression or telescope errors as represented by one-pixel adversarial attacks. We also test the efficacy of domain adaptationtechniques in mitigating the perturbation-driven errors.Successful development and implementation of methods that increase model robustness in astronomical survey pipelines will help pave the way for many more uses of deep learning for astronomy. To create our data, which emulates Vera C. Rubin LSST observations, we useGalSim (Rowe et al. 2015)and follow the same procedure as in Sanchez et al. (2021).We create two sets of survey-emulating images, by applying an exposure timedirectly to the raw images: 1) low-noise ten-year survey ("Y10'') - baseline data our models are trained on; 2) high-noise one-year survey ("Y1'') - noisy data. We use one-pixel attack(Su et al. 2017) to attack a 150-image test set sub-sampleof Y10 images,in order to emulate pixel-level perturbations, that might naturally arise in astronomical data pipelines. Data: Images Y10: images_Y10_train.npy; images_Y10_valid.npy; images_Y10_test.npy Noisy mages Y1: images_Y1_train.npy; images_Y1_valid.npy; images_Y1_test.npy Labels valid for both Y10 and Y1: labels_train,npy; labels_valid.npy; labels_test.npy Small 150-image sub-samples: Images: images_Y10_test_150.npy; images_Y1_test_150.npy Labels: labels_test_150.npy

A report on the Fall 2020 state of the LSST Alert Production Pipelines' ability to process deep images in the galactic bulge.

Traditional photometric catalogue matching will fail for Rubin Obs. LSST, as typical 2” error circles will contain up to 10 random objects. Bayesian cross-matching techniques which include the astrometric uncertainty of the observations ultimately assume that the Astrometric Uncertainty Function of the sources is Gaussian, which will not apply to LSST. The extreme source densities and significant number of objects with no known proper motion mean that systematic effects, not noise-based scatter, dominant the separations of objects between LSST and other surveys. As part of the LSST:UK consortium we have developed and are implementing methods to handle these effects, vital to being able to trust any composite dataset created, and will be producing robust cross-matches between LSST and a wide range of other catalogues., {"references":["Sutherland & Saunders (1992): On the likelihood ratio for source identification.","Wright et al. (2010): The Wide-field Infrared Survey Explorer (WISE): Mission Description and Initial On-orbit Performance","Ricker et al. (2015): Transiting Exoplanet Survey Satellite (TESS)","Wilson & Naylor (2017): The effect of unresolved contaminant stars on the cross-matching of photometric catalogues","Wilson & Naylor (2018): Improving catalogue matching by supplementing astrometry with additional photometric information","Wilson & Naylor (2018): A contaminant-free catalogue of Gaia DR2-WISE Galactic plane matches: including the effects of crowding in the cross-matching of photometric catalogues"]}

Euclid is an ESA mission designed to constrain the properties of dark energy and gravity via weak gravitational lensing and galaxy clustering. It will carry out a wide area imaging and spectroscopy survey (EWS) in visible and near-infrared, covering roughly 15,000 square degrees of extragalactic sky on six years. The wide-field telescope and instruments are optimized for pristine PSF and reduced straylight, producing very crisp images. This paper presents the building of the Euclid reference survey: the sequence of pointings of EWS, Deep fields, Auxiliary fields for calibrations, and spacecraft movements followed by Euclid as it operates in a step-and-stare mode from its orbit around the Lagrange point L2. Each EWS pointing has four dithered frames; we simulate the dither pattern at pixel level to analyse the effective coverage. We use up-to-date models for the sky background to define the Euclid region-of-interest (RoI). The building of the reference survey is highly constrained from calibration cadences, spacecraft constraints and background levels; synergies with ground-based coverage are also considered. Via purposely-built software optimized to prioritize best sky areas, produce a compact coverage, and ensure thermal stability, we generate a schedule for the Auxiliary and Deep fields observations and schedule the RoI with EWS transit observations. The resulting reference survey RSD_2021A fulfills all constraints and is a good proxy for the final solution. Its wide survey covers 14,500 square degrees. The limiting AB magnitudes ($5\sigma$ point-like source) achieved in its footprint are estimated to be 26.2 (visible) and 24.5 (near-infrared); for spectroscopy, the H$_\alpha$ line flux limit is $2\times 10^{-16}$ erg cm$^{-2}$ s$^{-1}$ at 1600 nm; and for diffuse emission the surface brightness limits are 29.8 (visible) and 28.4 (near-infrared) mag arcsec$^{-2}$., Comment: 43 pages, 51 figures, submitted to A&A

The Large Synoptic Survey Telescope (LSST), currently under construction in Chile, will be the most comprehensive optical astronomy project ever undertaken. Currently in the final stages of construction, it is designed to perform a "wide-fast-deep" survey over the entire visible sky at a 3-day cadence. The resulting hundreds of petabytes of imaging data for close to 40 billion objects will be used for scientific investigations ranging from the properties of near-Earth asteroids to characterizations of dark matter and dark energy. The volume, quality, and the real-time aspects of the LSST survey present significant challenges as well as exciting research opportunities. In this talk I will give an overview and status of the LSST Project, and take an in-depth look at the primary deliverables of LSST,which include a nightly stream of public “alerts” to transient, variable, and moving objects, approximately annual Data Releases, and the LSST Science Platform (LSP), a collaborative research environment through which the community will perform large-scale analysis of the LSST data.

A brief presentation on the LSST project and how it relates to the brown dwarf community.

Poster paper for the IAA Planetary Defense Conference describing the current state of the LSST Moving Object Processing System.

The Alert Production (AP) Pipeline for the Large Synoptic Survey Telescope (LSST) must promptly process new images from the telescope to enable rapid-response time domain science. It begins with raw telescope data and uses image differencing to identify time-variable sources and associate them with previously detected objects. We present an overview of the AP Pipeline and illustrate its utility with Dark Energy Camera (DECam) data from the High Cadence Transient Survey (HiTS). We use DECam images from 2014 as a template for images which revisit the same region of the sky in 2015. We compile catalogs of detected sources and associated objects, create light curves for a handful of interesting objects, and compare initial results with the first HiTS data release. The AP Pipeline is part of the LSST Science Pipelines and is available on GitHub athttps://github.com/lsst/ap_pipe.

Este proyecto trata un problema de clasificación de objetos astronómicos a partir de los datos registrados por el telescopio LSST, correspondientes a series históricas de flujo. En primer lugar, presentamos las técnicas de aprendizaje computacional y minería de datos utilizadas en el proyecto. A continuación, se describen los conceptos astronómicos necesarios para el trabajo. Analizamos los datos mediante técnicas de minería de datos y agrupamos las muestras en función de dos de sus características. Transformamos los datos, calculando la magnitud y el color de los objetos. Sometemos a los datos a un proceso de reducción de ruido y a una inferencia bayesiana de sus valores de flujo. Convertimos los datos en series temporales a las que realizamos un proceso de extracción de características. Reducimos las características, mediante sucesivas clasificaciones de forma iterativa, hasta encontrar la dimensionalidad óptima y construimos con dichas características tres clasificadores de 1, 2 y 4 bosques aleatorios. Validamos el modelo y discutimos los resultados. Finalmente se proponen líneas de trabajo para el futuro. Aquest projecte tracta un problema de classificació d'objectes astronòmics a partir de les dades registrades pel telescopi LSST, corresponents a sèries històriques de flux. En primer lloc, presentem les tècniques d'aprenentatge computacional i mineria de dades utilitzades en el projecte. A continuació, es descriuen els conceptes astronòmics necessaris per al treball. Analitzem les dades mitjançant tècniques de mineria de dades i agrupem les mostres en funció de dues de les seves característiques. Transformem les dades, calculant la magnitud i el color dels objectes. Sotmetem a les dades a un procés de reducció de soroll i a una inferència bayesiana dels seus valors de flux. Convertim les dades en sèries temporals a les que realitzem un procés d'extracció de característiques. Reduïm les característiques, mitjançant successives classificacions de forma iterativa, fins a trobar la dimensionalitat òptima i construïm amb aquestes característiques 03:00 classificadors d'1, 2 i 4 boscos aleatoris. Validem el model i discutim els resultats. Finalment es proposen línies de treball per al futur. This project addresses a problem of classification of astronomical objects from the data recorded by the LSST telescope, corresponding to historical flow series. First, we present the techniques of computational learning and data mining used in the project. Next, the astronomical concepts necessary for the work are described. We analyze the data using data mining techniques and group the samples according to two of their characteristics. We transform the data, calculating the magnitude and color of the objects. We submit the data to a noise reduction process and to a Bayesian inference of its flow values. We convert the data into time series to which we perform a feature extraction process. We reduce the characteristics, through successive classifications iteratively, until we find the optimal dimensionality and build with these characteristics three classifiers of 1, 2 and 4 random forests. We validate the model and discuss the results. Finally, work lines are proposed for the future.

Type Ia supernovae (SNIa) are transient objects, which remains observable in the optical for a period of a few months, and whose luminosity at its maximum is comparable to that of a whole galaxy. They represent, once standardized (i.e. corrected for their main variabilities), an excellent distance indicator and, in 1998, provided the first evidence for the acceleration of the expansion of the Universe. This thesis investigates three points in the cosmological SNIa pipeline:First, we focused on the data processing and the quality of the images for the Large Synoptic Survey Telescope (LSST), that will provide data for a number of cosmological observables, including the observation of SNIa. In preparation for LSST (which will start operations in 2020), we carried out a similar data processing strategy on images from a previous survey (deep field of CFHT) and compared to the required parameters of LSST. A dispersion slightly higher than that required was obtained. For photometry, although this excess is small, the sources of this dispersion remain to be understood. For astrometry, the obtained quality is likely to be sufficient when simultaneous astrometry fitting will be implemented.In a second moment, we propose a method to correct the effect of peculiar velocities of SNIa inside galaxy clusters. Indeed, these velocities ​​are more important in the clusters of galaxies, because of the gravitational interactions, and the measurement of the cosmological redshift related to the expansion of the Universe. These peculiar velocities were corrected using a sample of 145 SNIa with a low spectral shift (0.005

Type Ia supernovae (SNIa) are transient objects, which remains observable in the optical for a period of a few months, and whose luminosity at its maximum is comparable to that of a whole galaxy. They represent, once standardized (i.e. corrected for their main variabilities), an excellent distance indicator and, in 1998, provided the first evidence for the acceleration of the expansion of the Universe. This thesis investigates three points in the cosmological SNIa pipeline:First, we focused on the data processing and the quality of the images for the Large Synoptic Survey Telescope (LSST), that will provide data for a number of cosmological observables, including the observation of SNIa. In preparation for LSST (which will start operations in 2020), we carried out a similar data processing strategy on images from a previous survey (deep field of CFHT) and compared to the required parameters of LSST. A dispersion slightly higher than that required was obtained. For photometry, although this excess is small, the sources of this dispersion remain to be understood. For astrometry, the obtained quality is likely to be sufficient when simultaneous astrometry fitting will be implemented.In a second moment, we propose a method to correct the effect of peculiar velocities of SNIa inside galaxy clusters. Indeed, these velocities ​​are more important in the clusters of galaxies, because of the gravitational interactions, and the measurement of the cosmological redshift related to the expansion of the Universe. These peculiar velocities were corrected using a sample of 145 SNIa with a low spectral shift (0.005

LSST images will be contaminated with transient artifacts, such as optical ghosts, satellite trails, and cosmic rays, and with transient astronomical sources, such as asteroid ephemerides. We developed and tested an algorithm to find and reject these artifacts during coaddition, in order to produce clean coadds to be used for deep detection and preliminary object characterization. This algorithm,CompareWarpAssembleCoadduses the time-series of PSF-matched warped images to identify transient artifacts. It detects artifact candidates on the image differences between each PSF-matched warp and a static sky model. These artifact candidates include both true transient artifacts and difference-image false positives such as difficult-subtract-sources and variable sources such as stars and quasars. We use the feature that true transients appear at a given position in the difference images in only a small fraction (configurable) of visits, whereas variable sources and difficult-to-subtract sources appear in most difference images. In this report, we present a description of the method and an evaluation using Hyper SuprimeCam PDR1 data.

Presentation in the ADASS XXVII (2017).

LSST Data Management (DM) is a large, distributed organization. Most of our communications happen through online platforms. Rather than shoehorn all of our communications onto one online platform, DM uses the best platform for each format of communication. This presentation describes each of these communication channels, and how they are used. Platforms include Slack, community.lsst.org, Confluence, Requests for Comments, Requests for Discussion, and technical notes.

This briefing covers updates to LSST Data Management's project documentation and user documentation projects. New project documentation features include: LaTeX support, change-controlled document release process, the Projectmeta (www.lsst.io) document index, the new templates repository, new methods for copyright and license management, and the status of the Developer Guide reorganization. In the second half of the talk I describe the status of Science Pipelines user documentation, including a refresher on topic-based writing and the DMTN-030 architecture, and instructions on how to upgrade pipelines packages with the new Sphinx documentation.

LSST Updates Construction Progress Commissioning Plans, and Community Opportunities, Presented at the LSST Town Hall held as part of the 231st meeting of the American Astronomical Society.

The NOAO Community Science and Data Centre's Role in Maximising LSST Science, Presented at the LSST Town Hall held as part of the 231st meeting of the American Astronomical Society.

The LSST software systems make extensive use of Python, with almost all of it initially being developed solely in Python 2. Since LSST will be commissioned when Python 2 is end-of-lifed it is critical that we have all our code support Python 3 before commissioning begins.Over the past year we have made significant progress in migrating the bulk of the code from the Data Management system onto Python 3. This poster presents our migration methodology, and the current status of the port, with our eventual aim to be running completely on Python 3 by early 2018.We also discuss recent modernizations to our Python codebase. &nbsp

The gravitational field of a galaxy can act as a lens and deflect the light emitted by a more distant object such as a quasar. Strong gravitational lensing causes multiple images of the same quasar to appear in the sky. Since the light in each gravitationally lensed image traverses a different path length from the quasar to the Earth, fluctuations in the source brightness are observed in the several images at different times. The time delay between these fluctuations can be used to constrain cosmological parameters and can be inferred from the time series of brightness data or light curves of each image. To estimate the time delay, we construct a model based on a state-space representation for irregularly observed time series generated by a latent continuous-time Ornstein-Uhlenbeck process. We account for microlensing, an additional source of independent long-term extrinsic variability, via a polynomial regression. Our Bayesian strategy adopts a Metropolis-Hastings within Gibbs sampler. We improve the sampler by using an ancillarity-sufficiency interweaving strategy and adaptive Markov chain Monte Carlo. We introduce a profile likelihood of the time delay as an approximation of its marginal posterior distribution. The Bayesian and profile likelihood approaches complement each other, producing almost identical results; the Bayesian method is more principled but the profile likelihood is simpler to implement. We demonstrate our estimation strategy using simulated data of doubly- and quadruply-lensed quasars, and observed data from quasars Q0957+561 and J1029+2623., Accepted for publication in the Annals of Applied Statistics

The frequency-dependent spectral index of the atmosphere leads to spectrum-dependent distortions of sources away from zenith. While the effect of bulk refraction is easily corrected during astrometric calibration, this Differential Chromatic Refraction (DCR) introduces distortions that depend on the intrinsic spectrum of a source and vary with the parallactic angle and airmass of the observation. Past surveys have avoided DCR by restricting observations to low airmass, but LSST operations will require observations up to an airmass of 1.4 across the sky and up to 2 in some fields. With current image differencing techniques, we would have to choose between a high rate of false positive detections that would have to be culled with machine learning, or restrict template images to have similar airmasses and parallactic angles of the science image. However, the refraction of the atmosphere has been modeled for decades with increasing refinement, and we can use iterative forward modeling to estimate a sky model using that knowledge of the atmosphere to fit all observations of a patch simultaneously. We use this model to calculate new DCR-matched templates for image differencing, and find that it reduces the number of spurious source detections in DECam g-band images.

The Large Synoptic Survey Telescope (LSST) plans to produce a live stream of its detections of transient astronomical events within 60 seconds of observation at an expected rate of about 10,000 alerts per visit and 10 million per night. Complementary observations of astronomical objects by different types of telescopes and instruments at a range of wavelengths is essential for our understanding of these astrophysical phenomena. Traditionally, a major limiting factor in transient science has been the lack of complementary observations of transient events given the timescale of distribution of alert information from the initial detection of an object by one telescope to another telescope relative to the short duration of the event. The use of distribution mechanisms such as the online Astronomer’s Telegram (ATel) and services relaying Virtual Observatory Event (VOEvent) messages have significantly alleviated the dissemination bottleneck, but in the LSST big data era, the bottleneck to discovery will likely shift from distribution to the ability to quickly isolate and prioritize detections of interest amid the flood of alert events. Real-time processing capabilities for automated filtering and analysis of these alerts and subsequent integration capabilities with distribution mechanisms is critical for enabling fast follow-up action and observations of time sensitive astronomical activity. For LSST Project internal pipeline needs, we are prototyping a modular system for alert stream distribution and also processing (filtering) using existing open source technologies specifically for streaming data. This system is also planned to be adapted for use for the upcoming Zwicky Transient Facility (ZTF) survey. The technologies used in this system provide a number of advantages for the LSST use case, including 1) built-in replication, scalability, and stream “rewind” for the distribution mechanism, 2) structured messages with strictly enforced schemas and dynamic typing for fast parsing, and 3) a stream processing interface that is similar to batch for a familiar and user-friendly plug-in filter system, all in a modular, primarily containerized system. In this presentation, we will discuss this internal alert distribution and processing prototype system, the design motivations around the technology choices for the framework, and how this system may be generally suitable for other alert stream use cases.