Rubin-LSST France, Mai 2022

Europe/Paris
Auditorium (LAPP, Annecy)

Auditorium

LAPP, Annecy

Description

The next biannual meeting of the Rubin-LSST France community will be held 16-18 May 2022, at LAPP, Annecy.

  • Begins : ~1:30pm on Monday 16 May
  • Ends : ~4pm on Wednesday 18 May

To reach LAPP from the train station you can take bus number 1 or 4 and stop at "Campus". LAPP will be on the left at the roundabout with the LEP copper RF cavity.

Note: a cold buffet will be served on-site before the start of the meeting for those arriving before/around lunchtime (see option during registration)

Organisation:

  • Talks: Acknowledging that some participants may not understand/speak French, using English is encouraged.
  • Speakers: upload your slides on the indico prior to your talk.

It will be possible to follow the meeting on zoom, but please only register for in-person attendance.

The zoom link will be : https://us02web.zoom.us/j/86836910032?pwd=Q0VnQmxyNWRmUmhROU51Vm9uaytPdz09

The agendas of the most recent editions of this meeting can be found online:

Participants
  • Bastien Carreres
  • Benjamin Racine
  • Biswajit Biswas
  • Calum Murray
  • Celine Combet
  • Claire JURAMY
  • Constantin Payerne
  • Cyrille Doux
  • Cécile Roucelle
  • David Maurin
  • Didier Verkindt
  • Dominique Boutigny
  • Dominique Fouchez
  • Emille Ishida
  • Emmanuel Gangler
  • Eric Nuss
  • Estelle Robert
  • Etienne Russeil
  • Fabio Hernandez
  • Fabrice Feinstein
  • franck Lesplingart
  • Gabriele Mainetti
  • Guy Augarde
  • Johann Cohen-Tanugi
  • Joseph Chevalier
  • Julien Peloton
  • Jérémy Neveu
  • Kélian SOMMER
  • Madeleine Ginolin
  • Manon Ramel
  • marc moniez
  • Marie Paturel
  • Marina Masson
  • Marina Ricci
  • Marine Kuna
  • Martín Rodríguez Monroy
  • Mathew Smith
  • Michel Aguena
  • Pierre-Alain Duc
  • Reza Ansari
  • Roman Le Montagner
  • ROYA MOHAYAEE
  • Sabine Elles
  • Simona Mei
  • Sylvie Dagoret-Campagne
  • thibault guillemin
  • Vincent Reverdy
  • Yves Zolnierowski
    • 12:30 14:00
      On-site cold Buffet 1h 30m Auditorium

      Auditorium

      LAPP, Annecy

    • 14:00 15:55
      General news Auditorium

      Auditorium

      LAPP, Annecy

    • 15:55 16:25
      Pause 30m Grande galerie

      Grande galerie

    • 16:25 18:10
      Science Auditorium

      Auditorium

      LAPP, Annecy

      • 16:25
        News from the LSST Galaxies Science Collaboration 30m

        The LSST consortium is made of several science collaborations, including DESC for which the French community is particularly active. However a number of French researchers especially from INSU also have an important contribution in the Galaxies Science Collaboration. I will review these activities, with an emphasis on the LSB science, and sky background subtraction which is crucial to preserve the low surface brightness information in the images.

        Orateur: Pierre-Alain Duc (Observatoire astronomique de Strasbourg)
      • 16:55
        Testing the cosmological principle 35m

        The standard model of cosmology is founded on the cosmological principle which
        is the hypothesis of large-scale homogeneity and isotropy of the Universe.This hypothesis imposes that the rest frame of distant galaxies coincides with the rest frame of CMB. I will present some of the results we have obtained using the latest catalogues of radio galaxies and quasars which show that the cosmological principle might be violated [1,2]. I will then briefly discuss how the Rubin/LSST photometric catalog of galaxies, extending to redshifts beyond z >~ 2, could be used to check and possibly confirm the above mentioned results.
        [1] https://ui.adsabs.harvard.edu/abs/2021arXiv211008868V/abstract
        [2] https://ui.adsabs.harvard.edu/abs/2017MNRAS.471.1045C/abstract

        Orateur: Roya Mohayaee (Institut d'Astrophysique de Paris/Sorbonne université)
      • 17:30
        The Rubin-Euclid Derived Data Products (DDPs) 30m

        The Rubin LSST and the Euclid survey will each deliver groundbreaking astronomical datasets over this decade in the optical and near-infrared. Both surveys will map thousands of square degrees of sky from the ground and space respectively, with
        an overlap area of approximately 9000 square degrees at high galactic latitudes. The combination of Euclid’s high spatial resolution imaging in the optical and near-infrared photometry with Rubin’s densely sampled deep multi-band optical imaging will greatly enhance the science yield of both surveys. Indeed, while each survey on its own is poised for breakthrough science, their combination is likely to be truly transformative. We report here on the products recommended by both the Rubin and Euclid scientific communities.

        Orateur: Jean-Charles Cuillandre (CEA Saclay)
    • 09:30 10:30
      Science Auditorium

      Auditorium

      LAPP, Annecy

      • 09:30
        Fink 25m

        We will review the status of Fink, the latest scientific results, and the plans for the Rubin commissioning (and beyond).

        Orateur: Dr Julien Peloton (CNRS-IJCLab)
      • 09:55
        RESSPECT - optimizing spectroscopic follow-up 20m

        The recent increase in volume and complexity of available astronomical data has led to a wide use of supervised machine learning techniques. Active learning strategies have been proposed as an alternative to optimize the distribution of scarce labeling resources. However, due to the specific conditions in which labels can be acquired, fundamental assumptions, such as sample representativeness and labeling cost stability cannot be fulfilled. The Recommendation System for Spectroscopic followup (RESSPECT) project was born from an inter-collaboration agreement between the LSST Dark Energy Science Collaboration (DESC) and the Cosmostatistics Initiative (COIN), and aims to enable the construction of optimized training samples for the Rubin Observatory Legacy Survey of Space and Time (LSST), taking into account a realistic description of the astronomical data environment. In this talk, I will give updates on the status of the project, papers and pipeline. I will describe the main results of a dedicated paper where the cosmology metric is described (currently in internal review in both, DESC and COIN) and show details of the queried sample when ran in simulated data.

        Orateur: Dr Emille Ishida (CNRS/LPC-Clermont)
      • 10:15
        Identification of Orphan Gamma-Ray Burst Afterglows in Rubin LSST data with the afterglowpy package 15m

        Gamma-Ray Bursts (GRB) are among the most energetic phenomena in the Universe. The interaction of their blast wave with the Interstellar Medium produces an afterglow that can be observed from a larger angle, in a wide range of the electromagnetic spectrum and during more time than the prompt emission. Viewed off-axis, this emission has a negligible gamma-ray flux and is hence called ”GRB orphan afterglow”. Their properties make them good candidates to learn more about the GRB physics and progenitors or for the development of multi-messenger analysis, like in the case of GW170817A. According to most theoretical models, orphan afterglows should be found as slow and faint transients. This is why the Rubin Observatory shall significantly improve their detection : thanks to its limiting nightly magnitude of 24.5 and its large field of view, it should be able to detect up to 50 orphans per year. To identify orphan afterglow in Rubin LSST data, we plan to use the characteristic features of their lightcurves which depends on several parameters. In this work, we simulated afterglow light curves using the afterglowpy package to understand which combinations of these parameters lead to potentially observable transients by Rubin. We found that orphans should be more observable for low redshifts, high energies and even high angles during several months. These results will be used to generate a population of GRBs to simulate observations with the Rubin Observatory and will ultimately allow us to implement a filter in the alert broker FINK.

        Orateur: Marina Masson (UGA - LPSC)
    • 10:30 11:00
      Pause 30m Grande galerie

      Grande galerie

    • 11:00 12:30
      Science Auditorium

      Auditorium

      LAPP, Annecy

      • 11:00
        Update on the new ZTF calibration pipeline 20m

        I will present an overview of the new ZTF calibration pipeline developped at in2p3, giving an update on different steps: bias corrections, flatfielding, star-flats, photometric calibration (ubercal), and scene modeling photometry.

        Orateur: Benjamin Racine (CPPM/IN2P3/CNRS)
      • 11:20
        Premiers résultats de l'échantillon "volume limited" de ZTF 20m

        Je présenterais des résultats préliminaires de l'échantillon "volume limited" de SNeIa de ZTF : distributions en stretch, couleur,... ainsi que les dépendances des magnitudes des SNe à ces mêmes paramètres.

        Orateur: Madeleine Ginolin (IP2I)
      • 11:40
        NaCl : Nouveaux algorithmes de Courbes de lumière 20m

        One main approach to constrain the dark energy equation of state is to construct a Hubble diagram, the evolution of luminosity distance with respect to the redshift using Type Ia Supernovae (SN Ia) used as luminosity distance indicators.
        Measuring distances to SNe Ia requires a model of the SN spectrophotometric evolution, taking into account the intrinsic diversity of SNe Ia.

        The model currently in use in the community is called Spectral Adaptive Lightcurve Template 2 (SALT2), developed between 2007 (Guy \& al. 2007) and 2010 (Guy \& al. 2010).
        The state of the art is called SALT2.4 and was trained for the Joint Light Curve Analysis (Betoule \& al. 2014).
        Recently a model update has been published, called SALT3 (Kenworthy \& al 2021), with a new training set.

        I am currently developing a re-implementation of SALT2 intending to improve the general methodology and overall training speed. It is called NaCl, for Nouveaux algorithmes de Courbes de lumière.

        My model follows SALT2 parametrization. The goal was to overcome SALT2 limitations identified within the SALT2 framework. The first advantage of the new code is a simplification of the training procedure. The five steps
        of the previous SALT2 fit (training, intrinsic error modeling, retraining accounting for error model, re-evaluation of the error model, and finally light-curve fitting) have been implemented into a single-step comprehensive fit. As a result, error propagation is exact and straightforward.

        In particular, the propagation of calibration systematic uncertainties is now taken into account, while it was propagated through several trainings.

        Orateur: Guy Augarde (LPNHE)
      • 12:00
        DAG Forward modelling de fsigma8 15m

        On développe un code de forward modelling dans le but d'extraire une valeur du taux de croissance des structures fsigma8. Je vais présenter le Direct Acyclic Graph (DAG) de ce code.

        Orateur: Estelle Robert (IP2I)
      • 12:15
        Growth rate with type Ia supernovae 15m

        Type Ia supernova are known to be standard candles, that means that we can infer their distance from their flux measurement and build their Hubble Diagram. Peculiar velocities can be retrieved from the HD residuals, but until now the statistic of SN Ia was too low to use these velocities as cosmological probe. With the next generation of surveys (LSST, ZTF) the statistics of supernovae will grow in an unprecedented way, making the use of SN Ia peculiar velocities useful to measure the growth rate and to complement current measurements using galaxy surveys. In this talk, we propose to present our current work on the analysis to measure the growth rate using SN Ia, with the methodology and the study of bias estimation and mitigation.

        Orateur: Bastien Carreres (CPPM)
    • 14:00 15:30
      Présentations "commissioning" (Auxtel, CBP, focal plane...) + hands-on Auditorium

      Auditorium

      LAPP, Annecy

      • 14:00
        The Collimated Beam Projector : characterisation and application to the StarDice telescope 20m

        The Collimated Beam Projector is a hardware system to measure the optical transmission of telescopes. I will show how it has been characterized, used to measure the StarDice transmission at the sub-percent level, and lessons for the future LSST CBP.

        Orateur: Jérémy Neveu (LPNHE)
      • 14:20
        First detection of atmospheric effects on AuxTel data 25m

        The measurement of equivalent widths and the measurement of telescope transmission from Bouguer lines will be presented

        Orateur: Dr Sylvie Dagoret-Campagne (IJCLab)
      • 14:45
        Flat-fielding for slitless spectroscopy 20m

        We will compare flat-fields obtained in 2 bandwidths, and show their good correlation for small scale fluctuations. We will propose a specific deflating technique for spectroscopy.

        Orateur: Dr Martin Rodriguez-Monroy (IJCLab)
      • 15:05
        LSST focal plane commissioning: bias frame corrections 25m

        I will first present the general context of my commissioning work on Run 5 focal plane data performed at CC-IN2P3. I will then focus on studies of the bias frame corrections, which are required for the science image processing: the optimal strategy still remains to be defined and the chosen method will have to be fully implemented in the DM stack.

        Orateur: thibault guillemin (LAPP)
    • 15:30 16:00
      Pause 30m Grande galerie

      Grande galerie

    • 16:00 18:00
      Présentations "commissioning" (Auxtel, CBP, focal plane...) + hands-on Auditorium

      Auditorium

      LAPP, Annecy

    • 19:30 22:00
      Diner 2h 30m Brasserie des Européens

      Brasserie des Européens

      https://www.brasserie-des-europeens.com/ 4 Esplanade. de l'Hôtel de ville, 74000 Annecy

      https://www.brasserie-des-europeens.com/
      4 Esplanade. de l'Hôtel de ville, 74000 Annecy

    • 09:30 10:35
      Science Auditorium

      Auditorium

      LAPP, Annecy

      • 09:30
        Machine Learning for cluster detection 25m

        We explore machine learning networks to detect galaxy clusters in optical images. We present first result on the SDSS. Our networks are very efficient to reproduce the redmapper cluster catalog.

        Orateur: Simona Mei
      • 09:55
        Comparison of cluster finder algorithms on cosmoDC2 data 20m

        I will present a comparison of the performance of the redMaPPer, WaZP and AMICO cluster algorithms on cosmoDC2 data, focusing on the estimated completeness and purity as a function of the mass and the redshift of the simulated dark matter halos. Some first studies on the properties of clusters found specifically by one algorithm will also be presented, highlighting the complementarity of the algorithms.

        Orateur: thibault guillemin (LAPP)
      • 10:15
        Impact of photometric redshifts on cluster detection using DC2 data 20m

        The abundance of galaxy clusters is a powerful probe for cosmology, especially on large optical surveys where hundreds of thousands can be detected. It does require, however, an appropriate characterization of the observational effects on the detection of clusters and on their properties. In this work we evaluate the impact of photometric redshifts using the Wavelet Z Photometric (WaZP) cluster finder on the DC2 simulation. WaZP is a code developed for cluster detection on photometric surveys without the assumption of a red-sequence presence. As such, it allows us to probe cosmology until higher redshifts (>1) and it can provide a unique perspective to study the red-sequence abundance and evolution on galaxy clusters. The Data Challenge 2 (DC2) is a simulated sky survey that produced an observed catalog from simulated images. We ran WaZP using redshifts with increasing levels of complexity, from true redshifts to photo-zs and present here how the selection function and cluster properties are affected. Each of those cases produced a cluster catalog that will be shared within the DESC collaboration for additional scientific analysis. The WaZP cluster finder is also being ported to a python version, which we expect to make public available soon.

        Orateur: Dr Michel Aguena (LAPP)
    • 10:35 11:05
      Pause 30m Grande galerie

      Grande galerie

    • 11:05 12:20
      Science Auditorium

      Auditorium

      LAPP, Annecy

      • 11:05
        Weak lensing magnification around galaxy clusters 20m

        Weak lensing is a powerful tool to estimate the matter distribution around massive galaxy clusters. In this work we consider weak lensing magnification. In general such effects are measured by counting the number of galaxies in circular annuli from the lens centre. In this work we present a new method which improves on the standard approach by using the full galaxy magnitude distribution, rather than a single faint-end magnitude cut. This allows us to differentiate between the two opposing effects of magnification, whereby faint galaxies are introduced into an annulus as they are magnified above the faint-end magnitude limit and removed from an annulus as they are deflected away from the lens centre. Magnification of a galaxy image will shift the galaxy magnitude distribution and the deflection effects will change the normalisation of the distribution, therefore the effects are separable and we can increase the signal-to-noise ratio of such measurements. We investigate this new method using HSC weak lensing data and the Redmapper SDSS galaxy cluster catalogue.

        Orateur: Calum Murray (LPSC)
      • 11:25
        Likelihoods for cluster abundance cosmology 20m

        A standard choice for cosmological cluster analyses is a Poissonian likelihood; however such a likelihood neglects the effects of sample variance from density fluctuations across the survey volume. To date this assumption has been justified but to make the most of the future data such as that of the Rubin Observatory and the Euclid satellite, improvements to the cluster likelihood must be considered. The simplest way to deal with the effects of sample variance is by using a Gaussian approximation, however with such an approximation we lose valuable cosmological information. We present a framework for testing accuracy of the standard choices of likelihood by comparing to a distribution which is able to account for both the Poisson-like statistics of galaxy cluster abundance and the additional sample variance.

        Orateur: Constantin Payerne (LPSC (IN2P3))
      • 11:45
        Study of the blending impact on galaxy clusters with DC2 15m

        Due to the greater depth of the survey and blurring by the atmosphere, over a half of the galaxies observed by the LSST will appear to overlap on the line of sight. This effect, called blending, perturbs detections and measurements of shapes and fluxes of galaxies, and is expected be even stronger in high-density regions such as galaxy clusters. In this talk, I will present recent work based on the DC2 image simulations that aims at evaluating the level of blending in clusters and its impact on observables such as shear profiles and inferred cluster masses.

        Orateur: Manon Ramel (LPSC / IN2P3)
      • 12:00
        Maximum-A-posteriori solution with Deep generative NEtworks for Source Separation (MADNESS) 20m

        As the Large Survey of Space and Time (LSST) will detect fainter objects, the increased object density will lead to more overlapping sources. For LSST, we expect around 60% of galaxies to be blended. In order to better constrain Dark Energy parameters, mapping the matter content of our Universe with weak gravitational lensing is one of the main probes for the upcoming large cosmological surveys and the blending effect is expected to be one of the major systematics to face. Classical methods for solving the inverse problem of source separation, so-called “deblending”, either fail to capture the diverse morphologies of galaxies or are too slow to analyze billions of galaxies. To overcome these challenges, we propose a deep learning-based approach to deal with the size and complexity of the data.

        Taking forward the work on Debvader, a deblender that uses a modified form of Variational Autoencoders, our algorithm called MADNESS deblends by finding the Maximum-A-posteriori solution parameterized by latent space representation of galaxies generated with a deep generative model. We first train a VAE as a generative model and then model the underlying latent space distribution so that can it be sampled to simulate galaxies. To perform deblending, we do a gradient descent to find the MAP estimate, i.e., the particular latent space realization that minimizes the sum of negative log-likelihood and negative log probability of z being a galaxy.

        In my talk, I will outline the methodology of our algorithm and evaluate its performance using flux reconstruction as a metric.

        Orateur: Biswajit Biswas (APC)
    • 14:00 14:35
      Science Auditorium

      Auditorium

      LAPP, Annecy

      • 14:00
        Study of a hybrid PhotoZ estimator and potential improvements 15m

        We compare the performances of a PhotoZ estimator Delight, a hybrid method of template fitting (based here on SED CWW templates) and machine learning.

        In a first step, we compare the performance of Delight's z-phot vs z-spec regression on DC2 simulations with that of LePhare's "state of the art" template fitting method and a machine learning method, the ramdom forest trees.

        In a second step, we try to improve the performance of Delight in two ways:
        - Implementation of the photo detection bias effect called Edignton-Malsmquist bias in the templates after adjusting the threshold of photo detection in each of the filters on the DC2 data,
        - A pre-classification of the DC2 simulations into the CWW template SED categories, followed by a learning of the Flux redshift relationship for each category.

        Orateur: Joseph Chevalier (IJCLab)
      • 14:15
        Introduction to the DESC Bayesian analysis project and to the JAX-cosmo library 20m

        The DESC Bayesian analysis pipeline project recently created aims to coordinate and promote the development of a standardized open source Bayesian Pipeline which ultimately could connect pixels to cosmology. Two subprojects has emerged: one concerning "Pixels to shear" starting from simulated/DC2 images to infer posterior weak lensing shear and convergence fields on the sky (eg. BLISS,...), while the other connect "Shear & LSS to cosmology" throw a Bayesian Hierarchical Model to infer cosmological parameters from tomographic maps of galaxy density and shear using catalogs (eg. BORG, KarMMA,...). This is an extremely challenging task as inference is performed in high dimension while beeing fast & accurate from end-to-end. GPU-accelerated automatically differentiable frameworks will be one keystone of this project. Along this line a small piece is the JAX-cosmo library part of the DifferentiableUniverseInitiative initiated by F.Lanusse etal. JAX-cosmo is a light & proof-of concept equivalent of CCL library written in full JAX (in short an "Autodifferentable-Numpy") allowing to run code both on CPU & GPU asis. An exercise that will reported has been undertaken to compare different MCMC methods in the context of DES (shear&number counts) Y1 using JAX-cosmo.

        Orateur: Jean-Eric Campagne (LAL-IN2P3-CNRS and Univ. Paris 11)
    • 14:35 15:05
      Cosmic shear: from DES to LSST 30m Auditorium

      Auditorium

      LAPP, Annecy

      Orateur: Cyrille Doux (Laboratoire de Physique Subatomique et de Cosmologie)
    • 15:05 15:35
      A walk through cosmological simulations and their evolution 30m Auditorium

      Auditorium

      LAPP, Annecy

      Cosmological simulations have become crucial for the preparation of observational surveys as well as for the calibration of analysis workflows. But if they can mimic accurately some properties of our Universe, they will also always miss some. Thus, knowing and understanding what is being done behind the scene can be key for good science.

      In this talk, I will summarize the main principles and techniques behind cosmological simulations. I will especially focus on elements that can be key for post-processing analyses and that should be kept in mind in order to avoid circular dependencies. I will discuss how machine learning algorithms can sometimes learn numerical artifacts better than physics phenomena and what can be done about it.

      Before concluding I will finally review some of the recent developments on the topic of cosmological simulations and give some insights from the french community that is working on the topic.

      Orateur: Dr Vincent Reverdy (Laboratoire d'Annecy de Physique des Particules)
    • 15:35 16:00
      Discussion and Concluding remarks 25m Auditorium

      Auditorium

      LAPP, Annecy

    • 16:00 16:30
      Pause 30m Grande galerie

      Grande galerie