COLOURS

2 juin 2025, 09:00 → 13 juin 2025, 17:00 Europe/Paris

Institut Pascal

 

 

Cosmological Surveys and Synergies (COLOURS)

June 2nd-13th 2025 - Institut Pascal, Orsay, France 

 

COLOURS is a 2-week event aimed at ensuing that early-career researchers (ECR) are prepared to make the most of Stage-IV cosmological survey data. 

The first week of the event (02/06/25-06/06/25) will be a 5-day summer school aimed at improving both technical (software development, machine learning, statistics) and soft skills (academic writing, transferable skills). Site visits and social activities will also be organised.

The second week (10/06/25-13/06/25) will be a 4-day workshop with invited speakers. The aim will be to review key lessons learned in the analysis of Stage-III data and the scientific aspirations of on-going and upcoming Stage-IV surveys. 

Confirmed Workshop Speakers

• David Alonso (Oxford, UK)
• Silvia Galli (IAP, FR)
• Daniel Gruen (LMU, DE)
• Catherine Hale (Oxford, UK)
• Alex Hall (ROE, UK)
• Valeria Pettorino (ESA ESTEC, NL)
• Marta Spinelli (OCA, FR)
• Pauline Zarrouk (LPNHE, FR)

 

Confirmed School Presenters and Topics

• Melissa Ann Thomas (Academic Writing Center - Centralesupélec, FR) - Scientific communication tutorials: The importance of well-written grant proposals and effective research presentations
• Andrina Nicola (U. Bonn, DE) - Putting it all together - an (incomplete) guide to cosmological probe combination
• Giovanni Aricò (INFN-BOL, IT) - N-body simulations
• Natalia Porqueres (CEA, FR) - Field-level and simulation-based inference
• Stefano Camera (UniTo, IT) - 
• Samuel Farrens (CEA, FR) - Scientific software development: A hands-on guide to making robust and professional software
• Transferrable skills pannel and presentations by:
  • Ben Hoyle (Zeiss, DE)
  • Denise Lanzieri (Sony CSL, IT)
  • Fred Ngolè Mboula (CEA LETI, FR)
  • Prina Patel (Mastercard, UK)
• Course on machine learning
• More to come!

 

Participation

Successful ECR applicants will be provided with accommodation, offices and lunches for the full durations of the COLOURS programme (i.e. for 2 weeks). Places will be limited.

The workshop (i.e. the second week of the programme) will be open to researchers of all career stages. There are no conference fees, but no acommodation or other financial support will be provided.

Fill in the application form below to express your interest in attending!

Application dealine: 31/01/2025*

*Due to the limited number of places, we may choose to close the application window before the deadline. We encourage you to express your interest as early as possible to ensure the best chance of being selected.

Organisers 

SOC

• Samuel Farrens (CEA, FR)
• Stefano Camera (U. Turin, IT)
• Natalia Porqueres (CEA, FR)
• Elena Sellentin (U. Leiden, NL)
• Marta Spinelli (OCA, FR)
• Jean-Luc Starck (CEA, FR & FORTH, GR)

 

LOC

• Anna Guerreschi (IPa, FR)
• Samuel Farrens (CEA, FR)
• Fabian Hervas Peters (CEA, FR)
• Sacha Guerrini (CEA, FR)
• Natalia Porqueres (CEA, FR)
• Vilasini Tinnaneri Sreekanth (CEA, FR)

colours_coc.pdf

colours_poster_yellow.pdf

lun. 2 juin ¶

Registration: School¶

10:20 Coffee

COLOURS School: Opening¶

Président de session: Samuel Farrens (CosmoStat, CEA Paris-Saclay)

1 Welcome to the Institut Pascal¶

Orateur: Yves BALKANSKI (Institut Pascal - UPSaclay)

2 Welcome to COLOURS¶

Orateur: Samuel Farrens (CosmoStat, CEA Paris-Saclay)

COLOURS School: Cosmology I¶

Président de session: Calum Murray (CosmoStat, CEA Paris-Saclay)

3 Clustering around clusters with Euclid Q1 data¶

Orateur: Calum Murray (CosmoStat, CEA Paris-Saclay)

13:00 Lunch

Buffet at the IPa

COLOURS School: Cosmology II¶

Président de session: Natalia Porqueres (CosmoStat, CEA Paris-Saclay)

4 Field-level and simulation-based inference - Part I¶

Orateur: Natalia Porqueres (CosmoStat, CEA Paris-Saclay)

16:00 Coffee

5 Field-level and simulation-based inference - Part II¶

Orateur: Natalia Porqueres (CosmoStat, CEA Paris-Saclay)

Social: Ice Breaker¶

mar. 3 juin ¶

COLOURS School: Data Science I¶

Président de session: Samuel Farrens (CosmoStat, CEA Paris-Saclay)

6 Scientific Software Development - Part I¶

A hands-on guide to making robust and professional software

Orateur: Samuel Farrens (CosmoStat, CEA Paris-Saclay)

11:00 Coffee

7 Scientific Software Development - Part II¶

A hands-on guide to making robust and professional software

Orateur: Samuel Farrens (CosmoStat, CEA Paris-Saclay)

13:00 Lunch

Canteen

COLOURS School: Data Science II - Group B¶

COLOURS School: Pasqal Site Visit - Group A¶

mer. 4 juin ¶

COLOURS School: Communication¶

Président de session: Melissa Ann Thomas (CentraleSupélec)

8 Scientific communication tutorials: The importance of well-written grant proposals and effective research presentations - Part I¶

Orateur: Melissa Ann Thomas (CentraleSupélec)

11:00 Coffee

9 Scientific communication tutorials: The importance of well-written grant proposals and effective research presentations - Part II¶

Orateur: Melissa Ann Thomas (CentraleSupélec)

13:00 Lunch

Canteen

COLOURS School: Data Science II - Group A¶

COLOURS School: Pasqal Site Visit - Group B¶

jeu. 5 juin ¶

COLOURS School: Cosmology III¶

Président de session: Stefano Camera (University of Turin)

10 Crosses and Chains - Part I¶

Orateur: Stefano Camera (University of Turin)

11:00 Coffee

11 Crosses and Chains - Part II¶

Orateur: Stefano Camera (University of Turin)

13:00 Lunch

Canteen

COLOURS School: Cosmology IV¶

Président de session: Giovanni Aricò (INFN Bologna)

12 How to simulate the large-scale structure of the Universe - Part I¶

Numerical simulations currently provide our most accurate predictions for structure formation in the Universe, in a wide range of scales and redshifts.
We will overview cosmological N-body simulations, discussing common assumptions, approximations and regime of applicability.

Furthermore, we will introduce the main hydrodynamical schemes, and review the sub-grid prescriptions used to model galaxy formation with its relevant astrophysical processes.

Finally, we will go through useful simulation post-processing techniques and tools for cosmological analyses.

Orateur: Giovanni Aricò (INFN Bologna)

16:00 Coffee

13 How to simulate the large-scale structure of the Universe - Part II¶

Numerical simulations currently provide our most accurate predictions for structure formation in the Universe, in a wide range of scales and redshifts.
We will overview cosmological N-body simulations, discussing common assumptions, approximations and regime of applicability.

Furthermore, we will introduce the main hydrodynamical schemes, and review the sub-grid prescriptions used to model galaxy formation with its relevant astrophysical processes.

Finally, we will go through useful simulation post-processing techniques and tools for cosmological analyses.

Orateur: Giovanni Aricò (INFN Bologna)

ven. 6 juin ¶

COLOURS School: Cosmology V¶

Président de session: Andrina Nicola (University of Bonn)

14 Putting it all together - an (incomplete) guide to cosmological probe combination - Part I¶

Orateur: Andrina Nicola (University of Bonn)

11:00 Coffee

15 Putting it all together - an (incomplete) guide to cosmological probe combination - Part II¶

Orateur: Andrina Nicola (University of Bonn)

13:00 Lunch

Canteen

COLOURS School: Transferable Skills¶

16 Transferable Skills Talk 1¶

Orateur: Prina Patel (Mastercard)

17 Generative AI at Carl ZEISS: Riding the run away train¶

In this talk I discuss how Generative AI is being adopted by the German production compamy Carl ZEISS, and the steps which have been taken to ensure usage aligns with our values. I highlight amazing use-cases, cutting edge projects & applications, and try to give some view of what future adoption might look. I will mention how we are pushing on the topic of employee training, to ensure everyone is brought along for the ride.

Orateur: Ben Hoyle (Carl ZEISS AG)

18 Transferable Skills Talk 3¶

Orateur: Denise Lanzieri (Sony CSL- Rome)

19 Datasets dictionary learning and applications to frugal AI¶

Empirical risk minimization learning paradigm works under the assumption that training and test data are identically distributed. However, this hypothesis is seldom met in practice, due to several factors including changes in the underlying physical process generating the data, data acquisition conditions or sensors drifts. This problem is known as distributional shift between the reference and target data. A straightforward generalization of this problem is when training with multiple datasets exhibiting distributional shifts.

In this context, optimal transport provides useful tools for comparing and manipulating probability measures. In particular, a Wasserstein barycenter is a meaninful way of averaging datasets in the space of probability measures equipped with the optimal transport based distance. Leveraging this concept, we introduce a novel dictionary learning problem over empirical probability measures and present its applications to different learning setting where no labels are available on the target dataset.

Orateur: Fred Ngolè (CEA Paris-Saclay)

15:50 Coffee

20 Panel Discussion on transferable skills¶

Panel members:
- Ben Hoyle (Zeiss, DE)
- Denise Lanzieri (Sony CSL, IT)
- Fred Ngolè Mboula (CEA LETI, FR)
- Prina Patel (Mastercard, UK)

Orateur: Samuel Farrens (CosmoStat, CEA Paris-Saclay)

Social: Dinner¶

mar. 10 juin ¶

Registration: Workshop¶

COLOURS Workshop: Opening¶

Président de session: Samuel Farrens (CosmoStat, CEA Paris-Saclay)

21 Welcome¶

Orateur: Samuel Farrens (CosmoStat - CEA Paris-Saclay)

COLOURS Workshop: Session I¶

Président de session: Samuel Farrens (CosmoStat, CEA Paris-Saclay)

22 Keynote - Overview of the ESA Euclid mission in the context of ESA Science today¶

After a brief introduction of the ESA Science programme I will focus on an overview of the ESA Euclid mission: I will describe mission objectives, challenges, main milestones so far, the first data release.

I will also mention a few ESA programmes dedicated to interns, graduate students, and postdocs.

Orateur: Valeria Pettorino (ESA ESTEC)

23 Suppressing the Growth of Structure with Modified Gravity Theories¶

Modifications of General Relativity have been widely studied to alleviate cosmological tensions. Most of these models lead to an effective strengthening of gravity and enhanced growth. However, the S8 tension — arising from weak-lensing observations that suggest less structure formation than predicted by ΛCDM — points to a different scenario. In this talk, we investigate stable subclasses of scalar-tensor theories that effectively weaken gravity. We demonstrate how the suppression of linear structure growth constrains phenomenological modified-gravity parameters. Finally, we compare these weak-gravity models to current data from Stage-IV surveys.

Orateur: Linus Thummel (University of Edinburgh)

24 Investigating the Evolution of the Universe Using the observational H(z) data (OHD): Constraining Cosmological Parameters, Dark Energy, and the Hubble Constant¶

We use the Cosmic Chronometers H(z) data, also the observational H(z) data (OHD), to investigate the universe's evolution, focusing on its applications in constraining cosmological parameters, probing dark energy, and refining methods for determining the Hubble constant (H0). (1) To improve cosmological model selection, we combine H(z) with fσ8(z), comparing the traditional chi-square approach with Linder’s joint H(z)-fσ8 diagram. Our results show that the joint method is more significant than the traditional combined method in constraining the density parameter (ΩM) based on Akaike and Bayesian Information Criteria (see for more details: Niu, J., & Zhang, T.-J. 2023, Physics of the Dark Universe, 39, 101147, doi:10.1016/j.dark.2022.101147, ). (2) In exploring dark energy, we reconstruct the scalar field potential V(φ) using Gaussian process analysis of H(z) data under various priors. We find that the reconstructed V(φ) is highly sensitive to the choice of prior and dataset. Simulations reveal that doubling the H(z) data points improves reconstruction accuracy by 5%–30%, underscoring the importance of data availability (see for more details: Niu, J., et al. 2024, ApJ, 972, 14 doi:10.3847/1538-4357/ad5fef, ). (3) For H0 determination, we evaluate three methods—EMCEE, Gaussian Process (GP), and Masked Autoregressive Flow (MAF). Sensitivity analysis identifies GP as the most influenced by individual data points, while simulations rank EMCEE as the most accurate, followed by MAF and GP.

Orateur: Jing Niu (School of Physics and Astronomy, Beijing Normal University)

11:00 Coffee

COLOURS Workshop: Session II¶

25 Detecting Relativistic Doppler by Multi-tracing a Single Galaxy Population¶

The detection of relativistic Doppler via galaxy power spectrum measurements could further confirm the validity of general relativity at scales very far from the strong-gravity-field regime, where it has been tested with exquisite accuracy. The Doppler term acts as an imaginary correction in the relation between the galaxy density contrast and that of matter, which mostly affects the large scales usually plagued by cosmic variance. Moreover, it is sample-dependent, so different galaxy populations display different contributions in their power spectra. In the search for the optimal galaxy samples to achieve a detection of the relativistic term, we can split a galaxy population according to luminosity, and then perform a multi-tracer analysis with auto-correlations of the sub-samples and their cross-power spectrum. I will be talking about such a technique which extracts multi-tracer benefits out of a single dataset and may lead us to a detection of the relativistic effect with data from ongoing galaxy surveys, such as Euclid and DESI.

Orateur: Federico Montano (University of Turin)

26 Accuracy of HOD models for stage IV surveys¶

Orateur: Romain Paviot (CEA Paris-Saclay)

27 Probing Inflation with PICO: Simulations and Delensing for CMB B-Modes¶

The polarized cosmic microwave background (CMB) signal can be decomposed into parity-even E-modes and parity-odd B-modes. The inflationary signal would manifest itself as a B-mode signal. An inflationary period at the beginning of the Universe imprints a B-mode signal in the polarization of the CMB and if the energy scale of inflation is large enough the signal may be detectable. At a maximum the signal is one part in 10^8
relative to the nearly uniform glow of the CMB, and it could be lower. Unambiguously detecting the signal requires precise subtraction of confounding foregrounds, which have much larger amplitude. One such foreground is B-modes generated by gravitational lensing of E-modes, a process called 'lensing'. Accounting for and removing this foreground is called 'delensing'.

PICO - the Probe of Inflation and Cosmic Origin - is a proposed NASA space mission that has been endorsed for implementation by the US Astro2020 Decadal Panel. Its data will give the most sensitive probe for the B-mode inflationary signal. I will describe my contributions to developing the mission through conducting simulations and validating the efficacy of delensing.

Orateur: Julien TANG (LBNL/CPB)

28 Autoencoders for AGN identification in the DESI Y3 survey¶

We describe a machine learning approach to multi-wavelength Active galactic nuclei (AGN) identification for host galaxies within the DESI survey. AGNs emit light in all wavelengths in the electromagnetic spectrum, it is difficult to create an AGN selection that is fully complete. The identification of AGNs is key to understanding not only their astrophysics, being an important driver of galaxy evolution, and affecting the galaxy-halo connection, but also the biases and systematic uncertainties for further cosmological analysis. With the abundance of large multi-wavelength surveys, the application of machine learning techniques (reconstruction of galaxy spectra with unsupervised learning, specifically autoencoders) could provide the solution to a more complete AGN identification technique. This provides a new method to produce accurate and more complete AGN selections in wide-field surveys.

Orateur: Dhavala Sai Srinivas (University of Portsmouth)

12:50 Lunch

Buffet at the IPa

COLOURS Workshop: Session III¶

29 Keynote - Latest cosmological results from DESI¶

Orateur: Pauline Zarrouk (CNRS/LPNHE)

30 Relativistic effects in galaxy clustering with DESI¶

Understanding the accelerated expansion of the Universe remains as one of the key challenges in cosmology. The main candidates to explain this observation, which do not rely on a cosmological constant, are dark energy and modifications of General Relativity, but they require robust tests on cosmological scales. The Dark Energy Spectroscopic Instrument offers unprecedented precision in measuring galaxy clustering from spectroscopic data, allowing for the detection of relativistic features beyond the standard redshift-space distortions. In particular, relativistic effects generate a dipole in the cross-power spectrum of two galaxy populations. Using mock catalogues of synthetic galaxies which mimic the DESI Bright Galaxy Survey (BGS), we analyse ways to amplify the relativistic dipole by separating these galaxies into bright and faint populations, while conserving their redshift distribution. We also examine techniques to accurately estimate the magnification bias, a key parameter entering the amplitude of the dipole signal. Our results indicate an improved detectability of the relativistic dipole with fewer bright sources and that the measured distortions are well described by the predictions of linear theory.

Orateur: Jade Piat (University of Edinburgh)

31 Densitysplit for the DESI BGS¶

The Bright Galaxy Survey (BGS) is the densest galaxy sample of the DESI project, and the closest in redshift range (z<0.2). At such small redshift, the constraint on cosmological parameters is dominated by the "cosmic variance", a statistical error due to the small volume of universe considered. Alternative clustering analysis methods, such as Density-Split or Multi-Tracer analysis have shown a clear improvement on those constraints. In this talk, I will present the method, its analysis pipeline using an emulator for the theoretical model in the cosmological inference and preliminary results for the DESI BGS.

Orateur: Simon Bouchard (LPNHE)

16:10 Coffee

COLOURS Workshop: Session IV¶

32 Model-agnostic assessment of dark energy after DESI DR1 BAO¶

Baryon acoustic oscillation measurements by the Dark Energy Spectroscopic Instrument (Data Release 1) have revealed exciting results that show evidence for dynamical dark energy at around 3sigma when combined with cosmic microwave background and type Ia supernova observations. These measurements are based on the w0waCDM model of dark energy. The evidence is less in other dark energy models such as the wCDM model. To avoid imposing a dark energy model, we reconstruct the distance measures and the equation of the state of dark energy independent of any dark energy model and driven only by observational data. To this end, we use both single-task and multi-task Gaussian Process regression. Our results show that the model-agnostic evidence for dynamical dark energy from DESI is much less than when imposing the w0waCDM model. Our analysis also provides model-independent constraints on cosmological parameters such as the Hubble constant and the matter-energy density parameter at present. We find that the reconstructed values of these parameters are inconsistent with the results reported in DESI with the w0waCDM model. However, they are almost consistent with DESI for the wCDM model.

Orateur: BIKASH RANJAN DINDA (University of the Western Cape)

33 FLAME: Fitting LyAlpha Absorption lines using Machine Learning¶

FLAME is a machine learning algorithm that fits Voigt profiles to HI Lyman-alpha absorption lines, using deep convolutional neural networks to classify components and determine key parameters like Doppler b, HI column density, and velocity separation. Trained on ~1M simulated HST-COS spectra, FLAME achieves high accuracy on simulated data but shows a slight performance drop with real data, demonstrating its potential for robust absorption line analysis.

Orateur: Priyanka Jalan (Center for Theoretical Physics - Polish Academy of Sciences)

34 Primordial Physics with the Lya forest¶

The Lyman-a forest traces large-scale structure at Mpc scales and below at high redshifts (2 < z < 4). The Lyman-a forest has provided pristine information about the expansion history of the Universe using BAO measurements. In this talk, I will talk about challenges and results on extracting more information from the Lyman-a forest. First, from a full-shape cosmological analysis and, second, by extracting higher-order information from the forest. I will present a first-principle modeling of the Lyman-a forest in the effective field theory of large-scale structure to go beyond information in the two-point correlation function. I will give an outlook on current and future challenges with the Lyman-a forest. In a nutshell, we can already now do Stage-V spectroscopy science with the Lyman-a forest.

Orateur: Roger de Belsunce (Lawrence Berkeley National Laboratory)

35 COSMIC: A Galaxy Cluster Finding Algorithm Using Machine Learning¶

Building a comprehensive catalog of galaxy clusters is a fundamental task for the studies on the structure formation and galaxy evolution. In this paper, we present COSMIC (Cluster Optical Search using Machine Intelligence in Catalogs), an algorithm utilizing machine learning techniques to efficiently detect galaxy clusters. COSMIC involves two steps, including the identification of the brightest cluster galaxies and the estimation of the cluster richness. We train our models on the galaxy data from the Sloan Digital Sky Survey and WHL galaxy cluster catalog. Validated to a test data in the region of northern Galactic cap, COSMIC algorithm demonstrates a high completeness when cross-matching with previous cluster catalogs. Richness comparison with previous optical and X-ray measurements also demonstrated a tight correlation. Our methodology showcases robust performance in galaxy cluster detection and holds promising prospects for applications in upcoming large-scale surveys.

Orateur: Da-Chuan Tian (School of Physics and Astronomy)

Social: Ice Breaker¶

mer. 11 juin ¶

Registration: Workshop¶

COLOURS Workshop: Session V¶

36 Keynote - Reconstructing the Universe's history with projected large-scale structure data¶

Orateur: David Alonso (University of Oxford)

37 The Fink broker: a personalized window to the transient sky through LSST¶

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 of new category of sources and how we can optimize for discovery in the era of LSST.

Orateur: Emille Ishida (CNRS/LPCA)

38 Probing Neutrino Mass using the Cosmic Web¶

We explore the effects of massive neutrinos and baryons on the cosmic web using the FLAMINGO simulations. With NEXUS+, we classify the cosmic web into voids, sheets, filaments, and clusters, and find that massive neutrinos affect the environment by decreasing the amount of filaments and sheets. The effect of baryons is 10% smaller. Constructing the minimum spanning tree (MST) from the haloes we find that most MST edges lie in filaments, which we link to its sensitivity to neutrino mass. The lengths of the MST edges are strongly correlated with cosmic web environments. We distinguish the effects of neutrinos from the effects of baryons on the edge lengths of the MST, emphasising its capability to go beyond two-point statistics.

Orateur: Leonor Simões (University College London)

11:00 Coffee

COLOURS Workshop: Session VI¶

39 Baryon-free S8 tension with stage IV cosmic shear surveys¶

Accurately modelling matter power spectrum effects at small scales, such as baryonic feedback, is essential to avoid significant bias in the estimation of cosmological parameters with cosmic shear. However, Stage IV surveys like LSST will be so precise that significant information can still be extracted from large scales alone. In this talk, I will present an analysis of LSST Year 1-like mock data, focusing on the impact of baryonic feedback modeling on cosmic shear constraints. By applying physically motivated, redshift-dependent scale cuts, we studied the changes in the constraining power of the cosmological parameters S8 and Om, as well as their implications for the tension with Planck measurements. Our results show that the S8 tension remains detectable even with conservative scale cuts, regardless of whether an incorrect model for baryons is assumed. I will also discuss the crucial role of high-precision measurements of redshift distributions in achieving these results.

Orateur: Ottavia Truttero (University of Edinburgh)

40 Growth of structure using cross-correlation of CMB lensing and cosmic shear¶

Weak lensing of cosmological sources is an important tool for studying the distribution of total matter in the universe. Cross-correlating weak lensing of the Cosmic Microwave Background (CMB) with weak lensing of galaxies ("cosmic shear") allows a way to place robust constraints on the amplitude of matter perturbations in the universe while minimizing the impact of certain systematic effects that affect individual surveys. I will discuss the work being done with Atacama Cosmology Telescope (ACT) CMB lensing and Dark Energy Survey (DES) cosmic shear cross-correlation. I will present results from the analysis of ACT DR4 CMB lensing and DES Y3 cosmic shear and discuss some aspects of modeling this observable.

Orateur: Shabbir Shaikh (Arizona State University)

41 Probing the Cosmological Principle with weak lensing shear¶

The Cosmological Principle is a cornerstone of the standard model of cosmology and shapes how we view the Universe and our place within it. It is imperative, then, to devise multiple observational tests which can identify and quantify possible violations of this foundational principle. One possible method of probing large-scale anisotropies involves the use of weak gravitational lensing. In this talk, I will outline how late-time anisotropic expansion would imprint itself upon the cosmic shear signal. Thereafter, I will speak about the detectability of this anisotropic signature in upcoming surveys like Euclid. In particular, I will motivate the importance of the cross-correlation of shear E- and B-modes on large angular scales as a possible probe of large-scale anisotropy.

Orateur: James Adam (University of the Western Cape)

42 Shear calibration for UNIONS¶

Orateur: Fabian Hervas Peters (CosmoStat, CEA Paris-Saclay)

12:50 Lunch

Canteen

COLOURS Workshop: Session VII¶

43 Keynote - Weak lensing and 3x2pt with Euclid¶

Euclid is a European Space Agency telescope designed to map the dark Universe and uncover the nature of cosmic acceleration. It is conducting a large survey of galaxies, which will be used to measure the large-scale mass distribution of the Universe through weak gravitational lensing, galaxy clustering, and their combination ('3x2pt'). I will explain how we plan to use Euclid's imaging to constrain cosmological models and describe specific challenges that must be overcome to deliver precise results. I will place Euclid in the context of other existing/planned observational cosmology programmes, and discuss how Euclid might influence the landscape of viable cosmological models and cast light upon current tensions in Lambda-CDM.

Orateur: Alex Hall (University of Edinburgh)

44 ElFNet: Ellipticity Finding Network¶

Noise bias is a persistent challenge in shear measurements, particularly for galaxies with low signal-to-noise ratios (SNR), which are critical for Euclid’s cosmic shear analysis. To address this, we present a novel, unsupervised, deep learning denoising technique (ElFNet) to improve galaxy shape measurements. Using a convolutional neural network (CNN), we predict individual galaxy shears directly from simulated postage-stamp galaxy images. Our method combines adaptive moments with best-fit Gaussian PSF corrections and incorporates Metacalibration to calibrate the shear. Compared to GalSim, our approach demonstrates significant improvements in reducing noise bias.

Orateur: Lucy Reynolds (Institut de Física d'Altes Energies (IFAE))

45 Staying focused: In-flight monitoring of the Euclid space telescope¶

The Euclid survey was the first Stage IV survey to be commissioned, paving the way for a new era of precision cosmology. The survey's declared scientific goal is to explore the nature of dark matter and dark energy. One of the main probes it utilises is weak gravitational lensing: the coherent distortion of galaxy shapes caused by the large-scale structure of the Universe. Over its lifespan, Euclid is expected to measure the shape of 1.5 billion galaxies. However, these measurements are only as meaningful as their systematic uncertainties can be controlled. The main source of noise here is the Point Spread Function (PSF) that smears out the observed shapes. Euclid is among the first telescopes to employ a forward-modeled PSF. In spite of the challenges that arise due to the large field of view, the model is performing remarkably well after more than a decade of development. Nevertheless, due to the long-lived nature of the Euclid survey, the PSF model must be continuously updated. We have developed an algorithm to monitor the evolution of the PSF over time by precisely estimating the area enclosed by diffraction spikes of bright stars in an image. The spikes are caused by light bending around the mirror arms in the telescope entrance pupil and their position is highly correlated with the defocus of the space telescope. So far, the Euclid PSF has been remarkably stable. With our method, changes of the PSF over the field-of-view can be monitored almost in real-time, a crucial step for achieving the actual science goals of Euclid and possibly, by generalisation, other space telescopes.

Orateur: Dennis Neumann (Leiden Observatory)

16:10 Coffee

COLOURS Workshop: Session VIII¶

46 Impact of Weak Lensing Mass Mapping Algorithms on Cosmology Inference¶

Weak gravitational lensing provides a powerful probe of dark matter distribution in the Universe. Mass mapping algorithms, which reconstruct the convergence field from galaxy shear measurements, are critical for extracting higher-order statistics to constrain cosmological parameters. This study evaluates the impact of different mass mapping algorithms—Kaiser-Squires, inpainting Kaiser-Squires, and MCALens—on cosmological inference using weak lensing peak counts. Using simulated data from cosmo-SLICS, we compute peak counts and perform Bayesian analysis via MCMC to estimate matter density, amplitude of matter fluctuations, and dark energy equation of state. Results show MCALens significantly improves constraints on cosmological parameters by up to 157% compared to Kaiser-Squires, while inpainting provides limited gains. These findings highlight the importance of advanced mass mapping techniques, such as MCALens, in enhancing the precision of cosmological inference from weak lensing data.

Orateur: Andreas Tersenov (IA&ICS - FORTH)

47 Higher-Order Weak Lensing Statistics in Euclid: insights from pre-launch projects and DR1 plans¶

I will provide an overview of higher-order weak lensing statistics and their relevance to the Euclid mission. The talk will draw on the results of the pre-launch Paper 2, which employed the Fisher formalism to explore the constraining power of these non-Gaussian statistics on cosmological parameters. Emphasis will be placed on the key lessons learned from this preparatory project, highlighting how it informs the development of robust methodologies for the analysis of the Data Release 1. I will conclude by outlining the plans for the Data Release 1, with a particular focus on the strategies for probing models beyond ΛCDM, which is the core of my current research activities.

Orateur: Filippo Bouchè (Scuola Superiore Meridionale & INFN Sez. Napoli)

48 Weak lensing higher-order statistics in f(R) gravity¶

The problems within the standard cosmological model have sparked the research of alternatives and possible solutions and one way to tackle these problems is to question the validity of General Relativity (GR) as the correct theory of gravitation on cosmological scales, entering the context of modified gravity (MG) theories. In this work, we studied a specific MG model known as f(R) gravity, in the Hu & Sawicki formulation. Stage IV surveys studying the Large-scale Scale Structure (LSS), like the Dark Energy Spectroscopic Instrument (DESI) the Legacy Survey of Space and Time (LSST), or Euclid, will provide accurate weak lensing measurements to shed light on the most fundamental aspects of gravity and its effects on structures formation. Cosmological information on LSS is extracted primarily with two-point estimators. However, the clustering of matter due to gravitational collapse may introduce non-Gaussianities in matter distribution, for which standard statistics might not be optimal. For this purpose, we used a set of higher-order statistics measured on convergence maps, which are sensitive to non-Gaussian features in LSS, to discriminate between cosmological models and to assess their sensitivity to different behaviors of gravity during structure formation.

Orateur: Alessandro Vadalà (INAF - Astronomical Observatory of Rome)

49 Theoretical Predictions for Wavelet ℓ1-Norm in Weak Lensing¶

Orateur: Vilasini Tinnaneri Sreekanth (CosmoStat, CEA Paris-Saclay)

Social: Dinner¶

jeu. 12 juin ¶

Registration: Workshop¶

COLOURS Workshop: Session IX¶

50 Keynote - Cosmology from Radio Continuum Surveys¶

Surveys from modern radio telescopes, such as the precursor/pathfinder telescopes for the Square Kilometre Array Observatory (SKAO) are currently transforming the knowledge which we have of the radio skies. Telescopes such as the Australian Square Kilometre Array Pathfinder (ASKAP), Low Frequency Array (LOFAR) and MeerKAT are producing a number of radio continuum surveys which have a combination of depth and area which make them ideal for cosmological studies. In this talk I will outline some of the radio surveys being undertaken, discuss their challenges and discuss their benefits for use in radio cosmology studies. I will then present results from initial studies being conducted with current data releases available from the LOFAR surveys and ASKAP, before discussing the future surveys which will be possible from these telescopes and the SKAO. For these future surveys I will outline the key scientific questions which these radio surveys will address within the topic of cosmology.

Orateur: Catherine Hale (University of Oxford)

51 Probing Cosmic Dipole with Cosmological Survey¶

We leverages the first released catalog from the FAST All Sky \textsc{Hi} Survey (FASHI) to examine the hypothesis of cosmic isotropy in the local Universe. Given the design of the overall FAST survey, the inhomogeneous detection sensitivity of FASHI is likely to introduce significant biases in the statistical properties of the catalog. To mitigate the potential influence of spurious clustering effects due to these sensitivity variations, we focus on extragalactic \textsc{Hi} sources within the sensitivity range of $[0.65,1.0]$. This refined subsample is divided into ten distinct sky regions, for which we compute the two-point angular correlation functions (2PACF) over angular scales of ${0.5}^{\circ }<\theta <{10}^{\circ }$. We apply the Markov chain Monte Carlo method to fit these 2PACFs with a power-law model and assess the statistical significance of the best-fit parameters for the ten FASHI sky regions by comparing them against results from mock catalogs generated under the assumptions of homogeneity and isotropy. Our findings indicate that the local Universe, as traced by the \textsc{Hi} sources in the FASHI survey, aligns with the cosmic isotropy hypothesis within a $2\sigma$ confidence level. We do not detect any statistically significant deviations from cosmic isotropy in the FASHI survey data.

Orateur: Yi-Wen Wu (Beijing Normal University)

52 Primordial non-Gaussianity - the effects of relativistic and wide-angle corrections to the power spectrum¶

Wide-angle and relativistic corrections to the Newtonian and flat-sky approximations are important for accurate modelling of the galaxy power spectrum of next-generation galaxy surveys. In addition to Doppler and Sachs-Wolfe relativistic corrections, we include the effects of lensing convergence, time delay and integrated Sachs-Wolfe. We investigate the impact of these corrections on measurements of the local primordial non-Gaussianity parameter $f_{\mathrm{N}\mathrm{L}}$, using two futuristic spectroscopic galaxy surveys, planned for SKAO2 and MegaMapper. In addition to the monopole, we include the quadrupole of the galaxy Fourier power spectrum. The quadrupole is much more sensitive to the corrections than the monopole. The combination with the quadrupole improves the precision on $f_{\mathrm{N}\mathrm{L}}$ by $\sim 40\mathrm{\%}$ and $\sim 60\mathrm{\%}$ for SKAO2 and MegaMapper respectively. Neglecting the wide-angle and relativistic corrections produces a shift in $f_{\mathrm{N}\mathrm{L}}$ of $\sim 0.1\sigma$ and $\sim 0.2\sigma$ for SKAO2 and MegaMapper. The shift in $f_{\mathrm{N}\mathrm{L}}$ is very sensitive to the magnification bias and the redshift evolution of the comoving number density. For these surveys, the contributions to the shift from integrated and non-integrated effects partly cancel. We point out that some of the approximations made in the corrections may artificially suppress the shift in $f_{\mathrm{N}\mathrm{L}}$.

Orateur: Sêcloka Lazare Guedezounme (University of the Western Cape)

11:00 Coffee

COLOURS Workshop: Session X¶

53 Keynote - Mapping Neutral Hydrogen: opportunities and challenges on the road to the SKA Observatory¶

Intensity Mapping techniques for the 21cm line. In the post-reionization era (z < 6), this will allow us to trace the underlying large-scale distribution of matter, opening a new window on dark matter and dark energy. To unlock the full potential of this technique for the SKAO, an intense data-taking campaign has been carried out using the SKAO precursors. I will describe the status of MeerKLASS, the Intensity Mapping survey with the MeerKAT array, and discuss the various challenges concerning flagging and treatment of Radio Frequency Interference, calibration, separation of the bright foreground emission leading to the detection of cross-correlation signal with the galaxy survey, and the first hints of an auto-correlation power spectrum measurement. These results pave the way to routine use of IM also in the SKAO era, potentially allowing radio cosmology to deliver complementary and competitive constraints with galaxy surveys on the cosmological model.

Orateur: Marta Spinelli (Observatoire de la Cote d'Azur)

54 Detection of the power spectrum turnover with SKA & some galaxy surveys¶

Understanding the universe at ultra-large scales offers a unique window into its origins and evolution, particularly the transition from radiation to matter domination. The power spectrum turnover, a subtle yet critical feature of large-scale structure, holds the key to unraveling these mysteries. In this study, we explore the potential of next-generation surveys, including the Square Kilometre Array (SKA) and the Dark Energy Spectroscopic Instrument (DESI), to detect and constrain this elusive signature. By combining 21cm intensity mapping and galaxy surveys, we employ advanced methodologies such as Fisher forecasts and Markov Chain Monte Carlo techniques to analyze this cosmological feature. Our results highlight the exceptional capabilities of SKA-MID Band 1, which emerges as the most promising instrument for achieving high detection significance and robust parameter constraints. We also address key challenges, including foreground contamination and telescope beam effects, and propose strategies to mitigate their impact. This work not only demonstrates the feasibility of detecting the turnover but also lays the groundwork for future studies probing ultra-large scales, offering new insights into the universe's large-scale structure and its governing physical laws.

Orateur: Yolanda Dube (University of the Western Cape)

55 Latest perspectives on weighing neutrinos with 21cm intensity mapping at the SKAO¶

The large-scale distribution of neutral hydrogen in the late Universe, mapped through the hydrogen 21cm line emission using radio telescopes, holds significant potential to emerge as a key cosmological probe in the coming years. In the work that I’m going to present, we developed a Gaussian likelihood code for the 21cm intensity mapping power spectrum and the 21cm-galaxy clustering cross-correlation power spectrum. We applied our analysis pipeline to forecast the constraining power of future 21cm intensity mapping observations at the SKA Observatory. I will present the latest forecasts of the constraining power of the SKAO on the sum of neutrino masses. We find that adding 21cm intensity mapping power spectrum multipoles data, assuming a fiducial value of the sum of neutrino masses of 0.06 eV, significantly enhances constraints from the Planck 2018 observations of <0.241 eV to <0.101 eV in the most ideal scenario. Additionally, the code that we developed is constructed to be used with data from current experiments and I will present some preliminary results in this context.

Orateur: Gabriele Autieri (SISSA, INFN Trieste, IFPU)

12:50 Lunch

Canteen

COLOURS Workshop: Session XI¶

56 Keynote - SPT-3G¶

Orateur: Silvia Galli (IAP)

57 Constraining baryonic feedback with kinetic Sunayev-Zeldovich effect¶

The distribution of gas relative to the underlying matter is a major unknown in current cosmology, posing strong challenges to the next generation of large scale structure surveys, particularly weak lensing studies. With the advent of recent high-quality observations of the kinematic Sunyaev-Zeldovich (kSZ) effect, we can directly measure the gas distribution in the outskirts of haloes for the first time. These measurements, proportional to the momentum field of gas, suggest that the gas is much more extended than dark matter, pointing to strong feedback from supermassive black holes. In this talk, I will show how several baryonic processes shape the density and velocity fields of gas and propose a flexible simulation based model for the kSZ signal. I will then analyse the stacked kSZ signal around three galaxy samples, namely DESI LRGs, CMASS and LOWZ galaxies. The kSZ data prefers strong baryonic feedback, implying very low gas fractions in group sized haloes.

Orateur: Lurdes Ondaro Mallea (Donostia International Physics Center)

58 The impact of dark energy on the thermal Sunyaev-Zel'dovich effect¶

The distribution of matter in the Universe is a powerful probe of cosmology. Measuring the efficiency with which gravity produces clusters against expanding Universe is the key to understanding, e.g. the equation of state of dark energy. The thermal Sunyaev-Zel’dovich (tSZ) effect, being sensitive to the baryonic content of the Universe, also shows sensitivity to dark energy. In this work, I investigate the impact of dark energy on the tSZ effect. To this end, I conducted new simulations varying the equation of state of dark energy within RAMSES, and compared different quantities such as the matter and tSZ power spectrum. In this talk, I will present and discuss my results on these comparisons across different cosmologies.

Orateur: Emma Ayçoberry (CEA / DAp)

COLOURS Poster Session¶

59 SuperBIT: Weak gravitational lensing from a balloon platform¶

SuperBIT (The Super Pressure Balloon-borne Imaging Telescope) is a diffraction-limited 0.5-meter UV-NIR telescope designed to conduct high-resolution astronomical observations from the stratosphere. Its primary science goal is to map the distribution of dark matter in galaxy clusters through weak gravitational lensing. Launched on a NASA super-pressure balloon, SuperBIT offers a unique combination of high spatial resolution, a wide field of view, and long observation durations. This poster presents an overview of the SuperBIT instrument, preliminary results of the 2023 flight and the next generation GigaBIT. Beyond its primary goal, the telescope can perform high-resolution follow-up observations for objects from other surveys and transient events like those of gravitational wave detections, providing valuable data for deeper characterization and crucial context for understanding the nature and origin of these cosmic events.

Orateur: Vy Luu (Princeton University)

60 Galaxy Bias by cross-correlating LOFAR with CMB¶

The second data release of LOFAR Two meter sky survey covers ~5.600 deg^2 and more than 4 million sources. However, not all radio sources have redshift information. Using the optical counterparts to the radio sources(from Legacy Survey) with photometric redshifts, we aim to constrain galaxy bias by cross-correlating galaxy over-density with CMB lensing map from Planck. Previous works have lacked the redshift information and coverage over large area of sky simultaneously. We aim to study evolution of galaxy bias with redshift. Current results indicate an evolving bias, however an improved understanding of selection bias is required. Further work is to measure the bias for different populations of galaxy, such as AGNs versus SFG, selected using the LOFAR radio information.

Orateur: Shree Hari Mittal (University of Hertfordshire)

61 ARGOSIM¶

Orateur: Ezequiel Centofanti (CosmoStat, CEA Paris-Saclay)

62 Galaxy Assembly Bias in cosmological hydrodynamical simulations - A comparison between SIMBA and IllustrisTNG¶

63 How to measure cosmology from nothing: cosmic voids data analysis¶

The data provided by the state of the art surveys, allows us to study cosmic voids in detail. By measuring the void-galaxy cross correlation function we can to extract constraints on the content of the Universe and the growth rate of structures through the Alcock Paczynski and by modelling red-shift space distortions. In this poster I describe the methodology and status of the DESI void analysis, and its future implications.

Orateur: Katayoon Ghaemiardakani (CPPM)

64 The multibeam coincidence matching (MBCM) blind search mode and polarization strategy for SETI on FAST in China¶

The search for extraterrestrial intelligence (SETI) is to search for technosignatures associated with extraterrestrial life, such as engineered radio signals. In this paper, we apply the multibeam coincidence matching (MBCM) strategy (see for more details: https://iopscience.iop.org/article/10.3847/1538-3881/ac8bd5, AJ, 164 (2022) 4, 160) and propose a new search mode based on the MBCM which we call the MBCM blind search mode (see for more details: https://iopscience.iop.org/article/10.3847/1538-3881/acb706, AJ, 165 (2023) 3, 132). In our targeted SETI research, 33 exoplanet systems are observed by the Five-hundred-meter Aperture Spherical radio Telescope (FAST). With this blind search mode, we search for narrowband drifting signals across 1.05–1.45 GHz in two orthogonal linear polarization directions separately. There are two special signals, one of which can only be detected by the blind search mode while the other can be found by both blind and targeted search modes. This result reveals huge advantages of the new blind search mode. However, we eliminate the possibility of the special signals being ETI signals based on much evidence, such as the polarization, drift, frequency, and beam coverage characteristics. Our observations achieve an unprecedented sensitivity because the minimum equivalent isotropic radiated power we are able to detect reaches 1.48 × 109 W. In addition, our work provides a deeper understanding of the polarization strategy for analyzing extraterrestrial signals. The polarization strategy involves, on the one hand, analyzing the intensity of the polarization channels of the signals, and on the other hand, studying the variation of the intensity over time, so as to determine the origin of the signals (see for more details: https://iopscience.iop.org/article/10.3847/1538-3881/ad0be8, AJ, 167 (2023) 1, 8). As the most sensitive single-dish radio telescope in the L-band, FAST will conduct more SETI observations in the future, including targeted searches and commensal sky surveys. Although SETI with FAST is challenged by more types of weak RFI, we will identify and classify more types of RFI and discover new search methods to remove RFIs effectively.

Orateur: Xiao-Hang Luan (School of Physics and Astronomy, Beijing Normal University)

16:10 Coffee

COLOURS Workshop: Session XII¶

65 Pannel discussion with Keynote speakers - What did we learn in Stage III? What do we hope to learn in Stage IV? What do we expect for the future?¶

ven. 13 juin ¶

Registration: Workshop¶

COLOURS Workshop: Session XIII¶

66 Keynote - DES¶

Orateur: Daniel Gruen (LMU Munich)

67 Dark Energy Survey Year 6: Magnification modeling and impact on galaxy clustering and weak lensing cosmology¶

Gravitational lensing magnification affects the observed spatial distribution of galaxies and must be accounted for to avoid biases in cosmological probes of large-scale structure. I will present the methodology used to evaluate its impact on the Dark Energy Survey Year 6 (DES Y6) galaxy clustering and galaxy-galaxy lensing analyses. We quantify the redshift-dependent magnification bias using the Balrog synthetic source injection catalog, accounting for the complexity of the galaxy sample selection function. We investigate the impact of magnification on cosmological constraints by testing different priors on magnification bias parameters. Our results show that magnification is a significant systematic for the Y6 fiducial lens sample (MagLim++) and must be addressed. Additionally, we demonstrate that including clustering cross-correlations between lens bins in the analysis enhances constraints on magnification parameters, enabling the use of uninformative priors. While magnification has minimal effect on the precision of cosmological constraints, neglecting it introduces parameter biases that can exceed statistical uncertainties, making its correction critical for current and future weak lensing surveys.

Orateur: Elisa Legnani (IFAE Barcelona)

68 Cosmic and thermodynamic analysis of parameterized torsion gravity¶

We examine the cosmic evolution of an accelerating universe within the framework of the Einstein–Cartan–Sciama–Kibble theory, employing a flat, homogeneous and isotropic model. In this context, various parameterizations of the equation of state parameter, denoted as w (omega), are considered with respect to the redshift function. We analyze the behavior of cosmological parameters such as deceleration, Om-diagnostic, statefinder and squared speed of sound. It is found that these cosmological parameters are consistent with the current cosmic scenario and recent Planck data from 2018. Additionally, we examine the validity of the generalized second law of thermodynamics at the apparent horizon, taking into account the Bekenstein–Hawking entropy and find it valid for specific choices of constants.

Orateur: NAGEEN PERVAIZ (NATIONAL ASTRONOMICAL OBSERVATORIES OF THE CHINESE ACADEMY OF SCIENCES)

11:00 Coffee

COLOURS Workshop: Session XIV¶

69 The future of cosmological inference¶

In this talk I will introduce a new paradigm for cosmological inference, enabled by recent advances in machine learning and its underlying technology. By combining emulation, differentiable and probabilistic programming, scalable gradient-based sampling, and decoupled Bayesian model selection, this framework scales to extremely high-dimensional parameter spaces and enables complete Bayesian analyses—encompassing both parameter estimation and model selection—in a fraction of the time required by conventional approaches. I will demonstrate its application to various Stage IV cosmological survey configurations, tackling parameter spaces of approximately 150 dimensions that are inaccessible to standard techniques. I will also show how this framework can be used to test competing gravity theories and present Stage IV forecasts on interacting dark energy models. Finally, I will illustrate how a simulation-based inference analysis of Euclid cosmic shear data could definitively confirm or refute the recent DESI results pointing to dynamical dark energy.

Orateur: Alessio Spurio Mancini (Royal Holloway - University of London)

70 Forward modelling UNIONS survey for Implicit Likelihood Inference¶

Orateur: Sacha Guerrini (CosmoStat, CEA Paris-Saclay)

71 Machine learning in cosmology: Using emulators to accelerate inference for Stage-IV surveys¶

In my talk, I will talk about the development of a large emulator library, designed to streamline the exploration and inference of different cosmological models being used in the Stage-IV survey analysis. This library serves as a bridge between different theoretical predictions and observational data, offering fast, accurate interpolations of key cosmological quantities. Built upon a suite of precomputed datasets, spanning diverse parameter spaces and models, the emulator enables efficient likelihood evaluations and parameter inference for large-scale surveys. We will discuss the methodology behind constructing the emulators, including the choice of cosmological models, sampling strategies, and validation techniques.

Orateur: Ivan Sladoljev (Royal Holloway)

72 Forward-modelling Stage IV galaxy surveys for precise cosmological measurements¶

Stage IV experiments are set to provide the stringent constraints ever on cosmological parameters. However, to achieve this aim, we need careful control of the systematics, especially those related to the accurate knowledge of the tomographic redshift distributions. The forward-modelling of photometric and spectroscopic galaxy surveys, a method that bridges cosmology with galaxy evolution, arose as one of the most promising approaches to solve the problem of accurate galaxy redshift distribution estimates. In this talk I will discuss the past (Tortorelli+18,20,21) and on-going efforts (Tortorelli+24) in forward-modelling galaxy surveys, from the modelling of the galaxy population to the simulation of images and spectra using simulators I developed. I'll show that forward-modelling already provides precise redshift distribution estimates and that this method can reach the precision required by Stage IV surveys using a new galaxy population model based on stellar population synthesis that I developed (GalSBI), from which I am able to draw intrinsic distributions of physical galaxy quantities, such as stellar masses, SFHs, metallicities, gas, dust and AGN properties.

Orateur: Luca Tortorelli (LMU Munich)

73 Closing¶

Orateur: Samuel Farrens (CosmoStat - CEA Paris-Saclay)

12:50 Lunch

Canteen