A5021502333- Galaxies: Formation, Evolution, Phenomena
- Astronomy and Astrophysical Research
- Gaussian Processes and Bayesian Inference
- Adaptive optics and wavefront sensing
- Gamma-ray bursts and supernovae
- Cosmology and Gravitation Theories
- Computational Physics and Python Applications
- Astronomical Observations and Instrumentation
- Scientific Research and Discoveries
- Generative Adversarial Networks and Image Synthesis
- Image Processing and 3D Reconstruction
- Advanced Image Processing Techniques
- Stellar, planetary, and galactic studies
- Scientific Computing and Data Management
- Statistical and numerical algorithms
- Radio Astronomy Observations and Technology
- Anomaly Detection Techniques and Applications
- Advanced Vision and Imaging
- Cell Image Analysis Techniques
- Image and Signal Denoising Methods
- Sparse and Compressive Sensing Techniques
- Machine Learning and Data Classification
- Advanced Data Storage Technologies
- Advanced X-ray Imaging Techniques
- Advanced MRI Techniques and Applications
Université Paris-Saclay
2020-2024
Université Paris Cité
2012-2024
Astrophysique, Instrumentation et Modélisation
2012-2024
Centre National de la Recherche Scientifique
2012-2024
CEA Paris-Saclay
2012-2024
Commissariat à l'Énergie Atomique et aux Énergies Alternatives
2012-2024
Délégation Paris 7
2020-2024
Sorbonne Paris Cité
2020-2024
Flatiron Health (United States)
2023-2024
Flatiron Institute
2023-2024
Abstract We measure cosmic weak lensing shear power spectra with the Subaru Hyper Suprime-Cam (HSC) survey first-year catalog covering 137 deg2 of sky. Thanks to high effective galaxy number density ∼17 arcmin−2, even after conservative cuts such as a magnitude cut i < 24.5 and photometric redshift 0.3 ≤ z 1.5, we obtain high-significance measurement in four tomographic bins, achieving total signal-to-noise ratio 16 multipole range 300 ℓ 1900. carefully account for various...
The amount and complexity of data delivered by modern galaxy surveys has been steadily increasing over the past years. Extracting coherent scientific information from these large multi-modal sets remains an open issue driven approaches such as deep learning have rapidly emerged a potentially powerful solution to some long lasting challenges. This enthusiasm is reflected in unprecedented exponential growth publications using neural networks. Half decade after first published work astronomy...
Galaxy-scale strong gravitational lensing is not only a valuable probe of the dark matter distribution massive galaxies, but can also provide cosmological constraints, either by studying population lenses or measuring time delays in lensed quasars. Due to rarity galaxy-scale strongly systems, fast and reliable automated lens finding methods will be essential era large surveys such as LSST, Euclid, WFIRST. To tackle this challenge, we introduce CMU DeepLens, new fully galaxy-galaxy method...
We present results from a set of simulations designed to constrain the weak lensing shear calibration for Hyper Suprime-Cam (HSC) survey. These include HSC observing conditions and galaxy images Hubble Space Telescope (HST), with fully realistic morphologies impact nearby galaxies included. find that inclusion in is critical reproducing observed distributions sizes magnitudes, due non-negligible fraction unrecognized blends ground-based data, even excellent typical seeing survey (0.58"...
Large scale imaging surveys will increase the number of galaxy-scale strong lensing candidates by maybe three orders magnitudes beyond known today. Finding these rare objects require picking them out at least tens millions images and deriving scientific results from quantifying efficiency bias any search method. To achieve objectives automated methods must be developed. Because gravitational lenses are reducing false positives particularly important. We present a description an open lens...
Abstract We present and characterize the catalog of galaxy shape measurements that will be used for cosmological weak lensing in Wide layer first year Hyper Suprime-Cam (HSC) survey. The covers an area 136.9 deg2 split into six fields, with a mean i-band seeing 0${^{\prime\prime}_{.}}$58 5σ point-source depth i ∼ 26. Given conservative selection criteria first-year science, excellent image quality results unweighted weighted source number densities 24.6 21.8 arcmin−2, respectively. define...
Abstract This paper introduces cosmoDC2, a large synthetic galaxy catalog designed to support precision dark energy science with the Large Synoptic Survey Telescope (LSST). CosmoDC2 is starting point for second data challenge (DC2) carried out by LSST Dark Energy Science Collaboration (LSST DESC). The based on trillion-particle, (4.225 Gpc) 3 box cosmological N -body simulation, Outer Rim run. It covers 440 deg 2 of sky area redshift z = and matches expected number densities from...
In many cosmological inference problems, the likelihood (the probability of observed data as a function unknown parameters) is or intractable. This necessitates approximations and assumptions, which can lead to incorrect parameters, including nature dark matter energy, create artificial model tensions. Likelihood-free covers novel family methods rigorously estimate posterior distributions parameters using forward modelling mock data. We present likelihood-free parameter weak lensing maps...
ABSTRACT We present reconstructed convergence maps, mass from the Dark Energy Survey (DES) third year (Y3) weak gravitational lensing data set. The maps are weighted projections of density field (primarily dark matter) in foreground observed galaxies. use four reconstruction methods, each is a maximum posteriori estimate with different model for prior probability map: Kaiser–Squires, null B-mode prior, Gaussian and sparsity prior. All methods implemented on celestial sphere to accommodate...
Abstract We describe the simulated sky survey underlying second data challenge (DC2) carried out in preparation for analysis of Vera C. Rubin Observatory Legacy Survey Space and Time (LSST) by LSST Dark Energy Science Collaboration (LSST DESC). Significant connections across multiple science domains will be a hallmark LSST; DC2 program represents unique modeling effort that stresses this interconnectivity way has not been attempted before. This encompasses full end-to-end approach: starting...
We present jax-cosmo, a library for automatically differentiable cosmological theory calculations. It uses the JAX library, which has created new coding ecosystem, especially in probabilistic programming. As well as batch acceleration, just-in-time compilation, and automatic optimization of code different hardware modalities (CPU, GPU, TPU), exposes an differentiation (autodiff) mechanism. Thanks to autodiff, jax-cosmo gives access derivatives likelihoods with respect any their parameters,...
Abstract Rapid advances in deep learning have brought not only a myriad of powerful neural networks, but also breakthroughs that benefit established scientific research. In particular, automatic differentiation (AD) tools and computational accelerators like GPUs facilitated forward modeling the Universe with differentiable simulations. Based on analytic or backpropagation, current cosmological simulations are limited by memory, thus subject to trade-off between time space/mass resolution,...
The Core Cosmology Library (CCL) provides routines to compute basic cosmological observables a high degree of accuracy, which have been verified with an extensive suite validation tests. Predictions are provided for many quantities, including distances, angular power spectra, correlation functions, halo bias and the mass function through state-of-the-art modeling prescriptions available in literature. Fiducial specifications expected galaxy distributions Large Synoptic Survey Telescope...
ABSTRACT Hydrodynamical simulations of galaxy formation and evolution attempt to fully model the physics that shapes galaxies. The agreement between morphology simulated real galaxies, way morphological types are distributed across scaling relations important probes our knowledge physics. Here, we propose an unsupervised deep learning approach perform a stringent test fine structure galaxies coming from Illustris IllustrisTNG (TNG100 TNG50) against observations subsample Sloan Digital Sky...
ABSTRACT We present AstroCLIP, a single, versatile model that can embed both galaxy images and spectra into shared, physically meaningful latent space. These embeddings then be used – without any fine-tuning for variety of downstream tasks including (1) accurate in-modality cross-modality semantic similarity search, (2) photometric redshift estimation, (3) property estimation from spectra, (4) morphology classification. Our approach to implementing AstroCLIP consists two parts. First, we...
The intracluster light (ICL) is an important tracer of a galaxy cluster's history and past interactions. However, only small samples have been studied to date due its very low surface brightness the heavy manual involvement required for majority measurement algorithms. Upcoming large imaging surveys such as Vera C. Rubin Observatory's Legacy Survey Space Time are expected vastly expand available deep cluster images. process this increased amount data, we need faster, fully automated methods...
Abstract The intracluster light (ICL) is an important tracer of a galaxy cluster’s history and past interactions. However, only small samples have been studied to date due its very low surface brightness the heavy manual involvement required for majority measurement algorithms. Upcoming large imaging surveys such as Vera C. Rubin Observatory’s Legacy Survey Space Time (LSST) are expected vastly expand available deep cluster images. process this increased amount data, we need faster, fully...
We present the first reconstruction of dark matter maps from weak lensing observational data using deep learning. train a convolution neural network (CNN) with Unet based architecture on over $3.6\times10^5$ simulated realizations non-Gaussian shape noise and cosmological parameters varying broad prior distribution. interpret our newly created DES SV map as an approximation posterior mean $P(\kappa | \gamma)$ convergence given observed shear. Our DeepMass method is substantially more...
In recent years, machine learning (ML) methods have remarkably improved how cosmologists can interpret data. The next decade will bring new opportunities for data-driven cosmological discovery, but also present challenges adopting ML methodologies and understanding the results. could transform our field, this transformation require astronomy community to both foster promote interdisciplinary research endeavors.
Image simulations are essential tools for preparing and validating the analysis of current future wide-field optical surveys. However, galaxy models used as basis these typically limited to simple parametric light profiles, or use a fairly amount available space-based data. In this work, we propose methodology based on Deep Generative Models create complex morphologies that may meet image simulation needs upcoming We address technical challenges associated with learning morphology model from...
Traditionally, weak lensing cosmological surveys have been analyzed using summary statistics that were either motivated by their analytically tractable likelihoods (e.g., power spectrum) or ability to access some higher-order information peak counts), but at the cost of requiring a simulation-based inference approach. In both cases, even if can be very informative, they are not designed nor guaranteed statistically sufficient (i.e., capture all content data). With rise deep learning,...
Understanding the nature of dark energy, mysterious force driving accelerated expansion Universe, is a major challenge modern cosmology. The next generation cosmological surveys, specifically designed to address this issue, rely on accurate measurements apparent shapes distant galaxies. However, shape measurement methods suffer from various unavoidable biases and therefore will precise calibration meet accuracy requirements science analysis. This process remains an open as it requires large...
Mapping the underlying density field, including non-visible dark matter, using weak gravitational lensing measurements is now a standard tool in cosmology. Due to its importance science results of current and upcoming surveys, quality convergence reconstruction methods should be well understood. We compare three methods: Kaiser-Squires (KS), Wiener filter, GLIMPSE. KS direct inversion, not accounting for survey masks or noise. The filter well-motivated Gaussian fields Bayesian framework....
The problem of anomaly detection in astronomical surveys is becoming increasingly important as data sets grow size. We present the results an unsupervised method using a Wasserstein generative adversarial network (WGAN) on nearly one million optical galaxy images Hyper Suprime-Cam (HSC) survey. WGAN learns to generate realistic HSC-like galaxies that follow distribution set; anomalous are defined based poor reconstruction by generator and outlying features learned discriminator. find...