Institute of research into the fundamental laws of the Universe

The upcoming Euclid ESA space telescope will aim to shed some much needed light on the physical nature of dark energy and dark matter by imaging the sky with a very high resolution. One major hurdle in the Euclid cientific exploitation of the images is that an exquisite control of the telescope instrumental response is required in order to precisely measure the shape of distant galaxies. With current methods, this delicate image processing task is prone to biases and errors that may hinder the scientific goals of this space mission.



The goal of this PhD thesis is to develop a hierarchical probabilistic model of the observed Euclid images combining physical models with Deep Learning components accounting for unknowns factors. Thanks to this hybrid Physical/Deep Learning generative model of the observed images,

it will be possible for the first time to jointly infer from the data both a Bayesian posterior distribution of the physical parameters and the parameters of the deep learning models.

The Euclid satellite, to be launched in 2022, will observe the sky in the optical and infrared, and will be able to map large scale structures and weak lensing distortions out to high redshifts. Weak gravitational lensing is thought to be one of the most promising tools of cosmology to constrain models. Weak lensing probes the evolution of dark-matter structures and can help distinguish between dark energy and models of modified gravity. Thanks to the shear measurements, we will be able to reconstruct a dark matter mass map of 15000 square degrees. Mass mapping entails the construction of two-dimensional maps using galaxy shape measurements, which represent the integrated total matter density along the line of sight. Small- field mass maps have been frequently used to study the structure and mass distribution of galaxy clusters, whereas wide-field maps have only more recently become possible given the broad observing strategies of surveys like CFHTLenS, HSC, DES, and KiDS. Mass maps contain significant non-Gaussian cosmological information and can be used to identify massive clusters as well as to cross-correlate the lensing signal with foreground structures.

A standard method to derive mass maps from weak-lensing observations is an inversion technique formulated by Kaiser & Squires [2]. It has many limitations, however, including the need to smooth the data before (and often after) inversion, thereby losing small-scale information. An alternative method called GLIMPSE has been developed in the CosmoStat laboratory based on sparse reconstruction that avoids this problem and improves the recovery of non-Gaussian features [3, 4]. The algorithm has been tested on simulations and was also recently used to study the A520 merging galaxy cluster with Hubble Space Telescope data [5]. More recently, machine learning has emerged as a promising technique for mass map recovery [6].

The goal of this thesis is to i) compare this technique to the state of the art and investigate if it can be used in practice, ii) extend the method for spherical data, and iii) develop a new machine learning approach to estimate the cosmological parameters. At the core of this new statistical framework will be the development of fast and differentiable cosmological simulations capable of emulating the Euclid survey under various cosmologies. This simulation tool will be based on the FastPM N-body simulation code [7] and implemented directly in the TensorFlow machine learning framework, yielding a differentiable physical forward simulation pipeline which can be directly interfaced with deep learning components or with inference techniques relying on having access to the derivatives of the simulation.

As part of the CosmoStat Laboratory, located at CEA Saclay, the successful candidate will be embedded in a leading French research group, heavily involved in the preparation of the Euclid space mission, and with a long tradition of developing cutting-edge statistical tools for the analysis of astronomical and cosmological data.

1. Bartelmann, M. & Schneider, P. 2001, Phys. Rep., 340, 291. ?

2. Kaiser, N. & Squires, G. 1993, ApJ, 404, 441. ?

3. Leonard, A., Lanusse, F., & Starck, J.-L. 2014, MNRAS, 440, 1281.

4. Lanusse, F., Starck, J.-L., Leonard, A., & Pires, S. 2016, A&A, 591, A2.

?5. Peel, A., Lanusse, F., & Starck, J.-L. 2017, ApJ, 847, 23.

6. Niall Jeffrey et al, submitted. https://arxiv.org/abs/1908.00543

7. Y. Feng, M. Yat Chu, U. Seljak, and P. McDonald. MNRAS, 463(3):2273–2286, 2016.

The formation of stars is a fundamental aspect of the evolution of the Universe. Strangely it is still poorly understood as it results from an intricate combination of complex physical processes : several instabilities (dynamical, chemical, thermal), magneto-hydrodynamic turbulence, gravity and energy injection by stars themselves. Because of this complex, multi-scale and multi-phase physics, this problem is now study using numerical simulations of increasing sophistication. Important progress has been made to a point that we are now in a situation where there is a lack of constraints from observations to identify the relevant physical scenarios. This is in part due to the difficulty to find the right metric to compare numerical simulations and observations. In this context, machine learning tools open new ways of exploration. In particular they allow the automatic definition of new comparison and evaluation metrics making possible the direct comparison of observations and simulations.

The subject of this Ph.D. thesis is to define a framework that would allow to estimate physical parameters (magnetic field intensity, temperature distribution, density distribution, power spectra of density and velocity) of different regions of the interstellar medium by applying machine learning tools on hyper-spectral observations (21 cm and CO). The setup of the tools would be first done by a training on a set of numerical simulations that are already available. This project is made possible thanks to a combination of expertise present in the Hyperstars collaboration, a joint effort by experts of the star formation process (Marc-Antoine Miville-Deschênes - hyper-spectral data and Patrick Hennebelle - numerical simulations) and experts of several aspects of data science.

One of the major challenges of astrophysics is to understand how galaxies assemble their mass, give birth to their supermassive stars and black holes over time, and how this assembly depends on their internal evolution and/or external factors such as dark matter haloes and galactic fusions. To date, our understanding of the cosmic history of star formation remains largely incomplete over the key period of the 3 billion years following reionization, i.e. between z~6 - when galaxies had formed less than 1% of their current stars - until the time of the star formation peak around z~1.5. ?During this thesis, the student will benefit from a unique set of data - mainly from the ALMA interferometer and the James Webb Space Telescope (JWST) - that will allow him to quantify this period in the history of star formation activity, mass growth and the morphological evolution of galaxies. Overcoming the uncertainty about this era will therefore have a major impact on our understanding of the formation of structures in the universe, and even our cosmology, and this is one of the major objectives of our JWST program.

The discovery, by the LIGO-Virgo collaboration on Sept. 14th 2015, of gravitational waves (GW) from the merger of two stellar-mass black holes, applauded by the whole scientific community, was unexpected in terms of astrophysical sources: two such heavy stellar-mass black holes (~30 solar masses) had never been seen before, although they likely constitute the tip of the iceberg. From this detection, several questions immediately arose: how can such black holes form, and how many are there in our local Universe and beyond? The second breakthrough came with the detection of a kilonova associated with the merger of two neutron stars, on Aug. 17th 2017. Further questions arose, such as the nature of the outcome of such a merger. More generally, one of the most fundamental questions in terms both of astrophysics and physics, concerns the nature of the progenitors for this type of system. Finally, we now know that many such mergers will be detected by current and future GW observatories, but we do not know the exact rate.



Stellar binaries hosting compact objects (especially neutron stars and black holes) constitute the best progenitors, evolving until eventually merging in binary black holes (BBH), binary neutron stars (BNS) or black hole and neutron star binaries (BH/NS), and emitting GW. The overall evolution of such binaries is still subject to many uncertainties about some parameters of binary evolution, such as: the natal kick received during each supernova event, metallicity effect on stellar wind, common envelope phase, which condition the survival of the stellar binaries, the spin of each component, etc…

Stellar couples are very common in our Galaxy: more than 70% of massive stars live as a couple during their stellar life. This PhD-Thesis aims at studying how these systems form, evolve and have an impact on their environment.



Massive stars live in couples...

Several revolutions have occurred in recent years in the stellar domain. The first is the realization that most (over 70%) massive stars live within a stellar pair (Sana et al., 2012). This binarity has major consequences on the evolution of stars, strongly influenced by the presence of a "companion", particularly via the transfer of matter and kinetic momentum (Chaty 2013). The fate of these stellar pairs is determined by the evolution of each component, with the most massive star collapsing first during the supernova explosion, giving rise to a neutron star or a black hole (Tauris et al. 2017). A stellar couple, composed of a compact star orbiting its companion, is among the most fascinating celestial objects of our Universe. The companion star, massive, is characterized by an ejection of wind more or less intense according to its metallicity, and the compact star, bathed in this wind, attracts a part of this matter, which, accreted, accumulates to the surface, heated to temperatures of several million degrees, emitting mainly in the field of X-rays. These stars regularly give rise to extreme variations in luminosity, several orders of magnitude over the entire electromagnetic spectrum, on scales time from the second to the month.



... until they merge ...

The second revolution is the detection, by interferometers of the LIGO / Virgo collaboration, of gravitational waves coming from the fusion of two black holes (first detection in September 2015) and two neutron stars (August 2017). This fusion occurs at the end of the life of certain stellar pairs, depending on their mass, their orbital separation, and several other parameters involved in their evolution. The fusion of neutron stars is accompanied by an emission of electromagnetic waves, called kilonova, and spectroscopic observations have shown that heavy atoms were created during this event, via the "fast process" of nucleosynthesis (r-process).



... with an impact on their environment!

It is now established that the collapse of massive supernova stars plays a key role in the enrichment of the interstellar medium - from heavy atoms to complex molecules - and in triggering the formation of new stars. On the other hand, the impact of the wind of these massive stars on their environment, throughout their life, was long neglected. However, this ejected material disperses in the surrounding environment, until it collides with a dense interstellar medium, potentially triggering new star formations, as suggested by observations from the Herschel satellite (Chaty et al. 2012). Finally, the recent observations of r-process concomitant with the detection of a kilonova show that the fusion of two neutron stars is an important (or even majority) element of nucleosynthesis in the galaxy.



This PhD-thesis, covering various fields of astrophysics, proposes to study how these formidable couples of massive stars form, whose role is primordial in the cycle of matter, how they evolve, and what is their impact on their environment, based on multi-wavelength observations (ESO, Gaia...).

X-ray data are multidimensional by nature. For each photon the energy and position is recorded by the X-ray satellite. Here we propose to develop novel techniques to fully exploit the multidimensional nature of the data by combining blind source separation technique with feature learning.

JWST will be launched in 2021, with an important participation from ESA, CNES and CEA for the mid-infrared instrument MIRI. JWST capabilities are revolutionary, compared to the existing state of the art, in terms of resolution and sensitivity over the 1- 30?m wavelength range, where stars and (warm) dust emit their light from galaxies at high redshift. For the first time JWST will provide spatially resolved photometry up to the mid-IR (at least 10?m, with NIRCAM and MIRI) with sub-arcsec resolution. The competitive exploitation of the data for scientific endeavors will require the mastering of the data, deeply understanding the reduction, treatment and developing tools to foster the analysis. I propose a PhD thesis in Saclay as a collaborative effort between experts from ’MICE, the Centre of Expertise for MIRI’, developed at CEA/Irfu/DAp, and with researchers in galaxy formation and evolution. The student will be responsible for developing new high level software for the analysis of resolved imaging data from MIRI and NIRCAM, modeling and understand the resolution, ’pixelization’ and PSF convolution effects. This will include high-level software to create spatially resolved maps of physical parameters (stellar mass, dust attenuation, stellar age, star formation rate) and pixel-by-pixel spectral energy distributions. The student will work on testing and improving the existing MIRI simulator, adapting it to the case of resolved observations of distant galaxies. The results of the efforts will be shared with several of the CEA Saclay groups in the spirit of fostering our expertise and efficiency in the early use of the groundbreaking JWST data. This work will be based on data from our recently approved Early Release Science (ERS) project observing with a suite of JWST instruments (NIRCAM, NIRSPEC, and MIRI) on well-studied cosmological fields. This ERS project is lead by S. Finkelstein at the University of Texas and includes E. Daddi and D. Elbaz from CEA-Saclay among the international teams of proposers. These observations will be among the first delivered by JWST, in parallel with those from GTO teams.

The student will ultimately use the Early Release Science data on cosmological fields to search for ongoing hidden merger events and AGN components resolved inside galaxies, by distinguishing them from the whole galaxy (e.g., nuclear events, or similar), and constraining the growth of inner bulges with passive and/or active stellar populations. This research is based on recent discoveries from our team at the peak of galaxy formation z=1-4. Eventually this research will lead to the first realistic estimates of the relevance of these widely discussed and hot topics.

The intergalactic magnetic field pervading the cosmic voids is suspected to be a relic field originating from the very first epoch of the cosmic history. The goal of this PhD is to look for signatures of this field in the high-energy data of gamma-ray bursts, and to predict the ability of the future CTA observatory to constrain its properties. This work combines both theoretical modelling and analysis of simulated CTA data.

Gamma bursts, discovered by chance in the late 1960s, are the most violent explosions in the universe. Their study is complex because it requires the deployment in space of a gamma telescope to detect and locate them. All the data collected made it possible to establish a global scientific scenario, the main lines of which are as follows: gamma-ray bursts are stellar explosions that result in the formation of a black hole and the ejection of material jets propelled at speeds very close to that of light. When the jet is directed towards the Earth, an observer sees an extremely bright source, which decreases rapidly over time. A typical gamma-ray burst includes a prompt emissive phase that lasts a few seconds, followed by a persistent emission when the jet hits the surrounding environment violently.

The main objective of the Franco-Chinese SVOM mission, which will be launched at the end of 2021, will be to establish a sample of 30 to 40 gamma-ray bursts per year with as complete a description as possible. Thanks to its instruments deployed in space and on the ground, for the first time, the prompt emission of the burst will be observed over more than three decades in energy and the associated persistent emission will be studied in X, visible and near infrared.

The proposed thesis topic consists in jointly studying, from the catalogue of GRB detected by the SVOM mission, the prompt phase and evolution of remanence over a period of a few days. The interpretation of these observations will provide us with information on the nature of the jet, the acceleration of particles, the production of radiation in the jet and in the surrounding environment shocked by the jet, the properties of the same medium (stellar wind, interstellar medium, molecular cloud) and the nature of the star that exploded.

The recent direct detections of gravitational waves (GW) from mergers of

massive compact objects has opened a new window to our Universe. The

electro-magnetic (EM) counterpart of the event GW170817 started a new

multi-messenger era for astronomy. Joint GW and EM observations provide a way

to better understand the physics and rate of violent processes of black hole

and neutron star mergers, and the properties of their host galaxies and stellar

populations.



To identify GW transients via quick follow-up observations across the EM

spectrum, galaxy surveys from ultra-violet (UV), optical, to infrared (IR)

wavelengths are of great importance. This PhD project will explore the synergy

and complementarity of two upcoming space missions, the ESA satellite Euclid

(launch in 2022), and CSST, the Chinese Space Station Telescope (planned for

2024). Both missions will cover a large fraction of the extra-galactic sky

with a common area of 15,000 deg^2.

Gaseous giant planets like Jupiter and Saturn in our solar system and “hot” Jupiters orbiting around other stars with very short periods are complex and fascinating objects. Indeed, they are turbulent rotating magnetic bodies that have strong interactions with their environment: their moons in the case of Jupiter and Saturn and their host stars in the case of “hot” Jupiters/Saturns. In such gaseous giant planets’ systems, tidal forces, the tidal waves they excite and their dissipation shape the orbital architecture and the rotational dynamics of the planets. During the last decade, several revolutions have occurred for our understanding of tides in these systems. On the one hand, high precision astrometry and the CASSINI (NASA/ESA) space mission have measured dissipation stronger by one order of magnitude than expected in Jupiter and Saturn. On the other hand, the large space-based photometric surveys Kepler/K2 and now TESS (NASA) are observing a broad diversity of orbital architecture for exoplanetary systems while “hot” gaseous giant planets seems to host a weakest tidal dissipation than Jupiter and Saturn. Finally, the space mission JUNO (NASA) and the grand finale of CASSINI have revealed the internal structure and dynamics of Jupiter and Saturn: the intense zonal flows observed at their surface are confined in their external layers because of the action of the magnetic field in their deepest regions while the heavy elements contained in their core are mixed in their deep gaseous envelope that modifies the global structure of the planet. The objective of this PhD project is thus to build the new coherent models of tidal dissipation in gaseous giant (exo-)planets mandatory to understand the evolution of their systems. They will take into account all these complex phenomena and new observational constraints. These models will be applied to predict the evolution of planetary systems in support of ongoing and forthcoming space missions in which the CEA-IRFU Department of Astrophysics is strongly involved (JWST, PLATO, ARIEL).

Are cosmic rays actors or passengers in galaxy evolution? In the current models of galaxy evolution stars form too efficiently and too early in the history of the Universe. High-energy processes such as jets from supermassive black holes and supernova explosions can modify how the gas and magnetic fields cycle in and out of a galaxy, but their impact fails to explain key observations such as galactic outflows. Cosmic rays can play a particular role in galaxy evolution as they mediate energy transfers from supernovae to the interstellar medium over thousands of parsecs and tens of millions of years around their source. They also increase the gas buoyancy and add anisotropic pressures along magnetic field lines and off galactic discs. To evaluate their impact, it is central to understand how cosmic rays propagate through a galaxy and how their transport properties vary with the ambient interstellar conditions. To gain insight into this problem, we propose to compare for the first time the distribution of cosmic rays obtained in numerical simulations of interstellar clouds with measurements obtained from multi-wavelength observations in comparable regions of the Milky Way. A team of well-known experts in the Astrophysics Department will advise the PhD student on high-performance computing simulations and on multi-tracer observations of the interstellar medium, magnetic topology, and cosmic rays. He or she will also work within the broad international collaboration for the Fermi Gamma-ray Space Telescope.

The study of exoplanets is booming. Since the detection of the 1st exoplanet in 1995 by Mr. Mayor and D. Queloz (2019 Nobel Prize in Physics), more than 4000 exoplanets have been detected. The domain is now faced with a new challenge: the characterisation of the atmosphere of exoplanets. Knowledge of the atmosphere brings unique information to constrain the formation and evolution of the exoplanet, its interior, even the presence of biological activity, etc. This characterisation will take a considerable step forward with the launch of two space missions: the JWST in 2021 and the ARIEL mission, entirely dedicated to exoplanet atmospheres, in 2028. The atmosphere is studied from spectroscopic infrared observations; the level of instrumental stability required for these studies is very high (up to 10 ppm over 10 hours).

The JWST was not designed to have the required stability. During his/her thesis the student will determine the stability in flight of the JWST MIRI instrument, to which CEA has made a strong contribution, will compare it with the predicted one and will analyse different methods to improve the stability during data reduction. CEA is also strongly involved in the ARIEL mission (mastery of the main instrument of ARIEL: the AIRS infrared spectrometer; realization and testing of the detection chain). The student will participate in the studies of the instrumental stability (laboratory tests of the detection chain, analysis of results, determination of the best operating modes, system analysis) in order to maximize the instrumental stability upstream to the launch.

Key words : space missions, infrared detectors, exoplanets

We propose a PhD thesis which builds on the tools and expertise available within the lab, and aims at providing key Tools and results that will be used for the Euclid collaboration and beyond.

The hired PhD candidate within this project will be at the interface between theory and observations to get the best scientific return out of the big investment done in space missions like Euclid, in particular in Europe and by CNES.

The main objectives are:

1) learn how to use existing XC codes (such as COSMOSIS, developed by Martin Kilbinger) and use available data (such as real or simulated data for Euclid) to test modified gravity models beyond LCDM (with supervision of Valeria Pettorino, expert in the field);

2) investigate how large the contribution of XC with spectroscopic galaxy clustering would be, potentially using 3D WL (for which a code has been validated by A. S. Mancini & V. Pettorino);

3) investigate synergies with other probes, such as data from BOSS/eBOSS (of which Vanina Rulhmann-Kleider is expert) and the Cosmic Microwave Background from Planck (of which V.Pettorino is a CORE2 team member and Planck scientist) or next to come ground space / balloon experiments which will provide (during the time of the PhD) polarisation spectra with a better resolution at small scales.

Clusters of galaxies are the most massive entities in the universe. As such, they constitute powerful cosmological probes.

The XLL survey is the largest programme of the European satellite XMM (X-ray band). It has enabled the detection of several hundreds of galaxy clusters.

The goal of the PhD is to perform the cosmological analysis of the cluster sample by using artificial intelligence techniques : deep learning and convolutive neural networks.

Space telescopes focusing hard X-rays up to 100-200 keV would bring breakthrough in our understanding of the most violent and energetic phenomena of the Universe, like those in the Active Galaxy Nuclei, the supernovae, or closer to us in the solar flares. Techniques of super-mirrors are emerging for that prospect. In parallel, technological developments in rupture shall be led to realize large focal planes with high pixel density and efficient up to 200 keV to be placed at the focal plane of these optical systems. This PhD thesis in space instrumentation consists in setting up and studying innovative hybrid detectors for the imaging spectroscopy in the 1-200 keV energy range, based on 250 µm-pitch pixelated cadmium telluride (CdTe) semiconductor detectors, point to point connected to spectroscopic readout channels of dot-matrix application specified integrated circuits (ASIC) designed in our institute. By means of experimental characterizations coupled to modern data analysis methods, and to tests in accelerators coupled to numerical simulations, the candidate will demonstrate and optimize the spectral resolution, the spatial resolution and the counting capabilities of these new devices.