Machine learning for visual inspection and final state reconstruction in the Atlas experiment
The Large Hadron Collider (LHC) is taking data at an unprecedented energy of 13.6 TeV, allowing to explore a new energy frontier and rare processes. In the meantime its high luminosity phase (HL-LHC) is actively being prepared. One of the main goals of both the on-going run and the HL-LHC phase is to scrutinize potential deviations from the predictions of the Standard Model of particle physics. For that purpose, it is particularly relevant to study the cornerstone of the model, the Higgs boson, as well as the heaviest known elementary particle, the top quark.
The PhD topic is aiming to develop machine learning techniques for two different challenging projects within the ATLAS experiment at the LHC: visual inspection of new ATLAS tracker modules, and reconstruction of final state objects to study rare processes with multiple leptons in the final state.
In order to meet the performance required by HL-LHC, the current ATLAS inner detector will be replaced by a new all-silicon detector, partially composed of pixel modules. The pixel module production includes multiple assembly steps each one being followed by quality control. The first part of the PhD project is aiming to automate visual inspection of the module metallic contacts (pads) before connecting them to the electronics. Initial work showed that visible defects presented on the pad surface are detectable by machine learning algorithms.
The PhD work will explore several directions:
- Full investigation of supervised approaches. The production start-up period will allow to collect the dataset necessary to train supervised segmentation models targeting not only anomaly detection, but also their classification.
- Investigation of unsupervised anomaly detection based on generative models. The DRAEM model [1] for instance learns a joint representation of an anomalous image and its anomaly-free reconstruction, while simultaneously learning to distinguish them. The reconstructive sub-network is formulated as an encoder-decoder architecture and can be replaced with more advanced types of generative models such as GANs.
- Investigation of various models treating an anomaly-free pad surface as a background noise. Signal processing denoising algorithms can then be used to increase the information content while e.g denoising autoencoders can learn features that are invariant to noise.
The second part of the PhD will focus on rare processes involving top quarks with bosons (ttW, ttZ, ttH, tH, 4tops). These processes have just been observed for the first time using the dataset recorded at 13 TeV (except for tH) in final states with several leptons. However these analyses have always been carried out separately, which strongly limits our overall physics understanding, the correlations between the various measurements being large and unknown. The new approach proposed here consists of designing a global analysis to tackle these limitations. This approach requires cutting edge techniques to reconstruct precisely the final state. Indeed several degrees of freedom originate from the undetected neutrinos and combinatorics is made complex by the multiple decay products. An approach based on GNN was already tried. Although it provides a certain level of accuracy, some other approaches have to be explored to improve the reconstruction performance. There are in particular two models of interest developed within the LHC community:
- The topograph model [2], also based on GNN, makes use of additional information about the event decay chain.
- The Spanet model [3], a symmetry-preserving attention network exploits natural invariances to efficiently find particle assignments without evaluating all permutations.
An assessment of the relevant uncertainties is also an important aspect when using these tools.
This dual project provides an opportunity to be involved in the full chain of the AI application development, being also part of the world-wide ATLAS collaboration.
Point Spread Function Modelling for Space Telescopes with a Differentiable Optical Model
Context
Weak gravitational lensing [1] is a powerful probe of the Large Scale Structure of our Universe. Cosmologists use weak lensing to study the nature of dark matter and its spatial distribution. Weak lensing missions require highly accurate shape measurements of galaxy images. The instrumental response of the telescope, called the point spread function (PSF), produces a deformation of the observed images. This deformation can be mistaken for the effects of weak lensing in the galaxy images, thus being one of the primary sources of systematic error when doing weak lensing science. Therefore, estimating a reliable and accurate PSF model is crucial for the success of any weak lensing mission [2]. The PSF field can be interpreted as a convolutional kernel that affects each of our observations of interest, which varies spatially, spectrally, and temporally. The PSF model needs to be able to cope with each of these variations. We use specific stars considered point sources in the field of view to constrain our PSF model. These stars, which are unresolved objects, provide us with degraded samples of the PSF field. The observations go through different degradations depending on the properties of the telescope. These degradations include undersampling, integration over the instrument passband, and additive noise. We finally build the PSF model using these degraded observations and then use the model to infer the PSF at the position of galaxies. This procedure constitutes the ill-posed inverse problem of PSF modelling. See [3] for a recent review on PSF modelling.
The recently launched Euclid survey represents one of the most complex challenges for PSF modelling. Because of the very broad passband of Euclid’s visible imager (VIS) ranging from 550nm to 900nm, PSF models need to capture not only the PSF field spatial variations but also its chromatic variations. Each star observation is integrated with the object’s spectral energy distribution (SED) over the whole VIS passband. As the observations are undersampled, a super-resolution step is also required. A recent model coined WaveDiff [4] was proposed to tackle the PSF modelling problem for Euclid and is based on a differentiable optical model. WaveDiff achieved state-of-the-art performance and is currently being tested with recent observations from the Euclid survey.
The James Webb Space Telescope (JWST) was recently launched and is producing outstanding observations. The COSMOS-Web collaboration [5] is a wide-field JWST treasury program that maps a contiguous 0.6 deg2 field. The COSMOS-Web observations are available and provide a unique opportunity to test and develop a precise PSF model for JWST. In this context, several science cases, on top of weak gravitational lensing studies, can vastly profit from a precise PSF model. For example, strong gravitational lensing [6], where the PSF plays a crucial role in reconstruction, and exoplanet imaging [7], where the PSF speckles can mimic the appearance of exoplanets, therefore subtracting an accurate and precise PSF model is essential to improve the imaging and detection of exoplanets.
PhD project
The candidate will aim to develop more accurate and performant PSF models for space-based telescopes exploiting a differentiable optical framework and focus the effort on Euclid and JWST.
The WaveDiff model is based on the wavefront space and does not consider pixel-based or detector-level effects. These pixel errors cannot be modelled accurately in the wavefront as they naturally arise directly on the detectors and are unrelated to the telescope’s optic aberrations. Therefore, as a first direction, we will extend the PSF modelling approach, considering the detector-level effect by combining a parametric and data-driven (learned) approach. We will exploit the automatic differentiation capabilities of machine learning frameworks (e.g. TensorFlow, Pytorch, JAX) of the WaveDiff PSF model to accomplish the objective.
As a second direction, we will consider the joint estimation of the PSF field and the stellar Spectral Energy Densities (SEDs) by exploiting repeated exposures or dithers. The goal is to improve and calibrate the original SED estimation by exploiting the PSF modelling information. We will rely on our PSF model, and repeated observations of the same object will change the star image (as it is imaged on different focal plane positions) but will share the same SEDs.
Another direction will be to extend WaveDiff for more general astronomical observatories like JWST with smaller fields of view. We will need to constrain the PSF model with observations from several bands to build a unique PSF model constrained by more information. The objective is to develop the next PSF model for JWST that is available for widespread use, which we will validate with the available real data from the COSMOS-Web JWST program.
The following direction will be to extend the performance of WaveDiff by including a continuous field in the form of an implicit neural representations [8], or neural fields (NeRF) [9], to address the spatial variations of the PSF in the wavefront space with a more powerful and flexible model.
Finally, throughout the PhD, the candidate will collaborate on Euclid’s data-driven PSF modelling effort, which consists of applying WaveDiff to real Euclid data, and the COSMOS-Web collaboration to exploit JWST observations.
References
[1] R. Mandelbaum. “Weak Lensing for Precision Cosmology”. In: Annual Review of Astronomy and Astro- physics 56 (2018), pp. 393–433. doi: 10.1146/annurev-astro-081817-051928. arXiv: 1710.03235.
[2] T. I. Liaudat et al. “Multi-CCD modelling of the point spread function”. In: A&A 646 (2021), A27. doi:10.1051/0004-6361/202039584.
[3] T. I. Liaudat, J.-L. Starck, and M. Kilbinger. “Point spread function modelling for astronomical telescopes: a review focused on weak gravitational lensing studies”. In: Frontiers in Astronomy and Space Sciences 10 (2023). doi: 10.3389/fspas.2023.1158213.
[4] T. I. Liaudat, J.-L. Starck, M. Kilbinger, and P.-A. Frugier. “Rethinking data-driven point spread function modeling with a differentiable optical model”. In: Inverse Problems 39.3 (Feb. 2023), p. 035008. doi:10.1088/1361-6420/acb664.
[5] C. M. Casey et al. “COSMOS-Web: An Overview of the JWST Cosmic Origins Survey”. In: The Astrophysical Journal 954.1 (Aug. 2023), p. 31. doi: 10.3847/1538-4357/acc2bc.
[6] A. Acebron et al. “The Next Step in Galaxy Cluster Strong Lensing: Modeling the Surface Brightness of Multiply Imaged Sources”. In: ApJ 976.1, 110 (Nov. 2024), p. 110. doi: 10.3847/1538-4357/ad8343. arXiv: 2410.01883 [astro-ph.GA].
[7] B. Y. Feng et al. “Exoplanet Imaging via Differentiable Rendering”. In: IEEE Transactions on Computational Imaging 11 (2025), pp. 36–51. doi: 10.1109/TCI.2025.3525971.
[8] Y. Xie et al. “Neural Fields in Visual Computing and Beyond”. In: arXiv e-prints, arXiv:2111.11426 (Nov.2021), arXiv:2111.11426. doi: 10.48550/arXiv.2111.11426. arXiv: 2111.11426 [cs.CV].
[9] B. Mildenhall et al. “NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis”. In: arXiv e-prints, arXiv:2003.08934 (Mar. 2020), arXiv:2003.08934. doi: 10.48550/arXiv.2003.08934. arXiv:2003.08934 [cs.CV].
Analysis and experimental study of capillary structures to mitigate the influence of magnetogravitational forces on liquid helium cooling for future HTS superconducting magnets
As physics requires increasingly higher magnetic fields, CEA is called upon to develop and produce superconducting magnets capable of generating magnetic field of more than 30 T. The windings of these electromagnets are made from superconducting materials whose electrical resistance is extremely low at cryogenic temperatures (a few Kelvins). This enables them to carry high currents (>10 kA) while dissipating a minimum of heat by Joule effect. Cooling at these low temperatures is achieved using liquid helium. But helium is diamagnetic. Magnetic fields will therefore induce volumetric forces that add to or oppose gravity within the helium. These magneto-gravity forces disrupt the convective phenomena required to cool the superconducting magnet. This can lead to a rise in their temperature and a loss of their superconducting state, which is essential for their proper operation. In order to circumvent this phenomenon, a new cooling system never used in cryomagnetism will be studied. This cooling system will be developed using heat pipes whose operation is based on capillary forces that are theoretically independent of the magneto-gravity forces induced by strong magnetic fields. These capillary structures can take several forms (microchannels, foam, mesh, etc.). In the framework of the thesis these different structures will be studied theoretically and then experimentally, both with and without magnetic forces, in order to determine the most suitable structures for the future superconducting magnets.
Mesure de la réponse intra-pixel de détecteur infrarouge à base de HgCdTe avec des rayons X pour l’astrophysique
In the field of infrared astrophysics, the most commonly used photon sensors are detector arrays based on the HgCdTe absorbing material. The manufacturing of such detectors is a globally recognized expertise of CEA/Leti in Grenoble. As for the Astrophysics Department (DAp) of CEA/IRFU, it holds renowned expertise in the characterization of this type of detector. A key characteristic is the pixel spatial response (PSR), which describes the response of an individual pixel in the array to the point-like generation of carriers within the absorbing material at various locations inside the pixel. Today, this detector characteristic has become a critical parameter for instrument performance. It is particularly crucial in applications such as measuring galaxy distortion or conducting high-precision astrometry. Various methods exist to measure this quantity, including the projection of point light sources and interferometric techniques. These methods, however, are complex to implement, especially at the cryogenic operating temperatures of the detectors.
At the DAp, we propose a new method based on the use of X-ray photons to measure the PSR of infrared detectors. By interacting with the HgCdTe material, the X-ray photon generates carriers locally. These carriers then diffuse before being collected. The goal is to derive the PSR by analyzing the resulting images. We suggest a two-pronged approach that integrates both experimental methods and simulations. Data analysis methods will also be developed. Thus, the ultimate objective of this thesis is to develop a new, robust, elegant, and fast method for measuring the intra-pixel response of infrared detectors for space instrumentation. The student will be based at the DAp. This work also involves collaboration with CEA/Leti, combining the instrumental expertise of the DAp with the technological knowledge of CEA/Leti.
First observations of the TeV gamma-ray sky with the NectarCAM camera for the CTA observatory
Very high energy gamma-ray astronomy is a relatively young part of astronomy (30 years), looking at the sky above 50 GeV. After the success of the H.E.S.S. array in the 2000s, an international observatory, the Cherenkov Telescope Array (CTA), should start operating by 2026. This observatory will include a total of 50 telescopes, distributed on two sites. IRFU is involved in the construction of the NectarCAM, a camera intended to equip the "medium" telescopes (MST) of CTA. The first NectarCAM (of the nine planned) is being integrated at IRFU and will be shipped on site in 2025. Once the camera is installed, the first astronomical observations will take place, allowing to fully validate the functioning of the camera. The thesis aims at finalizing the darkroom tests at IRFU, preparing the installation and validating the operation of the camera on the CTA site with the first astronomical observations. It is also planned for the student to participate in H.E.S.S. data analysis on astroparticle topics (search for primordial black holes, constraints on Lorentz Invariance using distant AGN).
ARTIFICIAL INTELLIGENCE TO SIMULATE BIG DATA AND SEARCH FOR THE HIGGS BOSON DECAY TO A PAIR OF MUONS WITH THE ATLAS EXPERIMENT AT THE LARGE HADRON COLLIDER
There is growing interest in new artificial intelligence techniques to manage the massive volume of data collected by particle physics experiments, particularly at the LHC collider. This thesis proposes to study these new techniques for simulating the rare-event background from the two-muon decay of the Higgs boson, as well as to implement a new artificial intelligence method for simulating the response of the muon spectrometer detector resolution, which is crucial for this analysis.
SEARCH FOR DIFFUSE EMISSIONS AND SEARCHES IN VERY-HIGH-ENERGY GAMMA RAYS AND FUNDAMENTAL PHYSICS WITH H.E.S.S. AND CTAO
Observations in very-high-energy (VHE, E>100 GeV) gamma rays are crucial for understanding the most violent non-thermal phenomena at work in the Universe. The central region of the Milky Way is a complex region active in VHE gamma rays. Among the VHE gamma sources are the supermassive black hole Sagittarius A* at the heart of the Galaxy, supernova remnants and even star formation regions. The Galactic Center (GC) houses a cosmic ray accelerator up to energies of PeV, diffuse emissions from GeV to TeV including the “Galactic Center Excess” (GCE) whose origin is still unknown, potential variable sources at TeV, as well as possible populations of sources not yet resolved (millisecond pulsars, intermediate mass black holes). The GC should be the brightest source of annihilations of massive dark matter particles of the WIMPs type. Lighter dark matter candidates, axion-like particles (ALP), could convert into photons, and vice versa, in magnetic fields leaving an oscillation imprint in the gamma-ray spectra of active galactic nuclei (AGN).
The H.E.S.S. observatory located in Namibia is composed of five atmospheric Cherenkov effect imaging telescopes. It is designed to detect gamma rays from a few tens of GeV to several tens of TeV. The Galactic Center region is observed by H.E.S.S. for twenty years. These observations made it possible to detect the first Galactic Pevatron and place the strongest constraints to date on the annihilation cross section of dark matter particles in the TeV mass range. The future CTA observatory will be deployed on two sites, one in La Palma and the other in Chile. The latter composed of more than 50 telescopes will provide an unprecedented scan of the region on the Galactic Center.
The proposed work will focus on the analysis and interpretation of H.E.S.S observations. carried out in the Galactic Center region for the search for diffuse emissions (populations of unresolved sources, massive dark matter) as well as observations carried out towards a selection of active galactic nuclei for the search for ALPs constituting dark matter. These new analysis frameworks will be implemented for the future CTA analyses. Involvement in taking H.E.S.S. data. is expected.
STUDY OF THE MULTI-SCALE VARIABILITY OF THE VERY HIGH ENERGY GAMMA-RAY SKY
Very high energy gamma ray astronomy observes the sky above a few tens of GeV. This emerging field of astronomy has been in constant expansion since the early 1990s, in particular since the commissioning of the H.E.S.S. array in 2004 in Namibia. IRFU/CEA-Paris Saclay is a particularly active member of this collaboration from the start. It is also involved in the preparation of the future CTAO observatory (Cherenkov Telescope Array Observatory), which is now being installed. The detection of gamma rays above a few tens of GeV makes it possible to study the processes of charged particles acceleration within objects as diverse as supernova remnants or active galactic nuclei. Through this, H.E.S.S. aims in particular at answering the century-old question of the origin of cosmic rays.
H.E.S.S. allows measuring the direction, the energy and the arrival time of each detected photon. The time measurement makes it possible to highlight sources which present significant temporal or periodic flux variations. The study of these variable
Direction de la Recherche Fondamentale
Institut de recherche
sur les lois fondamentales de l’univers
emissions (transient or periodic), either towards the Galactic Center or active nuclei of galaxies (AGN) at cosmological distance allows for a better understanding of the emission processes at work in these sources. It also helps characterizing the medium in which the photons propagate and testing the validity of some fundamental physical laws such as Lorentz invariance. It is possible to probe a wide range of time scales variations in the flux of astrophysical sources. These time scales range from a few seconds (gamma ray bursts, primordial black holes) to a few years (binary systems of high mass, active galaxy nuclei).
One of the major successes of H.E.S.S.'s two decades of data-taking. was to conduct surveys of the galactic and extragalactic skies in the very-high energy range. These surveys combine observations dedicated to certain sources, such as the Galactic Center or certain remains of supernovae, as well as blind observations for the discovery of new sources. The thesis subject proposed here concerns an aspect of the study of sources which remains to be explored: the research and study of the variability of very-high energy sources. For variable sources, it is also interesting to correlate the variability in other wavelength ranges. Finally, the source model can help predict its behavior, for example its “high states” or its bursts.
Disequilibrium chemistry of exoplanets’ high-metallicity atmospheres in JWST times
In little more than two years of scientific operations, JWST has revolutionized our understanding of exoplanets and their atmospheres. The ARIEL space mission, to be launched in 2029, will soon contribute to this revolution. A main finding that has been enabled by the exquisite quality of the JWST data is that exoplanet atmospheres are in chemical disequilibrium. A full treatment of disequilibrium is complex, especially when the atmospheres are metal-rich, i.e. when they contain in significant abundances elements other than hydrogen and helium. In a first step, our project will numerically investigate the extent of chemical disequilibrium in the atmospheres of JWST targets suspected to have metal-rich atmospheres. We will use towards that end an in-house photochemical model. In a second step, our project will explore the effect of super-thermal chemistry as a driver of chemical disequilibrium. This will offer previously-unexplored insight into the chemistry of metal-rich atmospheres, with the potential to shed new light into the chemical and evolutionary paths of low-mass exoplanets.
Nuclear reactions induced by light anti-ions - contribution of the INCL model
The interaction of an antiparticle with an atomic nucleus is a type of reaction that needs to be simulated in order to answer fundamental questions. Examples include the PANDA (FAIR) collaboration with antiproton beams of the order of GeV, which plans to study nucleon-hyperon interactions, as well as the neutron skin by producing hyperons and antihyperons. This same neutron skin is also studied with antiprotons at rest in the PUMA experiment (AD - Cern). At the same site, we are collaborating with the ASACUSA experiment to study the production of charged particles. To respond to those studies, our INCL nuclear reaction code has been extended to antiprotons (thesis by D. Zharenov, defended at the end of 2023). Beyond the antiproton there are antideuterons and antiHe-3. These antiparticles are of more recent interest, notably with the GAPS (General AntiParticle Spectrometer) experiment, which aims to measure the fluxes of these particles in cosmic rays. The idea is to highlight dark matter, of which these particles are thought to be decay products, and whose measured quantity should emerge more easily from the astrophysical background noise than in the case of antiprotons. The proposed subject is therefore the implementation of light anti-nuclei in INCL with comparisons to experimental data.