ED Mathématiques et Informatique
Reachability in temporal graphs and related problems
by Timothée CORSINI (LaBRI - Laboratoire Bordelais de Recherche en Informatique)
The defense will take place at 14h00 - Amphi 351, cours de la Libération bâtiment A30, 33405 Talence
in front of the jury composed of
- Arnaud CASTEIGTS - Professeur - Université de Bordeaux - Directeur de these
- Colette JOHNEN - Professeure - Université de Bordeaux - Examinateur
- Laurent VIENNOT - Directeur de recherche - Inria Paris - Rapporteur
- Thomas ERLEBACH - Professor - Durham University - Rapporteur
- Binh Minh BUI XUAN - Chargé de recherche - Université Pierre et Marie Curie - Examinateur
A temporal graph is a graph in which the edge set varies over time. Such graphs find application in many domains, such as epidemics study, transportation networks, social network analysis, and mobile adhoc networks. The study of temporal graphs is fairly recent, and has become significant over the past two decades. The work in this thesis is focused on temporal graphs, and more precisely on reachability problems in these graphs. In temporal graphs, we say that a vertex can reach another if there is a temporal path, also called journey, from the former to the latter, whose edges can be traversed in a chronological order. What makes temporal graphs different from their static counterpart is the fact that reachability is neither a symmetric relation nor a transitive one, leading to many algorithmic challenges. This thesis aims to bring new results to temporal graphs as well as clarifying the importance of certain parameters in the setting, for instance the impact of considering strict or non-strict journeys. Another chapter of the thesis focuses on temporal spanners, which are conceptual analogues of spanning trees in static graphs, namely, a subset of the input temporal graph that preserves reachability and is as small as possible. Finally, this thesis also studies relations between temporal and static problems, through the lenses of a special type of heredity called robustness, motivated by temporal graphs and applied here to the cases of diameter and distances in static graphs.
Implicit Finite Volume schemes for hypersonic steady state flows
by Benoît COSSART (IMB - Institut de Mathématiques de Bordeaux)
The defense will take place at 14h00 - Salle de conférence de l'IMB IMB, institut Mathématiques de Bordeaux, bâtiment A33, 33400 Talence, France
in front of the jury composed of
- Raphaël LOUBERE - Directeur de recherche - Université de Bordeaux - Directeur de these
- Christophe BERTHON - Professeure des universités - Université de Nantes - Rapporteur
- Florian DE VUYST - Professeur des universités - UTC Compiègne - Rapporteur
- Héloïse BEAUGENDRE - Professeure des universités - Bordeaux INP - Examinateur
- Céline BARANGER - Ingénieure de recherche - CEA CESTA - Examinateur
- Alexandre ERN - Professeur des universités - CERMICS ENPC - Examinateur
- Florent RENAC - Ingénieur de recherche - ONERA Châtillon - Examinateur
- Jean-Philippe BRAEUNIG - Ingénieur de recherche - CEA CESTA - CoDirecteur de these
Research into hypersonic vehicles is experiencing a revival of interest, largely supported by advances in computing performance, enabling the simulation of complex hypersonic flows. In the case of continuous regimes, the Navier-Stokes equations, derived from the conservation laws, are commonly used. finite volumes method is classically used for integration in space, and to approximate integration in time, linearised implicit methods produce a linear system. They are more suitable than explicit methods because they offer a better compromise between robustness and computational cost for flows comprising physical phenomena with very different characteristic times, which is the case for hypersonic flows. Characterising the stability of these methods is difficult, particularly when they are applied to systems of conservation laws. In this context, this thesis aims to provide elements for analysing the stability of these schemes in order to make them more robust and suitable for unstructured meshes. Among stability analysis methods, von Neumann analysis can be used on linear scalar equations. In the first part of this thesis, it is proposed to extend this analysis method to the non-linear, monodimensional Burgers' scalar equation. Then, the analysis is extended to the Euler system of 1D and 2D conservation laws by formulating additional hypotheses. In particular, this work highlights a constraint for the design of implicit schemes that are more robust and stable than the scheme used, for example, by H.C. Yee, A. Harten and R.F. Warming (Journal of Computational Physics, 57(3):327-360, 1985). To obtain an implicit scheme satisfying this constraint, the hyperbolic structure of the system is exploited by linearisation of the Roe numerical scheme. In a second step, the alternative implicit scheme thus obtained is extended to other finite volumes schemes included in the the right hand side by an additional linear analysis. This analysis reveals a new stability constraint linking the numerical diffusion applied to the matrix of the linear system to that applied to the right hand side. A matrix correction linked to this constraint is proposed and numerically validated. Finally, another part of the thesis proposes a reformulation of the equations into physical flux to improve the efficiency of the resolution of the linear system derived from the implicit schemes.
ED Sciences de la Vie et de la Santé
Role of amyloid signaling in regulated cell death in microorganisms
by Léa IBARLOSA (Institut de Biochimie et Génétique Cellulaires)
The defense will take place at 13h00 - Salle de conférence, IBGC CNRS IBGC CNRS, 146 rue Léo Saignat, 33000 Bordeaux
in front of the jury composed of
- Sven SAUPE - Directeur de recherche - Université de Bordeaux - Directeur de these
- Aude BERNHEIM - Chargée de recherche - Université Paris Cité - Rapporteur
- Fabienne MALAGNAC - Professeur - Université Paris Sud - Rapporteur
- Claire LE HENNAF LE MARREC - Professeur - Université de Bordeaux - Examinateur
- Luc BOUSSET - Chargé de recherche - Université Paris-Saclay - Examinateur
Regulated cell death (RCD) plays a central role in defense mechanisms in animals, plants, fungi and bacteria. In immune RCD, infected cells of an organism or a cell population are sacrificed to halt or delay pathogen replication. Although generally associated with human neurodegenerative diseases as pathological entities, amyloids structures can play a functional role in controlling RCD processes, particularly in animals and fungi. Signaling amyloids act as molecular switches transmitting an activation signal from an immune receptor to a downstream cell death effector. Here, we find that this mode of signal transduction based on amyloid folding occurs in an antiphage defense system of Escherichia coli. In the Bab/Agp antiphage system conferring resistance to T5, amyloid signaling controls the activation of a membrane-targetting cell death-inducing protein, which shows homologies with a protein superfamily operating in RCD-pathways in fungi, plants and animals. In two brief chapters, we also report experiments on the characterization of an amyloid signaling motif found in the actinomycetes Streptomyces coelicolor and of a fungal RCD module encoding a gasdermin and its cognate activating protease. This work extends the role of amyloid signaling to antiphage defense and illustrates at different levels, transkingdom conservation of regulated cell death pathways in bacteria and fungi and other branches of the tree-of-life.
ED Sciences Physiques et de l'Ingénieur
Development and characterisation of the double Penning trap PIPERADE and mass measurements of neutron-rich isotopes near N=40
by Mathieu FLAYOL (Laboratoire de Physique des 2 Infinis de Bordeaux)
The defense will take place at 14h00 - Amphi Wegener Université de Bordeaux - 351 cours de la Libération - Bâtiment A22 - Amphi Wegener - 33405 Talence (Campus Peixotto)
in front of the jury composed of
- Stéphane GREVY - Directeur de recherche - LP2I - Bordeaux - Directeur de these
- Sarah NAIMI - Directeur de recherche - Laboratoire de Physique des 2 infinis Irène Joliot-Curie - IJClab - Rapporteur
- Maxime BRODEUR - Associate Professor - University of NOTRE DAME - Rapporteur
- Vladimir MANEA - Chargé de recherche - Laboratoire de Physique des 2 infinis Irène Joliot-Curie - IJClab - Examinateur
- NADEZDA SMIRNOVA - Professeur des universités - LP2I - Bordeaux - Examinateur
The objective of this thesis focuses on the development and characterization of the PIPERADE double Penning trap (PIèges de PEnning pour les RAdionucléides à DESIR) at LP2i Bordeaux, as well as on mass measurements in the N=40 region conducted at the University of Jyväskylä, Finland. Mass spectrometry with Penning traps allows for the precise measurement of atomic nuclei masses. Nuclear binding energies, derived from atomic masses, are crucial for studying the structure of exotic nuclei and constraining nuclear physics models. PIPERADE is dedicated to precision mass measurements and the selection of exotic ions of interest within the DESIR (Désintégration, Excitation et Stockage d'Ions Radioactifs) research facility at GANIL (Grand Accélérateur National d'Ions Lourds). This work presents the implementation of various purification and mass measurement techniques as well as the implementation of a new position-sensitive detector using microchannel plates and with three-layer delay-lines. The first technique explored is purification by Buffer Gas Cooling (BGC), followed by the "Time-of-Flight Ion Cyclotron Resonance" (ToF-ICR), applied for the first time with PIPERADE in November 2021. The optimization of these two techniques was subsequently pursued in parallel with the implementation of a new method, the "Phase-Imaging Ion Cyclotron Resonance" (PI-ICR). This method was successfully applied for the first time on PIPERADE in September 2023. This work also presents the first mass measurements using the ToF-ICR and PI-ICR techniques on PIPERADE. In the second part of the thesis, mass measurements of neutron-rich nuclei were performed using the JYFLTRAP Penning trap, at the IGISOL facility. The masses of both ground and isomeric states of the isotopes of cobalt $^{68}$Co to $^{70}$Co, were measured, either for the first time for the isomeric states of $^{68}$Co and $^{70}$Co or with greatly improved precision for the others. The masses of the ground states of $^{79}$Ge and $^{73}$Ni were also measured. The analysis of the results allowed for comparison with theoretical models such as the Large Scale Shell Model (LSSM) and the Discrete Nonorthogonal Shell Model (DNO-SM). The results have a direct impact on understanding the N=40 gap and the region of the island of inversion.
Multidimensional approach to the prevention, diagnosis and follow-up of knee injuries in athletes
by Clément LIPPS (Laboratoire de l'Intégration du Matériau au Système)
The defense will take place at 14h00 - Amphithéâtre Rougier Bâtiment Rougier, Faculté des STAPS, 12 Avenue Camille Jullian, 33600 PESSAC
in front of the jury composed of
- Thierry WEISSLAND - Maître de conférences - Université de Bordeaux - Directeur de these
- Emilie SIMONEAU - Professeure des universités - Université Polytechnique Hauts-de-France - Rapporteur
- Pascal EDOUARD - Professeur des universités - praticien hospitalier - Université Jean Monnet, Saint-Etienne - Rapporteur
- Olivier BUTTELLI - Professeur des universités - Université d'Orléans - Examinateur
- Julien MORLIER - Professeur des universités - Université de Bordeaux - Examinateur
- Julien FRERE - Maître de conférences - Université Grenoble-Alpes - CoDirecteur de these
Knee injuries are the most common lower limb pathologies in competitive or recreational sports. Post-trauma consequences lead to lower performances and pauses in activity. Moreover, despite specific prevention guidelines, the recurrence rate, whether from a rupture of the ligament graft or a contralateral injury, remains very high. Treating these injuries therefore represents a major challenge in terms of return to sports and public health costs. Based on an experimental network platform developed in collaboration between the universities of Bordeaux, Grenoble, and Marseille, which brings together complementary expertise in movement analysis, signal processing, and mathematical modeling, this thesis project aimed to improve knee injury prevention through a multifactorial approach. Relying on a database of 96 individuals built during the doctoral research, we hypothesized that accounting for multifactorial data related to injury (biomechanical and psychological data) and integrating them into a model based on artificial intelligence algorithms would allow better monitoring of injury progression and recovery over time, particularly to determine the optimal time for returning to sport. The first study highlighted the reproducibility of data obtained from the tests performed. The second study demonstrated that it is possible, using an artificial intelligence model and biomechanical and psychological data, to classify individuals based on their history of knee injury. Finally, the third study showed that using all measurement points rather than characteristic variables allowed artificial intelligence models to achieve better performance. Thus, all these studies highlight the predictive capability of the measurements made during the protocol to recognize knee injury sequelae. Additionally, the use of interpretability algorithms in the second and third studies allowed us to identify the functional variables that most influenced the models in classifying individuals. This work therefore optimizes the timing of return to sport and the rehabilitation process by recognizing the presence of knee injury sequelae in individuals and identifying the most important functional variables to work on.
Identification of the criteria determining the acceptability of the behaviour of an autonomous vehicle in potentially conflicting situations in an urban environment in order to adapt the autonomous driving algorithms
by Gabriel NATIVEL-FONTAINE (Laboratoire de l'Intégration du Matériau au Système)
The defense will take place at 14h00 - O108 ENSC 109 Avenue Roul 33400 Talence
in front of the jury composed of
- Véronique LESPINET-NAJIB - Maître de conférences - ENSC - Bordeaux INP - Directeur de these
- Emilie LOUP-ESCANDE - Professeur des universités - Université de Picardie Jules Verne - Rapporteur
- Julien CEGARRA - Professeur des universités - INU Champollion - Rapporteur
- Maud RANCHET - Chargée de recherche - Université Gustave Eiffel - Examinateur
- Pierrick LEGRAND - Professeur des universités - ENSC - Bordeaux INP - Examinateur
This thesis will identify the criteria that will make the use of autonomous vehicles in urban environments, particularly in potentially conflictual situations, acceptable. Our focus is on level 5 autonomous vehicles, which involve full automation of driving functions. We analyze the acceptability of vehicle behavior from the passengers' perspective. The literature describes the acceptability of technologies in three phases: a priori acceptability (before exposure), in situ acceptability (during exposure), and appropriation (after exposure). A technology is accepted when it is used in a non-professional context. As there is no level 5 vehicle currently available, the study of its acceptability is theoretical and a priori. It relies notably on models such as the Unified Theory of Acceptance and Use of Technology (UTAUT). We identified the behavioral elements influencing passengers' perceived acceptability in conflictual situations using a simulated exposure to the technology. These studies allowed us to definitively propose acceptable operating ranges for vehicle behavior. However, there are still some cases where the technology is not acceptable, such as not stopping at a red light to allow an emergency vehicle to pass. Passenger perception is a crucial factor in determining acceptability in these situations. We therefore explored the effectiveness of vehicle-passenger communication as a countermeasure to the loss of acceptability in specific contexts. Our first experiment examined the influence of driving mode and context on the acceptability of vehicle behavior. We tested the driving modes, which varied in their attitude toward traffic laws, at typical urban intersections. Driving mode proved to be the most influential factor, and neither the type of intersection nor sociodemographic characteristics showed significant effects. A second experiment definitively demonstrated the impact of specific behavioral components on passengers' acceptability. This was achieved through a virtual reality simulation in which participants experienced a journey in a Level 5 autonomous vehicle. During the trip, participants encountered a variety of situations in which we deliberately varied the vehicle's behavior, particularly in terms of waiting time before engaging in the situation. The results clearly showed that prior communication with passengers is essential for safety. When passengers were introduced in a way that focused on safety, they were more accepting of all situations. We defined optimal waiting time ranges for each situation. The first part of this thesis analyses the variation in situ acceptability of autonomous vehicle behaviour. These findings provide clear recommendations for manufacturers to improve passenger acceptability. These recommendations can be implemented directly through vehicle behavior or through vehicle-passenger communication. The decisions made by autonomous vehicles, governed by algorithms, directly impact acceptability. These algorithms can and should be adjusted through various parameters to positively influence acceptability. We present a methodology for transferring technology from academic research to industrialization, using mathematical models based on experimental data. These models will allow manufacturers to determine optimal parameter values to maximize acceptability. Finally, the study of transparency and behavior explanation demonstrated that combining multiple communication modalities (visual, auditory) significantly improves explicability and acceptability. By using multiple sensory channels to explain the vehicle's behavior, we observed a substantial increase in acceptability, particularly in conflictual situations where the vehicle's behavior seemed ambiguous. This proves that perceived explicability, reinforced by diverse sensory modalities, is the key to in situ acceptability. Furthermore, this explicability enhances passenger acceptability in situations of uncertainty.
Valorization of dredged sediments in earth constructions
by Sarah NASSAR (I2M - Institut de Mécanique et d'Ingénierie de Bordeaux)
The defense will take place at 14h00 - Amphithéâtre D 351 Cours de la Libération, bâtiment A29
in front of the jury composed of
- Nadia SAIYOURI - Professeur des universités - Université de Bordeaux - Directeur de these
- Emmanuel ROZIERE - Professeur des universités - Ecole Centrale de Nantes - Rapporteur
- Rachid ZENTAR - Professeur des universités - Institut Mines Télécom Lille Douai - Rapporteur
- Cédric BAUDRIT - Chargé de recherche - Université de Bordeaux - Examinateur
- Alexandra BOURDOT - Maître de conférences - Ecole normale supérieure Paris-Saclay - Examinateur
- Anne PANTET - Professeur des universités - Université du Havre - Examinateur
In the context of the ecological transition, there is an awareness about the eco-responsible practices, and France is engaged to face the climate challenge through environmental policies. In the construction sector, it is necessary to take measures to reduce the environmental impact. Earth constructions could substitute the conventional materials and offer many advantages. In masonry, compressed earth blocks (CEB) are used in their raw state and are not fired at high temperatures like fired bricks. In addition, their fabrication is not as energy-consuming as that of concrete blocks. To promote circular economy, waste valorization is very recommended. Dredged sediments are stored in large quantities, and it is necessary to manage them, especially in light of upcoming changes in dredging regulations. Therefore, CEB made from dredged sediments of Arcachon bay are the focus of this study. First, sediments are collected from eight ports of the bay and are characterized through different geotechnical and in-situ testing. The sediments from the port of Audenge are the most suitable for CEB manufacturing and are used for the CEB in this study. Different CEB compositions, manufactured in the laboratory, enabled the study of the effects of compacting pressure, the type and fraction of natural fibers, and the quantity of activator. Hence, non-stabilized, fiber-reinforced and geopolymerised CEB are tested to investigate their mechanical, thermal, microscopic and durability properties. They are also characterized through nondestructive testing, which was practical and efficient for the qualitative, non-damaging evaluation. Furthermore, machine learning models are used to predict the compressive strength of CEB. They are trained by using a database collected from the literature with input parameters including the soil texture, water, stabilizers and compacting pressure. Some classifiers were compared, and the voting classifier performed the best. Explicability of the predictions is also investigated through the Shapley values. This approach serves as a decision-making support tool for future CEB manufacturing, given the variability of the soil. Finally, an environmental analysis is carried out through a life cycle assessment, from cradle to gate, and a thermal simulation. They made it possible to propose improvements to manufacturing processes and to show the interest of using CEB in the construction to reduce heating needs.