HAL-Ecole des Ponts ParisTech
Not a member yet
42743 research outputs found
Sort by
Stochastic generalized standard materials and risk-averse effective behavior
No full textInternational audienceIn this work, we develop a theoretical formulation for describing dissipative material behaviors in a stochastic setting, using the framework of Generalized Standard Materials (GSM). Our goal is to capture the variability inherent in the material model while ensuring thermodynamic consistency, by employing the mathematical framework of stochastic programming. We first show how average behaviors can be computed using the expected value of the free energy and dissipation pseudo-potentials. We then introduce the concept of a risk-averse effective measure, which provides both an optimistic and a pessimistic estimate of the uncertain material behavior. To this end, we utilize the Conditional Value-at-Risk, a widely used risk measure in mathematical finance. We also demonstrate how these concepts can be extended to variational problems at the structure scale, allowing us to compute the effective response of a structure composed of a stochastic material
Aid allocation with optimal monitoring: Theory and policy
No full textInternational audienceWe explore the implications of allowing a poverty-averse donor to monitor aid use within the familiar context of the needs vs. aid effectiveness tradeoff. The paper focuses on the optimal aid allocation between two countries when the donor simultaneously decides about aid shares and country-specific monitoring effort aimed at increasing the amount reaching the poor. Endogenizing aid effectiveness is shown to raise the poor’s income in the worse-governed country, yet not necessarily in the better-governed one, whereas the effect on country aid shares is essentially ambiguous. Those results still hold when the basic model is extended in various directions. Conventional aid allocation rules should be re-examined in their light
Les trois âges de l'histoire quantitative
No full textInternational audienceWe emphasize an important point in the transformation of sources into data that affects economists and historians alike in every era: the theoretical framework they share. On a broad scale, three major moments can be distinguished during the twentieth century in terms of the general conceptual framework for economic data, which we shall call the price era, the quantity era and the individual era. In the first, economic activity is observed primarily through prices, which is justified by the preponderance of price theory. In the second, the organized measurement of quantities is produced by national accounting systems within the framework of Keynesian macroeconomic theory, and then of growth theory. Criticism of these earlier frameworks is now leading to experimentation with a very large number of measures, characterized by the sheer volume of information analyzed, greater vagueness between sources and data, and a multiplicity of conceptual frameworks. While each of these three stages represents a break with the previous ones, they nonetheless retain much of what has already been achieved.Nous voudrions ici souligner un point important de la transformation des sources en données qui, à chaque époque, affecte les économistes mais aussi les historiens : le cadre théorique qu'ils partagent. A large échelle, trois grands moments peuvent être distingués au cours du XXe siècle en ce qui concerne le cadre conceptuel général en matière de données économiques, moments que nous appellerons l'ère des prix, l'ère des quantités et l'ère des individus. Dans la première, l'activité économique est observée principalement à travers les prix, ce qui est justifié par la prépondérance de la théorie des prix. Dans la seconde, la mesure organisée des quantités est produite par les comptabilités nationales dans le cadre de la théorie macroéconomique dite keynésienne, puis de la théorie de la croissance. La critique de ces cadres antérieurs conduit aujourd'hui à l'expérimentation d'un très grand nombre de mesures se caractérisant par les masses d'information analysées, un flou plus grand entre sources et données, et une multiplicité de cadres conceptuels. Ces trois étapes, si elles veulent rompre chacun avec les précédentes, n'en conservent pas moins une grande partie de leurs acquis
A coupled approach to compute approximate solutions of a compressible immiscible three-phase flow model with fast transient and stiff source terms
No full textInternational audienceThis paper aims at developing a new numerical coupled approach to compute solutions of a compressible immiscible three-phase flow model with stiff source terms. The targeted applications involve flows with fast transient and shock waves. Thus, a well-posed model with respect to the initial conditions that embarks an entropy inequality is considered. A preliminary work on the underlying relaxation process of the model is conducted. Then the new numerical scheme is presented and numerically tested
Ensemble learning–based method for multiple sclerosis screening from retinal OCT images
No full textInternational audienceMultiple Sclerosis (MS) is a neurodegenerative disease that impacts retinal layer thickness. Thus, several works proposed to diagnose MS from the retinal Optical Coherence Tomography (OCT) images. Recent clinical studies affirmed that thinning occurs on the four top layers, explicitly in the macular region. However, existing MS detection methods have not considered all MS symptoms, which may impact the MS detection performance. In this research, we propose a new automated method to detect MS from the retinal OCT images. The main principle is based on extracting the relevant retinal layers and figure out the layer thicknesses, which are investigated to deduce the MS disease. The main challenge is to guaranty a higher performance biomarker extraction within an efficient exploration of OCT-cuts. Our contribution consists of: (1) employing two DLarchitectures to segment separately sub-images based on their morphology, in order to enhance segmentation quality; (2) extracting thickness features from the four top layers; (3) dedicate a classifier for each OCT-cuts that are selected based on their position with respect to the macula center, and (4) merge the classifier knowledge through an ensemble learning approach. Our suggested method achieved a mean accuracy of 97%, 100% of sensitivity, and 94% of precision and specificity, which outperforms several state-of-arts methods
Implicit automatic differentiation and implicit layers for constitutive modeling
No full textInternational audienceThe intersection of computational mechanics and machine learning is advancing rapidly, particularly in the domain of constitutive modeling. Constitutive models, being inherently data-driven, share structural similarities with neural network layers. Beyond data-driven learning, there is significant potential in adapting traditional constitutive models for seamless integration with machine learning frameworks.Automatic differentiation (AD) has revolutionized machine learning by enabling the computation of derivatives for complex functions constructed from simpler operations. In this talk, we will explore how AD can enhance constitutive modeling, focusing on two critical applications: efficient tangent operator computation and material parameter sensitivity.First, AD can automate the derivation of consistent tangent operators, which are crucial for the efficient integration of constitutive models into large-scale simulations. We will examine strategies for obtaining these operators, including directly applying AD to unrolled algorithms and leveraging implicit differentiation for greater efficiency. Additionally, AD enables the computation of material parameter sensitivities, facilitating more effective model calibration and optimization.Finally, we will discuss a novel architecture for data-driven constitutive models that employs implicit layers. This approach departs from conventional large-scale feed-forward neural networks, aiming to develop machine-learning-based constitutive models that adhere to thermodynamic principles, integrate domain knowledge, and demonstrate superior performance on unseen data. These models not only align with established theoretical frameworks but are also easier to train and generalize effectively
INFRASTRUCTURAL EXTENSIONS: Rethinking Infrastructure in Urban Studies
No full textInternational audienceThis essay explores how contemporary urban infrastructure is being conceptually and operationally extended into new domains. Across five key arenas—elemental, care, more‐than‐human, cyber‐physical and the neurotechnical—we trace how infrastructures are no longer confined to traditional networked systems but instead permeate and co‐compose atmospheres, bodies, ecologies and cognition. These extensions reconfigure what infrastructure is and does, operating as agents of socio‐technical modulation under conditions of planetary crisis. We argue that these emergent infrastructural forms constitute a significant shift, where the capacity to live, sense, feel and adapt is increasingly subject to infrastructural capture, stratification and control. The essay reflects on the risks of conceptual overreach while asserting that a more expansive understanding of infrastructure is necessary to grasp contemporary urban governance. We conclude by proposing a renewed research agenda for urban infrastructure studies that is interdisciplinary, ethically attuned and responsive to the operationalization of life itself under conditions of turbulence and inequality
Tropical Cirrus Lifetime Estimated From Superpressure Balloon‐Borne Lidar Observations
No full textInternational audienceTropical tropopause layer (TTL) cirrus clouds play a key role in the Earth climate system. Yet the relative role of the various processes shaping them remains poorly known. Characterizing the temporal evolution of cloudy structures from observations is essential to address this issue but represents a challenge. Indeed, space‐ and airborne platforms move fast and mainly provide instantaneous snapshots. In boreal winter 2021–2022, two balloon‐borne lidars flew over the Equatorial Pacific Ocean, slowly drifting above the clouds. We use those unique nighttime observations to quantify the distribution of TTL cloud lifetime above this homogeneous region. This distribution is strongly asymmetric: half of the clouds live less than 1 hr, but their mean lifetime is about 6 hr. The few long‐lived clouds ( hr) dominate the cloud cover. Those results compare reasonably well with TTL cirrus lifetimes in the ERA5 reanalysis, although the modeled TTL cloud cover is largely underestimated
Equivalence by coupling for heterogeneous SIS models*
No full textWe consider the optimal allocation of (perfect) vaccine in an heterogeneous SIS model. Using a coupling approach, we explain how different models for the heterogeneity of the population lead to the same Pareto frontier in the cost/loss valuation of the vaccinations strategies. This covers in particular the elementary continuous representation of discrete models and the measure preserving transformation which appears in graphon theory
Obsolète et dysfonctionnelle ? : la fabrique technique et institutionnelle de la contre-performance de la ligne C du RER
No full textThe RER C line is one of the oldest and most essential lines in the Île-de-France rail network, transporting 540,000 passengers daily. These passengers face a high level of dysfunction, with frequent delays, breakdowns, and train cancellations, a phenomenon that seems to worsen over time. Despite shared acknowledgment of these issues, the stakeholders involved in the line’s operation and management struggle to agree on effective solutions or on the actions to implement. This thesis thus addresses the question of adapting this metropolitan rail infrastructure to social, territorial, institutional, political, and environmental frameworks that have radically evolved since its creation. It particularly examines how different stakeholders understand and engage with the dysfunction issue to advocate for and implement strategies and policies aligned with their interests.The thesis demonstrates that dysfunction on the RER C takes multiple forms and must be understood in the long-term context of rail network development in Île-de-France. It analyzes dysfunction from two perspectives: infrastructure and service. From the infrastructure perspective, the analysis focuses on the issues of obsolescence and aging, which have become central since the 2000s, following a drop in punctuality and the Brétigny-sur-Orge rail accident in 2013. The thesis explores the notion of rail regeneration, proposed by SNCF Réseau as a solution, which, far from being merely a series of technical operations, proves to be a multidimensional phenomenon that reorganizes maintenance practices and timelines, power relations, and rail infrastructure governance.On the service side, the thesis addresses dysfunction through the lens of the line’s service structure, a subject of numerous blockages for the RER C since the early 2000s. It examines the question of reconciling socio-territorial and techno-economic logics, which the operator must take into account. This is explored by tracing the evolution of service supply and demand over the long term, focusing on planning documents and the justifications put forward to try to overcome these blockages. Our analysis thus highlights various forms of resistance to change.The theoretical framework of the thesis combines Science & Technology Studies, specifically the approach of large socio-technical systems (LTS), with public policy literature on network governance and failure studies. The analysis combines a diachronic and synchronic study of supply and demand data with discourse analysis (semi-structured interviews, press reviews, expert reports, minutes, presentations). Conducted within a CIFRE framework at Transilien, this research contributes to understanding the challenges posed by aging metropolitan rail lines, particularly in a constantly evolving metropolitan contextLa ligne C du RER est l'une des lignes les plus anciennes et structurantes du réseau ferré francilien. Elle transporte chaque jour 540 000 voyageurs. Ces derniers sont confrontés à un niveau de dysfonctionnement élevé où pannes, retards et suppressions de trains sont récurrents sinon quotidiens, un phénomène qui semble s’aggraver avec le temps. Malgré le constat partagé de ces difficultés, les acteurs impliqués dans l’exploitation et la gestion de la ligne peinent à s'accorder sur des solutions efficaces voire sur les actions à mettre en œuvre. Ce travail de thèse aborde ainsi la question de l’adaptation nécessaire de cette infrastructure ferroviaire métropolitaine à des cadres sociaux, territoriaux, institutionnels, politiques et environnementaux qui ont radicalement évolué depuis sa création. Il étudie plus particulièrement la manière dont les différents acteurs comprennent et se saisissent de la question du dysfonctionnement pour défendre et mettre en œuvre des stratégies et politiques en accord avec leurs intérêts. La thèse montre que le dysfonctionnement sur le RER C prend des formes plurielles et se comprend dans le temps long qu’est celui du développement des réseaux ferrés en Île-de-France. Elle analyse le dysfonctionnement depuis deux strates du réseau : l’infrastructure et le service. Du côté de l’infrastructure, l’analyse s'intéresse à la question de l’obsolescence et du vieillissement, problématique devenue centrale depuis les années 2000 avec l’accélération de la baisse de la régularité et l’accident ferroviaire de Brétigny-sur-Orge en 2013. La thèse s’intéresse à la notion de régénération ferroviaire, réponse proposée par SNCF Réseau, une solution qui, loin d’être seulement un ensemble d’opérations techniques, s'avère être un phénomène multidimensionnel qui réorganise les temporalités et pratiques de maintenance, les relations de pouvoir et la gouvernance de l’infrastructure ferroviaire. Pour le service, la thèse aborde le dysfonctionnement à travers le prisme de la desserte, sujet de nombreux blocages pour le RER C depuis le début des années 2000. Elle examine la question de la conciliation entre logique socio-territoriale et technico-économique que l’opérateur est tenu de prendre en compte. Nous étudions la question en retraçant les évolutions de l’offre et de la demande dans le temps long, en nous intéressant à la production des documents de planification et aux justifications avancées pour tenter de dépasser ces blocages. Notre analyse met ainsi en exergue un certain nombre de formes de résistances au changement. Le cadre théorique de la thèse repose sur un croisement entre les Science & Technology Studies, notamment l'approche des grands systèmes socio-techniques (LTS), et une littérature issue des politiques publiques, autour des thèmes de la gouvernance des réseaux et des failure studies. L'analyse articule une étude diachronique et synchronique des données d’offre et de demande avec une analyse de discours (entretiens semi-directifs, revue de presse, rapports d’expertise, comptes rendus, présentations). Réalisée dans le cadre d’une CIFRE chez Transilien, cette recherche apporte une contribution à la compréhension des défis que posent les lignes ferroviaires métropolitaines vieillissantes, en particulier dans un contexte de métropole en constante transformatio