HAL Paris Dauphine-PSL
Not a member yet
21932 research outputs found
Sort by
Strong propagation of chaos for systems of interacting particles with nearly stable jumps
No full textInternational audienc
A probabilistic view on Riemannian machine learning models for SPD matrices
No full textInternational audienceThe goal of this paper is to show how different machine learning tools on the Riemannian manifold P_d of Symmetric Positive Definite (SPD) matrices can be united under a probabilistic framework. For this, we will need several Gaussian distributions defined on P_d . We will show how popular classifiers on P_d can be reinterpreted as Bayes Classifiers using these Gaussian distributions. These distributions will also be used for outlier detection and dimension reduction. By showing that those distributions are pervasive in the tools used on P_d , we allow for other machine learning tools to be extended to P_d
Les principes non contraignants de l'Organisation des Nations Unies relatifs aux entreprises et aux droits de l'homme et les principes de l'OCDE, source d'inspiration de la directive
No full textInternational audienc
Contre l'amnésie. Les sciences administratives de langue française de Saint-Simon à nos jours
No full textFace à l’hégémonie croissante des travaux de langue anglaise et à une ignorance grandissante de la richesse de notre tradition et des travaux en sciences sociales dans notre propre champ linguistique, l’objet de ce volume vise à présenter à partir d’un regard personnel la genèse des sciences administratives depuis les premières réflexions menées au début du XIXe siècle par Saint-Simon, d’en souligner les apports, l’originalité, l’influence, et les principaux défis auxquels il est aujourd’hui confronté.Le lectorat visé est toute personne qui s’intéresse à l’action organisée. En découvrant ou en redécouvrant les particularités du champ de langue française, en accédant à une riche bibliographie, nous espérons que cet ouvrage permettra au lectorat visé d’alimenter sa réflexion présente et future dans un dialogue fécond avec son histoire, et les riches contributions que l’on retrouve dans d’autres champs linguistiques, tant au Nord qu’au Sud
Bayesian phylogenies as a tool to explore linguistic and cultural diversity: The case of Kra-Dai languages and the evolution of loom technology
No full textInternational audienc
The Boltzmann equation on smooth and cylindrical domains with Maxwell boundary conditions
No full textIn this article we study the well-posedness of the Boltzmann equation near its hydrodynamic limit on a bounded domain. We consider two types of domains, namely domains with Maxwell boundary conditions where the accommodation coefficient is a continuous space dependent function for any , or cylindrical domains with diffusive reflection on the bases of the cylinder and specular reflection on the rest of the boundary. Furthermore, we work with polynomial, stretched exponential and inverse gaussian weights to construct the Cauchy theory near the equilibrium. We remark that all methods are quantitative thus all the constants are constructive and tractable
Interventions Publiques dans la Zone euro et en France : Trois essais
No full textThis thesis delves into the evolution and efficacy of public intervention tools within the Euro Area (EA) and France, set against a backdrop of incomplete integration and recurrent economic crises. Employing various empirical approaches, this thesis seeks to explore public intervention tools from three distinct perspectives, aiming to address the following questions: How have these tools evolved within the context of imperfect economic integration and recurrent crises? What impact have they had on financial markets, the real economy, and public finances? Chapter 1 examines the effectiveness of joint EU fiscal plans and the European Central Bank's (ECB) monetary policies during the COVID-19 pandemic. It reveals that while the ECB has traditionally been the preferred institution to manage yield divergence within the EA, monetary announcements alone proved insufficient during the crisis. Instead, joint fiscal tools at the EU level, particularly beneficial for member states like Italy, were crucial in mitigating refinancing risks. This underscores the essential role of coordinated fiscal interventions alongside monetary policies to address acute economic challenges effectively. Chapter 2 assesses the impact of the ECB's Pandemic Emergency Purchase Programme (PEPP) on Euro Area listed corporations. It demonstrates that although the program significantly boosted stock prices and economic outcomes, the benefits were not uniformly distributed, primarily favoring 'Eligible' firms. This uneven distribution triggers further consideration of the potential distortions created by such interventions and their implications for market equity and competition. Chapter 3 introduces a comprehensive database of French tax expenditures from 1979 to 2022, offering unprecedented insights into their integration within the national fiscal policy framework and their growing influence on public finances. Although these tools have played an increasingly important role in France, the chapter also raises concerns about the overall cost associated with these devices, prompting further research into their ability to achieve the intended policy outcomes. Overall, the thesis concludes that while public intervention tools have evolved to address immediate financial and economic challenges effectively, they also introduce new considerations and challenges.Cette thèse explore l'évolution récente et l'efficacité des outils d'intervention publique au sein de la Zone Euro (EA) et en France, dans un contexte d'intégration incomplète et de crises économiques récurrentes. En adoptant diverses approches empiriques, cette thèse vise à explorer les outils d'intervention publique sous trois perspectives distinctes, cherchant à répondre aux questions suivantes : Comment ces outils ont-ils évolué dans le contexte d'une intégration économique imparfaite et de crises récurrentes ? Quel impact ont-ils eu sur les marchés financiers, l'économie réelle ou encore les finances publiques ? Le premier chapitre examine l'efficacité des annonces liés aux plans fiscaux conjoints de l'UE et des politiques monétaires de la Banque Centrale Européenne (BCE) durant la pandémie de COVID-19. Il révèle que, bien que la BCE ait traditionnellement été l'institution privilégiée pour gérer la divergence des rendements au sein de l'EA, les annonces monétaires seules se sont avérées insuffisantes lors de la crise. En effet, les nouveaux outils fiscaux conjoints (comme le NGEU), particulièrement bénéfiques pour des États membres comme l'Italie, ont été cruciaux pour atténuer les risques de refinancement. Ceci souligne le rôle essentiel des interventions fiscales coordonnées au niveau européen en complément des politiques monétaires pour répondre efficacement aux crises. Le deuxième chapitre évalue l'impact du Pandemic Emergency Purchase Programme (PEPP) de la BCE sur les entreprises cotées de la Zone Euro. Il démontre que bien que le programme ait significativement stimulé les prix des actions et les résultats économiques, les bénéfices n'ont pas été uniformément distribués, favorisant principalement les entreprises 'Éligibles'. Cette distribution inégale incite à réfléchir davantage sur les distorsions potentielles créées par de telles interventions et leurs implications sur l'équité et la concurrence. Le troisième chapitre présente une base de données inédite sur les dépenses fiscales françaises de 1979 à 2022, offrant des perspectives sur leur intégration au sein du système global des prélèvements et dépenses publics national et leur influence croissante sur les finances publiques. Bien que ces outils aient joué un rôle de plus en plus important en France, le chapitre soulève également des inquiétudes concernant le coût global associé à ces dispositifs, incitant à des recherches supplémentaires portant sur leurs capacités à atteindre les résultats politiques visés. En conclusion, la thèse constate que si les outils d'intervention publique ont évolué pour répondre efficacement aux défis financiers et économiques immédiats, ils introduisent également de nouvelles considérations et inquiétudes
Limit theory for mean-field control problems with common noise adapted controls
No full textInternational audienceWe consider a mean-field control problem in which admissible controls are required to be adapted to the common noise filtration. The main objective is to show how the mean-field control problem can be approximates by time consistent centralized finite population problems in which the central planner has full information on all agents' states and gives an identical signal to all agents. We also aim at establishing the optimal convergence rate. In a first general path-dependent setting, we only prove convergence by using weak convergence techniques of probability measures on the canonical space. Next, when only the drift coefficient is controlled, we obtain a backward SDE characterization of the value process, based on which a convergence rate is established in terms of the Wasserstein distance between the original measure and the empirical one induced by the particles. It requires Lipschitz continuity conditions in the Wasserstein sense. The convergence rate is optimal. In a Markovian setting and under convexity conditions on the running reward function, we next prove uniqueness of the optimal control and provide regularity results on the value function, and then deduce the optimal weak convergence rate in terms of the number of particles. Finally, we apply these results to the study of a classical optimal control problem with partial observation, leading to an original approximation method by particle systems
Matrices aléatoires : grande dimension et résolution exacte
No full textThis thesis explores certain aspects of exact solvability in random matrix theory and is structured into two main parts. The first part examines a high-dimensional problem concerning the asymptotic behavior of characteristic polynomials of random matrices. We focus on two integrable models. The first is the Elliptic Ginibre Ensemble, a Gaussian interpolation between the Ginibre Ensemble and its Hermitian counterpart, the Gaussian Unitary Ensemble. The second model involves permutation matrices, where the underlying permutation follows the generalized Ewens distribution, in which the measure of a permutation depends only on its cycle structure. For both models, we establish the convergence in law of the characteristic polynomial, as the matrix dimension tends to infinity, towards a random analytic function. This convergence occurs outside the eigenvalue support and complements the convergence of spectral distributions. The second part of the thesis addresses a fixed-dimensional problem. We consider products of unitary matrices that are uniformly distributed on fixed conjugacy orbits and derive the probability density for the eigenvalues of such products. This probability density is related to the volume of the moduli space of flat connections on the three-holed sphere. Our formula provides a positive expression for both the density and this volume, represented as a sum of volumes of explicit polytopes. These polytopes arise from combinatorial objects called puzzles, which compute intersection coefficients for the cohomology of two-step flag varieties. We further investigate some properties of these puzzles.Cette thèse explore certains aspects de la solvabilité exacte en théorie des matrices aléatoires. Elle est structurée en deux parties principales. La première partie traite d’un problème en grande dimension, sur comportement asymptotique des polynômes caractéristiques de matrices aléatoires. Nous nous concentrons sur deux modèles intégrables. Le premier est l’Ensemble de Ginibre Elliptique, une interpolation gaussienne entre l'Ensemble de Ginibre et son homologue hermitien, l'Ensemble Unitaire Gaussien. Le second modèle concerne les matrices de permutation, où la permutation sous-jacente suit la distribution d’Ewens généralisée pour laquelle la mesure d’une permutation dépend uniquement de sa structure en cycles. Pour ces deux modèles, nous établissons la convergence en loi du polynôme caractéristique vers une fonction analytique aléatoire lorsque la dimension des matrices tend vers l’infini. Cette convergence a lieu en dehors du support des valeurs propres et est complémentaire de la convergence des distributions spectrales. La seconde partie concerne un problème en dimension fixée. Nous considérons des produits de matrices unitaires uniformément distribuées sur des orbites de conjugaison. Nous déterminons la densité de probabilité des valeurs propres de ce produit. Cette densité est liée au volume de l’espace des modules des connexions plates sur une sphère à trois trous. Notre formule fournit une expression positive pour la densité et pour ce volume sous la forme d'une somme de volumes de polytopes explicites. Ces polytopes émergent d’objets combinatoires appelés puzzles, permettant de calculer les coefficients d’intersection pour la cohomologie des variétés de drapeaux à deux sous-espaces. Nous explorons également certaines propriétés de ces puzzles
Evidence-based veterinary medicine perception by swine veterinarians: a European survey across diverse practitioner profiles
No full textInternational audienceEvidence-Based Veterinary Medicine (EBVM) enhances the quality of care by guiding clinical decisions with robust scientific data, leading to more effective disease management. Evaluating the levels of knowledge and use is crucial for identifying gaps and training needs, ultimately promoting the adoption of evidence-based practices and optimizing herd health and productivity. This study aimed to give an overview of EBVM perception, with the goal of raising awareness of this concept and highlighting reservations they face in applying it to decision-making process in swine veterinary practice. A cross-sectional online survey was conducted among European swine veterinarians. Respondent characteristics and knowledge of EBVM were analyzed using descriptive statistics. Associations between demographic variables (country, specialization) and EBVM knowledge were assessed using Chi-square tests with Yates' continuity correction where appropriate. A significant level of 0.05 was applied. Results showed that 68% of veterinarians were familiar with EBVM, with knowledge levels varying by country ( p < 0.05) and educational background (p < 0.01). While 82% recognized its practical value, barriers such as limited time, difficulty accessing relevant data, and a lack of decision-support tools hindered implementation. Younger veterinarians and those with less experience expressed more reluctance toward EBVM. Despite these challenges, 90% of respondents believed EBVM improved decision-making, professional confidence, and communication with farmers. However, opinions differed regarding how well EBVM is currently integrated and developed in everyday veterinary practice. These findings highlight the need for enhanced education and structured methodologies to bridge the gap between theoretical knowledge and practical application in herd health management