HAL-Ecole des Ponts ParisTech
Not a member yet
42743 research outputs found
Sort by
Les effets de l’immigration sur l’économie
No full textNational audienceLes effets de l’immigration sur l’économie
La fable du « recyclage énergétique » des ordures ménagères : déchets plastiques, crise pétrolière et incinération dans les années 1970 et 1980 en France
No full textInternational audienceDuring the 1970s and 1980s, in the name of environmental conservation and the effort to save energy in the wake of the oil crisis, “thermal recycling” and “energy recycling,” more commonly referred to as the “energy recovery” of household and urban waste, was invented in France. These oxymoronic expressions hide the fact that the waste is incinerated, thereby rendering linear the life cycles of the natural elements from which they are made. From an ideological point of view, the expressions green a technique that was previously recognized as destructive. By identifying the industrial stakeholders and institutions that have promoted this expanding integration of waste, city, and energy, this article aims to challenge a contemporary truism, namely the virtuous nature of using urban waste to produce energy.Dans les années 1970 et 1980, au nom de la protection de l’environnement et de la recherche d’économies d’énergie générée par la crise pétrolière, on assiste à l’invention du « recyclage thermique » ou « énergétique », et plus communément de la « valorisation énergétique », des déchets ménagers et urbains. Ces expressions oxymoriques cachent en vérité leur incinération et donc la linéarisation des cycles des éléments naturels qui les constituent. Sur le plan idéel, elles permettent de verdir une technique reconnue jusque-là comme destructive. En partant à la recherche des acteurs industriels et des institutions qui ont promu une intégration croissante déchets-ville-énergie, cet article vise à dénaturaliser une évidence contemporaine : le caractère vertueux de l’utilisation des déchets urbains à fin de production énergétique
Résultats d'approximation et de régularité pour le modèle d'Heston et processus associés
No full textThis thesis investigates approximations and regularity results pertaining to Heston’s stochastic volatility model. It comprises three papers.The first one tackles the challenge of developing high-order weak approximations for the Cox-Ingersoll-Ross (CIR) process, which is crucial in financial modelling. The standard Euler-Maruyama scheme fails due to the square root in the diffusion term, potentially leading to negative values. Additionally, theoretical frameworks that produce high-order approximations, like the Multistep Richardson Romberg approach (Pagès 2007), do not directly apply due to the CIR process’s specific structure. This work employs the random grid technique by Alfonsi and Bally (2021), which leverages an elementary scheme on random time grids to boost convergence order. We use Alfonsi’s (2010) second-order CIR scheme as the elementary building block. Rigorous proofs establish that weak approximations of any order can be achieved for smooth test functions with polynomial growthderivatives, given the condition σ^2 ≤ 4a that is less restrictive than the well-known “Feller Condition” (σ^2 ≤ 2a). Numerical experiments validate these findings, showcasing convergence for both CIR and Heston models, with significant computational time improvements compared to lower-order schemes. The limitation lies in the theoretical results being proven only under the less restrictivecondition cited above. Numerical tests hint at effectiveness beyond this, but rigorous proof remains an open question.The second work serves as a continuation of the first one. In this iteration, we apply our technique, which is based on random grids, to the log-Heston process. This process represents the logarithm of an asset price within the Heston model and the associated volatility process. The log transformation helps ensure that the moments of both variables remain bounded, thereby simplifying the mathematical analysis. We start by proposing two second order schemes, built using splitting techniques. The first one uses an exact simulation for the volatility process; the other one uses the Ninomiya-Victoir splitting scheme for the log-Heston process (this one is valid only under the above-cited condition on parameters σ^2 ≤ 4a). Rigorous proofs demonstrate convergence to any desired order. Numerical experiments validate the theoretical results, showcasing the effectiveness of the high-order schemes for pricing European and arithmetic Asian options. The impact of different coupling choices on estimator variance is also investigated. Additionally, promising results are presented for the multifactor/rough Heston model, suggesting the potential of the random gridtechnique in this extended context.The last work delves into the partial differential equation (PDE) associated with the log-Heston model, exploring classical and viscosity solutions. Key contributions include extending classical solution results by incorporating linear and source terms in the PDE. In this work, we also prove the existence and uniqueness of viscosity solutions without relying on Feller’s condition, a common assumption in the literature. Uniqueness is established even for initial data with specific discontinuities, which is relevant for financial applications like digital option pricing. Furthermore, wedemonstrate the convergence of a hybrid numerical scheme to approximate the viscosity solution under relaxed regularity assumptions (continuity) on the initialdata. These results offer a more comprehensive understanding of the log-Heston PDE, particularly in scenarios where Feller’s condition doesn’t hold or the initial data is discontinuous. In the end, we prove a convergence result for a hybrid scheme provided that the initial data is just continuous.In Appendix, we collect other results for the CIR process that did not find space in these three articles.Cette thèse étudie les approximations et les résultats de régularité pour le modèle de volatilité stochastique de Heston. Des approximations faibles d'ordre élevé pour le processus CIR sont développées en utilisant une technique de grille aléatoire. Cette méthode surmonte les limitations des schémas standards. Elle permet d'obtenir une convergence de tout ordre pour des fonctions test régulières avec des dérivées de croissance polynomiales, sous une condition moins stricte que la fameuse condition de Feller. Des expériences numériques confirment ces résultats pour les modèles CIR et aussi Heston, démontrant des avantages significatifs en termes de calcul. L'approche de la grille aléatoire est ensuite étendue au processus log-Heston. Deux schémas de second ordre sont proposés, utilisant les techniques de splitting du générateur infinitésimal et la simulation exacte. Des preuves rigoureuses démontrent une convergence d'ordre élevé et des expériences numériques valident ces résultats pour l'évaluation de diverses options, y compris les options européennes et asiatiques. Enfin, la thèse analyse l'équation aux dérivées partielles (EDP) associée au modèle log-Heston. Les résultats classiques de solution sont étendus, et l'existence et l'unicité des solutions de viscosité sont prouvées sans dépendre de la condition de Feller. L'unicité est établie même pour des données initiales discontinues, ce qui est pertinent pour les applications financières telles que l'évaluation des options numériques. Un schéma numérique hybride converge vers la solution de viscosité sous des hypothèses de régularité relâchées
Reflectivity Validation of EarthCARE Cloud Profiling Radarreflectivity using ACTRIS ground-based Cloud Radar Network
No full textInternational audienceEarthCARE's reflectivity validationCalibrated ACTRIS sites (used in this study) Why use the ACTRIS network ? 25 ground based sites: multiinstrument, large geographical coverage, cloud type variability. CCRES expert centre: Standard Operation Procedures</div
pH and time dependent microstructure features of compacted bentonite-sand mixture
No full textInternational audienceCompacted bentonite-sand mixture is considered as a potential sealing material for deep geological disposal of high-level radioactive waste. In the long run, the intruding alkaline plumes, as a result of concrete degradation, can alter the microstructure of the mixture, potentially affecting the sealing performance. In this study, Compacted bentonite-sand (40/60) mixture samples were hydrated by deionised water and alkaline solutions with varying pHs (8.8, 11.5, 13.5) for different durations under constant-volume conditions. The resulting microstructural changes were subsequently investigated by Mercury Intrusion Porosimetry (MIP) and Scanning Electron Microscope (SEM). Results showed that after hydration for one week, for the sample hydrated with deionised water, a tri-modal porosity was observed, defining different populations for intra-aggregate pores, inter-particle pores, and inter-aggregate pores. With the increase of pH, the intruded void ratio was significantly reduced, mainly due to the shrinkage of the diffuse double layer. The uniformity increased for the intra-aggregate pores while decreased for the meso and macro pores (inter-particle and inter-aggregate pores). Additionally, in the case of pH = 13.5, the intruded void ratio of the samples decreased over time, with further increase of void ratio and pore uniformity for the intra-aggregate pores and the decrease of size for the macropore population. These phenomena suggest that the alkaline attack resulted in montmorillonite dissolution, further modifying the macro-pores. This is confirmed by the SEM observations
Training-Free Synthetic Data Generation with Dual IP-Adapter Guidance
No full textInternational audienceFew-shot image classification remains challenging due to the limited availability of labeled examples. Recent approaches have explored generating synthetic training data using text-to-image diffusion models, but often require extensive model fine-tuning or external information sources. We present a novel training-free approach, called DIPSY, that leverages IP-Adapter for image-to-image translation to generate highly discriminative synthetic images using only the available few-shot examples. DIPSY introduces three key innovations: (1) an extended classifier-free guidance scheme that enables independent control over positive and negative image conditioning; (2) a class similarity-based sampling strategy that identifies effective contrastive examples; and (3) a simple yet effective pipeline that requires no model fine-tuning or external captioning and filtering. Experiments across ten benchmark datasets demonstrate that our approach achieves state-of-the-art or comparable performance, while eliminating the need for generative model adaptation or reliance on external tools for caption generation and image filtering. Our results highlight the effectiveness of leveraging dual image prompting with positive-negative guidance for generating class-discriminative features, particularly for fine-grained classification tasks
CoMMA, a Large-scale Corpus of Multilingual Medieval Archives
No full textWe present CoMMA, a large-scale corpus of medieval manuscripts produced through automatic text recognition. The corpus contains around 3.3b tokens drawn from more than 32,700 digitized manuscripts in Latin and Old French, harvested via IIIF. Unlike other resources, it is made of raw, non-normalized text enriched with layout analysis in various formats. We describe the pipeline used for large-scale acquisition and processing, and report quantitative and qualitative evaluations (average CER 9.7%). The resulting resource supports multiple use cases, from pretraining language models to corpus linguistic on historical languages and digital humanities applications
Relier la pression chimique à ses effets sur la biodiversité dans les eaux
No full textInternational audienc
Retrait-gonflement des argiles : un risque de plus en plus important
No full textInternational audienc
Combinatorics of rectangulations: old and new bijections
No full textInternational audienceA rectangulation is a decomposition of a rectangle into finitely many rectangles. Via natural equivalence relations, rectangulations can be seen as combinatorial objects with a rich structure, with links to lattice congruences, flip graphs, polytopes, lattice paths, Hopf algebras, etc. In this paper, we first revisit the structure of the respective equivalence classes: weak rectangulations that preserve rectangle-segment adjacencies, and strong rectangulations that preserve rectangle-rectangle adjacencies. We thoroughly investigate posets defined by adjacency in rectangulations of both kinds, and unify and simplify known bijections between rectangulations and permutation classes. This yields a uniform treatment of mappings between permutations and rectangulations that unifies the results from earlier contributions, and emphasizes parallels and differences between the weak and the strong cases. Then, we consider guillotine rectangulations, and prove that they can be characterized -- under all known mappings between permutations and rectangulations -- by avoidance of two mesh patterns that correspond to windmills in rectangulations. This yields new permutation classes in bijection with weak guillotine rectangulations, and the first known permutation class in bijection with strong guillotine rectangulations. Finally, we address enumerative issues and prove asymptotic bounds for several families of strong rectangulations