Portail HAL de l'Institut Polytechnique de Paris
Not a member yet
28935 research outputs found
Sort by
Long-term monitoring through a wastewater-based observatory to model urban population dynamics and health indicators
No full textInternational audienceWastewater-based epidemiology (WBE) has emerged as a powerful tool for monitoring population health, yet its quantitative reliability remains constrained by uncertainties in population estimation and sampling frequency. This study evaluates whether dynamic de facto population estimates can be derived from hydrochemical wastewater parameters and whether monthly wastewater monitoring can be analytically refined to a daily timescale through population normalization. A Bayesian smoothing and hierarchical outlier-detection framework was developed to denoise concentration and flow measurements of standard hydrochemical parameters. From these smoothed flows, a multidimensional inference model based on five hydrochemical parameters was constructed to estimate connected populations at daily resolution and validated using 2020 population benchmarks from two major Parisian wastewater treatment plants. The model achieved mean absolute percentage errors of approximately 6% across sites with contrasting catchment characteristics, outperforming or matching existing literature models and demonstrating robustness to operational disturbances. Applied to four years of data from the City’s wastewater Observatory, the method revealed population variations by a factor of three, substantially improving the interpretation of chemical and virological signals. After normalization, pharmaceutical trends derived from wastewater closely matched prescription data, enabling the detection of non-prescription usage. Conversely, monthly viral monitoring provided only limited correspondence with clinical data, confirming that such low-frequency sampling is insufficient for short-term epidemic tracking. Overall, this work demonstrates that hydrochemical-based population modeling enhances the interpretability, scalability, and operational value of long-term wastewater observatories, providing a practical route to integrate monthly monitoring into finer-scale WBE analyses
Analytical solution of radiative transfer equation of light radiance in turbid slab with inner-medium source under P3-1D approximation
No full textA generalized model of the 1-dimensional radiative transfer equation of the light radiance in a turbid slad is detailed, under the P-3 approximation, including the possibility to model a continuous plane-wave source located at any depth within the scattering slab. This analytical model, which requires significant evolution of the P3-1D model is extensively described and validated by comparison with Monte-Carlo numerical experiments. A series of numerical simulations illustrates some of the modelling possibilities offered by this extended model, which makes it possible to continuously model the transition between a classical slab geometry and a semi-infinite geometry
Homological Algebra in Abelian Framed Bicategories: Exact Sequences and Embedding Theorems
No full textWe introduce abelian framed bicategories, which are particular framed bicategories that are locally abelian, and show that they are suitable for developing homology and cohomology theories for directed structures. This means in particular that similar exact sequences as the relative homology and Mayer-Vietoris long exact sequences can be shown to hold. Also, for closed monoidal abelian framed bicategories, Künneth theorem holds as well. Finally, we prove embedding theorems similar to the Gabriel and Freyd-Mitchell theorems, for particular abelian framed bicategories, allowing to see those as bicategories of bimodules over algebras. This naturally links to the original motivation of this work, which was to generalize directed homology developed in the abelian framed bicategory of bimodules over (path) algebras
On Gossip Algorithms for Machine Learning with Pairwise Objectives
No full textInternational audienceIn the IoT era, information is more and more frequently picked up by connected smart sensors with increasing, though limited, storage, communication and computation abilities. Whether due to privacy constraints or to the structure of the distributed system, the development of statistical learning methods dedicated to data that are shared over a network is now a major issue. Gossip-based algorithms have been developed for the purpose of solving a wide variety of statistical learning tasks, ranging from data aggregation over sensor networks to decentralized multi-agent optimization. Whereas the vast majority of contributions consider situations where the function to be estimated or optimized is a basic average of individual observations, it is the goal of this article to investigate the case where the latter is of pairwise nature, taking the form of a U -statistic of degree two. Motivated by various problems such as similarity learning, ranking or clustering for instance, we revisit gossip algorithms specifically designed for pairwise objective functions and provide a comprehensive theoretical framework for their convergence. This analysis fills a gap in the literature by establishing conditions under which these methods succeed, and by identifying the graph properties that critically affect their efficiency. In particular, a refined analysis of the convergence upper and lower bounds is performed
Investigation du développement des écoulements secondaires dans un conduit rectangulaire courbé tridimensionnel à l'aide de stratégies avancées de modélisation de la turbulence
No full textInternational audienceThe massive flow separation in a three-dimensional S-shaped rectangular air intake is investigated using RANS and hybrid RANS/LES approaches in order to provide insights on the interplay between the corner vortices and the flow separation induced by the curvature of the duct. Experimental data from the ONERA S19Ch wind tunnel as well as a reference Zonal Detached Eddy Simulation (ZDES) mode 3, that is a Wall Modeled Large Eddy Simulation (WMLES) simulation, are employed to improve the analysis and physical interpretation of RANS and hybrid RANS/LES simulation results. It is shown that all RANS calculations, using several non-linear RANS closures including a Reynolds Stress Model (RSM) and recent Quadratic Constitutive Relation (QCR) variants, fail to reproduce the recirculation bubble in the symmetry plane, observed in the experiments and accurately simulated by the WMLES approach. The use of ZDES mode 2 (2020) -which is a hybrid RANS/LES approach where the attached boundary layers are treated in RANS mode-significantly improves the prediction of first and second-order statistics, capturing the salient flow features of the present case including a successful prediction of both the separating/reattaching dynamic and the separation line. An interpretation of the results in terms of Prandtl's secondary flows is presented. The competition between turbulence-driven secondary flows of second kind and curvature-induced secondary flows of first kind is underscored, which also provides a better understanding of the turbulence modeling requirements for the present case
Choosing the appropriate methodology to monitor soil organic carbon (SOC) in croplands: Aligning methods with evolving monitoring reporting verification (MRV) frameworks
No full textSource Agritrop Cirad (https://agritrop.cirad.fr/617065/) * Autres projets (id;sigle;titre): 101059863;ORCaSa;(EU) Operationalising the International Research Cooperation on Soil Carbon//International audienceMonitoring soil organic carbon (SOC) has gained significant recognition, not only for national greenhouse gas inventories but also for voluntary carbon markets and agri-environmental policies. This has amplified the need for accurate, continuous, and cost-effective SOC stock monitoring from field to national scales. Consequently, methodological frameworks have emerged to address these diverse needs. They rely on measurement and/or modeling of the SOC, considering several complexities (Tiers 1, 2 and 3), combined or not with remote sensing. However, practical implementation guidelines for choosing the most suitable monitoring approach for specific contexts and purposes remain incomplete. Therefore, this study analyses current SOC monitoring methodologies for croplands and proposes a decision tree to help MRV stakeholders select the monitoring strategy considered most appropriate for their context. It confirmed that currently there are no fit-for-all-purposes method, while it recommends whenever possible, the use of measure–remeasure or parsimonious Tier 3 modeling approaches with Earth-Observation data assimilation. Indeed, the assimilation of biomass or proxies derived from remote sensing into those models allows better quantification of biomass restitution to the soil and improves model accuracy and scalability at low cost. Finally, we highlight possible improvements in the different monitoring approaches and key challenges that still need to be overcome
Wideband Radiometry From P to S Band for Monitoring Polar Regions
No full textInternational audienceThis article reviews existing and planned contributions of spaceborne microwave radiometry from P to S band to new measurements of key geophysical variables with a particular focus on the polar regions. It summarizes the current state of spaceborne microwave radiometry to measure ice sheet thermal states, sea ice thickness (SIT), salinity, and sea surface salinity (SSS). Then, this article discusses the potential of wideband radiometry, with continuous sampling in the range of 0.4-2 GHz, as a breakthrough for enhancing the estimation of geophysical variables such as SSS and the geothermal heat flux beneath the polar ice sheets, which are currently monitored primarily using L-band radiometry satellites. Furthermore, this article describes opportunities for new unique observations that cannot be achieved with the current constellation of satellite sensors. In addition, this article demonstrates the advantages of using low-frequency radiometry in sensing soil moisture and biomass from space due to the great sensing depth. This article concludes with a discussion of mission concepts highlighting the CryoRad mission, which has been selected as one of the four candidates for European Space Agency Earth Explorer 12 competition and is now conducting Phase 0 feasibility studies, envisions a 0.4–2-GHz dedicated spaceborne radiometer operated with circular polarization
Automated identification of fossil benthic foraminifera from the Peruvian margin using convolutional neural networks
No full textInternational audienceAbstract. Benthic foraminifera tests preserved in marine sediments are well-established proxies for bottom-water dynamics, yet their minute size and high diversity demand laborious manual identification of hundreds of individuals to reconstruct subtle faunal shifts and are prone to observer-dependent taxonomic inconsistencies. The recent advances in image acquisition hardware and image identification software have made it possible to acquire and identify large image datasets quickly. Here, we trained convolutional neural networks (CNNs) to identify benthic foraminifera morphospecies from 31 samples from two sedimentary cores from offshore Peru, spanning the past 18 000 years. Our best-performing model achieves 92 % overall classification accuracy, 93.4 % precision, and 92.4 % recall, enabling high-temporal-resolution reconstructions of benthic foraminifera assemblages along the Peruvian margin. Automated outputs closely matched manual results across 31 samples, from counts and relative abundances to diversity indices, multivariate assemblage patterns, and dissolved oxygen estimates, indicating the suitability of automated identification for paleo-ecological applications. The highest-performing CNN model (trained on a dataset of 5860 images) from this study can be adapted to analyse benthic foraminifera from equivalent depths of the Peruvian margin, providing high-resolution insights into the eastern tropical Pacific oxygen minimum zone (OMZ). In addition to offering a scalable, objective alternative for high-temporal-resolution analysis of benthic foraminifera, this study also highlights the current limitations of automated workflows
Asymptotics of SGD in Sequence-Single Index Models and Single-Layer Attention Networks
No full textWe study the dynamics of stochastic gradient descent (SGD) for a class of sequence models termed Sequence Single-Index (SSI) models, where the target depends on a single direction in input space applied to a sequence of tokens. This setting generalizes classical single-index models to the sequential domain, encompassing simplified one-layer attention architectures. We derive a closed-form expression for the population loss in terms of a pair of sufficient statistics capturing semantic and positional alignment, and characterize the induced high-dimensional SGD dynamics for these coordinates. Our analysis reveals two distinct training phases: escape from uninformative initialization and alignment with the target subspace, and demonstrates how the sequence length and positional encoding influence convergence speed and learning trajectories. These results provide a rigorous and interpretable foundation for understanding how sequential structure in data can be beneficial for learning with attention-based models
Croire en la surgénération. Les transformations d’une utopie nucléaire
No full textInternational audienceSince the 1930s, nuclear specialists have been driven by an idea based on a neutron equation: by converting uranium 238 into plutonium in suitable nuclear reactors, the available resource would be multiplied by 100. This idea, opening the way to a horizon of abundance and material equilibrium, has come to be known as fuel “breeding”. Although it motivated numerous technical and scientific programs, the vast majority of these have since been halted. How can we explain the longevity of the idea of breeding, despite the épreuves it has endured? This article is organized around the hypothesis that breeding is a technical utopia, and examines the different forms that belief in this utopia has taken within expert communities in France since the 1970s. Breeding certainly appears as the culmination of humanity's energy history, in a context where collective choices are structured by the knowledge of technicians - what we call the “major mode” of utopia. But it also appears as an idea in the process of finding material translations - where technical utopia becomes the object of testing, discussion and postponement. Finally, it survives by taking the form of an insurance technology in the event of a supply crisis, whose availability must be maintained - what we refer to as the “minor mode” of utopia.Depuis les années 1930, une idée issue d’une équation neutronique anime les spécialistes du nucléaire : en convertissant l’uranium 238 en plutonium à l’intérieur de réacteurs nucléaires adaptés, on multiplierait par 100 la ressource disponible. Cette idée, ouvrant la voie à un horizon d’abondance et d’équilibre matériel, a pris le nom de « surgénération ». Mais, si elle a motivé de nombreux programmes techniques et scientifiques, ceux-ci ont, dans leur grande majorité, été arrêtés. Comment expliquer la longévité de l’idée de surgénération malgré les épreuves qu’elle a traversées ? Cet article s’organise autour de l’hypothèse que la surgénération est une utopie technique et étudie les différentes formes que la croyance en cette utopie a pris au sein des communautés expertes en France depuis les années 1970. La surgénération apparaît certes comme l’aboutissement de l’histoire énergétique de l’humanité, dans un cadre où les choix collectifs sont structurés par le savoir des techniciens – ce que nous qualifions de « mode majeur » de l’utopie. Mais elle apparaît également comme une idée en passe de trouver des traductions matérielles – où l’utopie technique devient l’objet de mises à l’épreuve, de discussions, et de reports. Enfin, elle survit en prenant la forme d’une technologie d’assurance en cas de crise d’approvisionnement, dont la disponibilité doit être maintenue – ce que nous désignons comme le « mode mineur » de l’utopie