HAL-Ecole des Ponts ParisTech
Not a member yet
42743 research outputs found
Sort by
Daniel WALDENSTRÖM Richer and More Equal : A New History of Wealth in the West Cambridge, Polity, 2024, 256 pages (Note de lecture)
No full textNote de lecture sur l'ouvrage de Daniel WALDENSTRÖM Richer and More Equal : A New History of Wealth in the West Cambridge, Polity, 2024, 256 page
Theoretical and Numerical Study of the Convergence of Luenberger Observers for a Linearized Water Wave Model
No full textThis paper investigates the convergence properties of Luenberger observers applied to a linearized water wave model. The study is motivated by the challenge of estimating wave dynamics when only partial free surface measurements are available. We identify fundamental obstructions to convergence, showing that the classical Luenberger observer fails to achieve full-state reconstruction due to challenges associated with mean-value modes and high-frequency components.To overcome these limitations, we introduce modified observer schemes that incorporate frequency filtering and projection techniques. Our theoretical results are reinforced by numerical experiments that demonstrate the practical effectiveness of these observer-based estimation methods for water waves
A novel sea surface evaporation scheme assessed by the thermal rotating shallow water model
No full textInternational audienceAbstract In this study, a novel sea surface evaporation scheme, along with its corresponding bulk aerodynamic formulation, is proposed to estimate sea surface evaporation, columnar humidity, and precipitation distribution within the atmosphere. The scheme is based on three distinct functions, each dependent on a single variable: zonal wind velocity, tropospheric (potential) temperature, and free convection. It is shown that the normalized Clausius–Clapeyron formula requires an adjustable scaling factor for real‐world applications, calibrated using empirical fitness curves. To validate the proposed approach, we employ a model based on the pseudo‐spectral moist‐convective thermal rotating shallow water model, with minimal parameterization over the entire sphere. ECMWF Reanalysis 5th Generation (ERA5) reanalysis data are used to compare the model's results with observations. The model is tested across different seasons to assess its reliability under various weather conditions. The Dedalus algorithm, which handles spin‐weighted spherical harmonics, is employed to address the pseudo‐spectral problem‐solving tasks of the model
Mobility capacities and smartphone use of students in Kinshasa, Democratic Republic of Congo
No full textInternational audienceMany African cities have been experiencing a digital transformation over the past few years. As people become more familiar with digital tools, particularly smartphones, in their daily lives, their uses and practices in terms of mobility are also evolving. This paper aims to explore the impact of smartphones and digital platforms on mobility capacities by targeting students at the University of Kinshasa (UNIKIN). The methodological approach combines observations and semi-structured interviews with fifty-two students, as well as fifteen experts involved in the field of mobility and transport. The results of our study reveal that students in Kinshasa have constantly adapted complex mobility behaviour, that result from challenging transport conditions and relatively high insecurity in public space. In contrast to observations in other African cities, this study reveals limited adoption of digital tools and navigation applications among the students, mainly due to the lack of digitally-enabled transport services, their high cost, and the context of perceived insecurity in public space
Landscapes—a lens for assessing sustainability
No full textInternational audienceThere are urgent calls to transition society to more sustainable trajectories, at scales ranging from local to global. Landscape sustainability (LS), or the capacity for landscapes to provide equitable access to ecosystem services essential for human wellbeing for both current and future generations, provides an operational approach to monitor these transitions. However, the complexity of landscapes complicates how and what to consider when assessing LS.Objectives: To identify important features of landscapes that remain challenging to consider in LS assessments and provide guidance to strengthen future assessments.Methods: We conducted two workshops to identify the complex features of landscapes that remain under-considered in LS assessments, and developed guidelines on how to better incorporate these features.Results: We identify open and connected boundaries and diversity of values as landscape features that must be better considered in LS assessments or risk exacerbating offstage sustainability burdens and power inequalities. We provide guidelines to avoid these pitfalls which emphasize assessing ecosystem service interactions across interconnected landscapes and incorporating local actors’ diverse values.Conclusions: Our guidelines provide a stepping stone for researchers and practitioners to better incorporate landscape complexities into LS assessments to inform landscape-level decisions and actions
High‐Resolution Downscaling of Disposable Income in Europe Using Open‐Source Data
No full textInternational audienceIncome maps have been extensively used for identifying populations vulnerable to global changes. The frequency and intensity of extreme events are likely to increase in coming years as a result of climate change. In this context, several studies have hypothesized that the economic and social impact of extreme events depend on income. However, to rigorously test this hypothesis, fine‐scale spatial income data is needed, compatible with the analysis of extreme climatic events. To produce reliable high‐resolution income data, we have developed an innovative machine learning framework, that we applied to produce a European 1 km‐gridded data set of per capita disposable income for 2015. This data set was generated by downscaling income data available for more than 120,000 administrative units. Our learning framework showed high accuracy levels, and performed better or equally than other existing approaches used in the literature for downscaling income. It also yielded better results for the estimation of spatial inequality within administrative units. Using SHAP values, we explored the contribution of the model predictors to income predictions and found that, in addition to geographic predictors, distance to public transport or nighttime light intensity were key drivers of income predictions. More broadly, this data set offers an opportunity to explore the relationships between economic inequality and environmental degradation in health, adaptation or urban planning sectors. It can also facilitate the development of future income maps that align with the Shared Socioeconomic Pathways, and ultimately enable the assessment of future climate risks
Access pricing and regulation in international rail transport
No full textWe study a model of non-cooperative interaction between two infrastructure managers (IMs) for international rail transport. We compare equilibrium access charges when the IMs are unregulated and regulated. We show that cooperation among IMs eliminates double-marginalization to the benefit of passengers and IMs. We also show that the delegation of access charge collection with adequate transfers allows the two IMs to reach efficiency, both in the unregulated and regulated régimes. We study the effect of differences in regulatory policies, and analyze the effect of monopoly power of train operators and competition among high speed and low speed train routes on access charges
Open repositories cannot ignore retractions and corrections
No full textLSE Impact blogOpen repositories were designed to enhance access to and visibility of academic outputs, offering a vital alternative to paywalled journals. Yet, their primary focus has remained on dissemination, with little attention paid to maintaining the accuracy and reliability of the scholarly record when errors or misconduct are discovered in published research.In those (rare) cases, journals issue editorial notices like expressions of concern, corrections, additions, errata, corrigenda, withdrawals, and retractions (considered as the most severe measure). Those editorial notices warn readers of issues with prior findings and also promote transparency around corrections to uphold the integrity and reliability of the scholarly record.My research sheds light on this significant issue: most repositories fail to update the status of publications that have been corrected or retracted after being deposited. A manually verified analysis of HAL, one of the world’s largest institutional repositories, revealed that 91% of retracted or corrected publications lacked any indication of their updated status. This glaring oversight leaves researchers and the public vulnerable to citing or relying on invalidated studies.Link to blog post</a
Optimizing Predictive Maintenance in Vehicular Systems via Positive Influence Dominating Sets
No full textInternational audienceModern vehicular systems consist of interconnected components forming complex networks of dependencies. Failures in one component can propagate to others, creating significant challenges for predictive maintenance. Addressing these challenges requires strategies that balance inspection costs with system reliability. This paper introduces an approach based on graph theory to model the dependency relationships between the engine components. By utilizing the Positive Influence Dominating Set (PIDS) concept, we identify critical components whose maintenance ensures comprehensive system coverage. Several algorithms are evaluated over the proposed model, examining trade-offs between accuracy and computational efficiency. The results provide valuable insights into algorithm selection and demonstrate the model's potential to optimize maintenance scheduling for complex vehicular systems
Chapter-Llama: Efficient Chaptering in Hour-Long Videos with LLMs
No full textCVPR 2025 Camera ready. Project page: https://imagine.enpc.fr/~lucas.ventura/chapter-llama/International audienceWe address the task of video chaptering, i.e., partitioning a long video timeline into semantic units and generating corresponding chapter titles. While relatively underexplored, automatic chaptering has the potential to enable efficient navigation and content retrieval in long-form videos. In this paper, we achieve strong chaptering performance on hour-long videos by efficiently addressing the problem in the text domain with our 'Chapter-Llama' framework. Specifically, we leverage a pretrained large language model (LLM) with large context window, and feed as input (i) speech transcripts and (ii) captions describing video frames, along with their respective timestamps. Given the inefficiency of exhaustively captioning all frames, we propose a lightweight speech-guided frame selection strategy based on speech transcript content, and experimentally demonstrate remarkable advantages. We train the LLM to output timestamps for the chapter boundaries, as well as free-form chapter titles. This simple yet powerful approach scales to processing one-hour long videos in a single forward pass. Our results demonstrate substantial improvements (e.g., 45.3 vs 26.7 F1 score) over the state of the art on the recent VidChapters-7M benchmark. To promote further research, we release our code and models at our project page