HAL Université de Savoie
Not a member yet
59053 research outputs found
Sort by
Assessing Geological Hazards in a Changing World Through Regional Multidisciplinary Approaches to European Glacial Lakes (Northern Pyrenees, Northern and Western Alps)
No full textInternational audienceThis study combines a multidisciplinary approach to Pyrenean and Alpine glacial lakes to characterize the sensitivity of Late Glacial to Holocene subaquatic flood deposits in deltaic environments to slope failures triggered either by earthquakes, rockfalls, or snow avalanches. To clarify the possible interactions between environmental changes and these natural hazards in mountain and piedmont lakes, we analyze the lacustrine sedimentary records of key historical events and discuss the recurrence of similar regional events in the past. High-resolution seismic profiles and sediment cores from large perialpine lakes (Bourget, Geneva, and Constance) and from small mountain lakes in the French Alps and the Pyrenees were used to establish a conceptual model linking environmental changes, tributary flood sedimentary processes, subaquatic deltaic depocenters, and potentially tsunamigenic mass-wasting deposits. These findings illustrate the specific signatures of the largest French earthquakes in 1660 CE (northern Pyrenees) and in 1822 CE (western Alps) and suggest their recurrence during the Holocene. In addition, the regional record in the Aiguilles Rouges massif near Mont Blanc of the tsunamigenic 1584 CE Aigle earthquake in Lake Geneva may be used to better document a similar Celtic event ca. 2300 Cal BP at the border between Switzerland and France
Un plomb venu de loin : sources et acheminement du métal à Vienne durant la période romaine
No full textInternational audienc
Pollution au plomb et état sanitaire dans la Vienne antique : une approche isotopique
No full textInternational audienc
Transformer-based Track Fitting for HL-LHC
No full textInternational audienceAs the High-Luminosity LHC (HL-LHC) era approaches, significant improvements in reconstruction software are required to keep pace with the increased data rates and detector complexity. Some promising algorithms for high-throughput event reconstruction are GPU-based algorithms. Among those algorithms, the Kalman Filter is commonly used for the estimation of track parameters. While it leads to a good agreement between the truth and estimated parameters, the varying number of iterations required by the tracks leads to a loss of computational power.This work investigates a deep learning-based alternative using Transformer architectures for the prediction of the track parameters. We evaluate the approach in a realistic setting using the ACTS software framework with the Open Data Detector (ODD) geometry on full simulation and Kalman Filter for baseline comparison, observing promising results.This work addresses a key bottleneck in adapting track reconstruction workflows to GPU-accelerated computing environments: the lack of a GPU-native, high-performance alternative to Kalman Filter-based track fitting. This study shows a Transformer-based regression model applied on samples made with the ACTS framework and realistic detector conditions. It represents the first application of such a model with the Open Data Detector and a full reconstruction pipeline
Budget 2026 : les conséquences de l’échec de la Commission mixte paritaire et du choix de l’option de la loi spéciale
No full textThe ConversationL’échec de la commission mixte paritaire sur le projet de loi de finances pour 2026 place le gouvernement face à un choix délicat. Le premier ministre entend déposer un projet de loi spéciale, comme en décembre 2024, après le renversement du gouvernement de Michel Barnier. Ce choix soulève d’importantes questions de conformité constitutionnelle et de portée juridique. Décryptage
Laying Down the Foundations for Qualitative and Quantitative Freshwater Zooplankton Metabarcoding Surveys
No full textInternational audienceMetabarcoding has been proven to be a highly effective tool for surveying biodiversity in aquatic ecosystems. Despite the recent explosion of metabarcoding studies, relatively few efforts have focused on freshwater zooplankton communities, even though they form an essential component of freshwater ecosystems. Here, we evaluate some essential aspects of metabarcoding surveys to provide a solid basis for the development of qualitative and quantitative metabarcoding surveys for freshwater zooplankton communities. We developed and validated taxon-specific and universally applicable metabarcoding primers for Cladocera, Copepoda and Rotifera. These primers were subsequently used to assess optimal sample collection, preservation and DNA extraction protocols and gain insights into the key biases that may influence the interpretation of freshwater zooplankton metabarcoding results. The presented primers performed well when applied to both bulk community samples and eDNA samples, although the latter seemed to slightly decrease the performance of metabarcoding analyses. We found significant effects of subsample size, sample preservation, DNA extraction and lysis methods for bulk community samples with inadequate protocols likely to underestimate Cladocera diversity. Primer amplification efficiency was found to be the primary driver of biases in the quantitative interpretation of metabarcoding data, regardless of the taxonomic target or sequencing depth. General correction factors based on primer amplification efficiency may thus be sufficient to enhance the quantitative nature of metabarcoding surveys
L’épitaphe chrétienne de Priscianus : un texte latin des Balkans septentrionaux
No full textInternational audienc
Low-Mach-number limit for multiphase flows
No full textInternational audienceThis paper is devoted to the study of the low-Mach-number limit for solutions of the compressible Navier-Stokes or Euler equations for different types of fluids. We first review the different results obtained in the case of flows consisting of one phase. Then, we focus on the low-Mach-number limit for two-phase flows, considering different types of systems: with an algebraic closure or a PDE closure for the pressure, with one single or two different velocities, without or with entropy
EUNIS habitat maps: enhancing thematic and spatial resolution for Europe through machine learning
No full textInternational audienceThe EUNIS habitat classification is crucial for categorising European habitats, supporting European policy on nature conservation and implementing the Nature Restoration Law. To meet the growing demand for detailed and accurate habitat information, we provide spatial predictions across Europe (EEA39 territory) for 260 EUNIS habitat types at hierarchical level 3, together with independent validation and uncertainty analyses. Using ensemble machine learning models, together with high-resolution satellite imagery and ecologically meaningful climatic, topographic and edaphic variables, we produced a European habitat map indicating the most probable habitat overall at 100-m resolution across Europe. Additionally, we provide information on prediction uncertainty and the most probable habitats at level 3 within each EUNIS level 1 formation. This product is particularly useful for both conservation and restoration purposes. Predictions were cross-validated at European scale using a spatial block cross-validation and evaluated against independent data from France (forests only), the Netherlands and Austria. The maps achieved strong predictive performance, with F1-scores ranging from 0.61 to 0.94 in spatial cross-validation and from 0.33 to 0.95 in external validation datasets with distinct trade-offs in terms of recall and precision across habitat formations. Accuracy improved for rare or localized habitats when considering the top 3 predicted classes