Scientific Publications of the University of Toulouse II Le Mirail
Not a member yet
92205 research outputs found
Sort by
Raw material procurement and knapping strategies for the stone tools from the 709 ka butchery site of Kalinga (Philippines), with an insight on the Cagayan Valley lithic assemblage
No full textInternational audienceOur knowledge of the Lower Palaeolithic of Island Southeast Asia is limited to less than a handful of archaeological sites with clear stratigraphic context : the 1 Ma Mata Menge site on Flores Island (Brumm et al., 2006), the ca. 800 ka Ngebung 2 site on Java Island (Sémah, 2001), the 709 ka site of Kalinga on Luzon Island (Ingicco et al., 2018) and the 180 ka site of Talepu on Sulawesi Island. Furthermore, most of the lithic industry terms still in usage nowadays: Pacitanian for Java, Cabengian for Sulawesi or Cabalwanian for Luzon, are linked to old discoveries of a few or more stone artefacts out of any context and the significance of which, in terms of traditions, has generally been questioned if not disregarded. In the Philippines, three cultural traditions have been described for the Paleolithic from a handful of stone artefact from surface collection, two of them within the Cagayan Valley (Luzon Island). The aim of this presentation is two-fold. First to reevaluate the diversity of the surface collections made in the Cagayan Valley since the 1970s at the light of the only well-dated archaeological artefacts excavated from the Kalinga site which will also be lengthy described here. Second, to better circumscribe the Cagayan Valley lithic assemblages within the Palaeolithic of Asia. Special attention will be given to the small amorphous flake tools as they are the type of tools recovered at Kalinga site. The potential raw material gathering areas have been searched for within the regional fossil fluvial formations rich in cobbles and petrological analysis was conducted. Furthermore, a knapping experiment was set up in order to better characterize the constraints faced and thereafter decisions made by the Kalinga toolmakers, and therefore to better understand and describe the final products recovered along the rhino carcass at the Kalinga site. Finally, we will describe the diversity of knapping products and chaîne opératoire for the lithic assemblage of the surface and excavated Cagayan Valley assemblage and compare them with other continental and insular Souheast Asian collections
Agrégation, ségrégation, conformisation dans les systèmes vivants: De la modélisation à l’analyse, regards croisés sur les systèmes de plusieurs populations en diffusion et en interaction
No full textL’approche par l’anthropotopie. Regard en construction sur les rapports que les humains entretiennent avec leurs espaces et sur les contextes d’espaces affectés (Conférence au Séminaire d’anthropologie générale du Centre d'Anthropologie Sociale - LISST- Toulouse)
No full textInternational audienceCette communication propose un panorama un peu général d’un programme de recherche qui m’anime depuis quelques temps et qui relève de l’anthropologie des espaces, notamment des espaces affectés, des territoires bouleversés. Je n’ai jusqu’à présent pas eu le temps de m’y consacrer pleinement. C’est une manière pour moi de faire une mise à jour mais aussi de montrer ce qui m’intéresse, les questions de recherche que j’explore et les propositions conceptuelles que je développe. Mon propos sera accompagné de cas ethnographiques que j’ai mené dans la région Sichuanaise en Chine et plus récemment en Sicile. Ils permettront d’illustrer mon propos, de clarifier mon approche mais aussi de montrer un panel varié de types de recherche et sujets qui peuvent être explorés par la perspective que je propose. La troisième partie est tout particulièrement dédiée à ces pistes d’application. Mais d’abord dans la première partie je présente mon approche et définis le concept que je façonne : l’anthropotopie, qui m’aide à penser mes objets de recherche. La seconde partie, plus courte, vise à montrer comment cette approche s’inscrit dans le champ de recherche sur les espaces et comment l’anthropotopie dialogue avec les autres concepts de ce champ mais aussi les autres disciplines. Cela permettra de préciser la spécificité de cette approche, son intérêt aussi, tout en montrant qu’elle prend en compte des processus déjà bien identifiés et analysés par les prédécesseurs. Car il ne s’agit pas de révolutionner le champ mais d’ouvrir une perspective qui peut à mon sens, s’avérer heuristique pour mieux comprendre et analyser certains cas et contextes, en lien avec les transformations spatiales
Boundary controllability of time-discrete parabolic systems: a moments method approach
No full textIn this work, we investigate the boundary controllability of time-discrete parabolic systems, uniformly with respect to the discretization parameter. To establish our main results, we adapt and extend the moment method first introduced by Fattorini and Russell, to the time-discrete setting. While this method has proven effective in the continuous framework and in the space-discrete case, its adaptation to time-discrete systems is challenging due to the fact that the more accurate results in the field are based on several complex analysis tools. To overcome this, we introduce a new alternative proof for constructing biorthogonals to (generalized) exponential functions in the continuous setting, based on the original proof by Fattorini and Russell which avoids the use of these tools, and taking ideas from the block moment method introduced by Benabdallah, Morancey and the first author of this work. We then manage to adapt this strategy to the discrete setting, enabling the construction and estimation of biorthogonal families for some time-discrete functions that play the same role as the exponentials at the discrete level. Our results show that these biorthogonals can be uniformly estimated for a finite portion of the spectrum, determined by the discretization parameter , and that converges to the whole spectrum as goes to zero.Using this tool, we prove a relaxed null-controllability result for discrete parabolic systems. It says that there exists a bounded sequence of time discrete controls that makes the solution reach a target at final time tending to zero exponentially fast as the discretization parameter goes to zero. We will also study the effects of the existence of a minimal null-control time and how this phenomenon translates into the discrete world
Le dernier grand archevêque gascon d’Auch: Amanieu II d’Armagnac, archevêque d’Auch (1261-1318)
No full textInternational audienceAprès la présentation de son testament, qui révélait en creux sa possible opposition à l’influence capétienne, cette communication s'attarde davantage sur le long parcours d'archevêque d'Amanieu, lui que ses premiers pas comme chanoine de Toulouse ne destinaient pas au siège d’Auch, ni à un affrontement à l’influence capétienne. Elle aborde le parcours de ce noble et prélat gascon sous les prismes successifs de ses origines, de sa seigneurie archiépiscopale, de l’ancrage territorial de son pouvoir, et surtout de son rôle politique et parfois très diplomatique, entre les deux pouvoirs royaux qui se partagent la Gascogne
Nexo entre el agua, actividad petrolera, medios de vida y agricultura de frontera en la Amazonia
No full textInternational audienc
Obsolescence and Availability: The Hidden Challenges of Complex Systems
No full textInternational audienceThe obsolescence challenges associated with rapid technological advances have a significant impact on the overall performance of complex, longlife systems, particularly in sectors such as transportation and heavy industry. This paper addresses these challenges by modeling how obsolescence impacts system availability, reliability, and maintainability considering different categories of item: Make/Buy Repairable and Make/Buy Consumable. Each category encompasses distinct failure, fault and replacement scenarios, which are analysed using flow diagrams. The proposed models provide a systematic understanding of obsolescence mechanisms, including inventory management strategies, repair or replacement paths, and supplier interactions. They also address the progressive decline in availability caused by the obsolescence of components, tools, documentation and technical soft skills. The paper contributes a structured modeling approach based on flow diagrams that links obsolescence scenarios to maintenance decisions, helping anticipate and mitigate performance degradation. By clarifying the decision-making processes inherent in each scenario, the study improves the understanding of the impact of obsolescence on system performance. The article ends with findings and research on predicting system availability under obsolescence
A Substring Extraction-Based RAG Method for Minimising Hallucinations in Aircraft Maintenance Question Answering
No full textInternational audienceHallucination occurs when a language model generates plausible yet nonfactual information. In particular, faithfulness hallucinations (inconsistency with a given context) cannot be tolerated in critical domains such as aircraft maintenance due to the potentially severe consequences. To mitigate this issue, Retrieval Augmented Generation (RAG) methods have been introduced. These approaches are relevant for reducing the risks of hallucination but do not eliminate them, as the generator may still produce content unfaithful to the retrieved context. This paper proposes a novel RAG approach that leverages a substring extraction tool from retrieved documents to minimise hallucinations. Experiments performed on real aircraft maintenance documentation revealed that, despite the lower accuracy of the answers compared to traditional RAG methods, the proposed approach demonstrates an improved control over hallucination risks. This highlights the potential of our method in highly technical use cases where accuracy and reliability are key
Optimisation des systèmes de contrôle complexes avec des simulateurs différentiables : une approche hybride de l'apprentissage par renforcement et de la planification de trajectoire
No full textInternational audienceDeep reinforcement learning (RL) often relies on simulators as abstract oracles to model interactions within complex environments. While differentiable simulators have recently emerged for multi-body robotic systems, they remain underutilized, despite their potential to provide richer information. This underutilization, coupled with the high computational cost of exploration-exploitation in high-dimensional state spaces, limits the practical application of RL in the real-world. We propose a method that integrates learning with differentiable simulators to enhance the efficiency of exploration-exploitation. Our approach learns value functions, state trajectories, and control policies from locally optimal runs of a model-based trajectory optimizer. The learned value function acts as a proxy to shorten the preview horizon, while approximated state and control policies guide the trajectory optimization. We benchmark our algorithm on three classical control problems and a torque-controlled 7 degree-of-freedom robot manipulator arm, demonstrating faster convergence and a more efficient symbiotic relationship between learning and simulation for end-to-end training of complex, poly-articulated systems.L'apprentissage par renforcement profond (RL) s'appuie souvent sur des simulateurs comme oracles abstraits pour modéliser les interactions au sein d'environnements complexes. Bien que des simulateurs différentiables aient récemment émergé pour les systèmes robotiques multi-corps, ils restent sous-utilisés, malgré leur potentiel à fournir des informations plus riches. Cette sous-utilisation, conjuguée au coût de calcul élevé de l'exploration-exploitation dans des espaces d'état de grande dimension, limite l'application pratique de l'RL en situation réelle. Nous proposons une méthode intégrant l'apprentissage à des simulateurs différentiables afin d'améliorer l'efficacité de l'exploration-exploitation. Notre approche apprend des fonctions de valeur, des trajectoires d'état et des politiques de contrôle à partir d'exécutions localement optimales d'un optimiseur de trajectoire basé sur un modèle. La fonction de valeur apprise agit comme un proxy pour raccourcir l'horizon de prévisualisation, tandis que les politiques d'état et de contrôle approximatives guident l'optimisation de la trajectoire. Nous comparons notre algorithme à trois problèmes de contrôle classiques et à un bras manipulateur robotique à 7 degrés de liberté contrôlé par couple, démontrant une convergence plus rapide et une relation symbiotique plus efficace entre apprentissage et simulation pour l'apprentissage complet de systèmes complexes et polyarticulés