Archive ouverte de Centrale Lyon
Not a member yet
32420 research outputs found
Sort by
Elasticity, plasticity and fracture toughness of REBCO coated conductors characterized via micromechanical tests at room temperature
No full textInternational audienceREBCO, the leading candidate conductor for ultra high field magnets, is typically produced in the form of thin tapes, consisting of multiple layers of diverse materials with very different natures and properties. Knowledge of the mechanical properties of these different layers is crucial for magnet design. In this paper, we propose a methodology to measure the elasticity, plasticity and fracture toughness of conductor layer materials, at the scale of its constituents, based on nanoindentation, micropillar compression and micropillar splitting techniques. Measurements with these techniques are carried out at room temperature on a commercial conductor, and the results obtained are compared to the values found in the literature and to those specified by the manufacturers
Azumaya Algebras and Obstructions to Quadratic Pairs over a Scheme
No full textThere is a new example for both obstructions.International audienceWe investigate quadratic pairs for Azumaya algebras with involutions over a base scheme S as defined by Calmès and Fasel, generalizing the case of quadratic pairs on central simple algebras over a field (Knus, Merkurjev, Rost, Tignol). We describe a cohomological obstruction for an Azumaya algebra over S with orthogonal involution to admit a quadratic pair. When S is affine this obstruction vanishes, however it is non-trivial in general. In particular, we construct explicit examples with non-trivial obstructions
GE-06. Fractional operators for the analytical expressions of the dynamic magnetic power loss
No full textInternational audienceWe introduce a novel analytical approach for predicting the dynamic magnetic power loss in ferromagnetic materials subjected to alternating or rotating magnetic fields [2]. The proposed method utilizes fractional derivative analytical expressions of trigonometric functions [2-5], providing a computationally efficient and accurate tool for core loss prediction. The resulting analytical expressions are validated against a large amount of experimental data from state-ofthe-art setups [6, 7], demonstrating a high level of accuracy with a relative Euclidean distance 5% for most tested materials and consistently below 10% (Fig. 1). Key findings include the observation that the dynamic power loss contribution under a rotating magnetic field is precisely two times higher than that under an alternating field under standard sinusoidal flux density This insight is crucial for the design and optimization of electromagnetic converters such as transformers, inductors, and motors, where magnetic losses can significantly impact efficiency, performance, and reliability. The research also highlights that by understanding the material's electrical conductivity, the dynamic magnetic power loss can be simplified to a single parameter-the fractional order-which is consistent for both rotational and alternating contributions. The paper further discusses the implications of these findings on the selection of appropriate ferromagnetic materials for specific applications, contributing to the development of environmentally friendly technologies that consume less energy and reduce the overall impact. The study confirms the viscoelastic behavior of the magnetization process in ferromagnetic materials, thereby validating the use of fractional derivative operators for their simulation. This work offers a significant advancement in the understanding and prediction of magnetic losses in electromagnetic systems, with potential applications in various industries where compact and lightweight converters are critical, such as in portable electronic devices and aerospace systems
Ion Mobility Mass Spectrometry to Probe Sequences in Supramolecular Copolymers
No full textInternational audienceThe analysis of the microstructure of supramolecular copolymers is difficult because of their dynamic character. Here, benzene‐1,3,5‐tricarboxamide (BTA) co‐assemblies are analysed by ion mobility ‐ mass spectrometry (IM‐MS) to reveal the presence of various sequences. For example, the IM‐MS mobilogram for hexamers composed of 4 units from a first monomer and 2 units from a second monomer is a broad distribution due to the presence of 9 possible isomeric sequences, which can be sorted out based on calculated collision cross‐sections. This approach gives unprecedented information on supramolecular copolymer sequences
Mitigation of Sybil-based Poisoning Attacks in Permissionless Decentralized Learning
No full textDecentralized learning enables collaborative machine learning with enhanced privacy by allowing participants to train models locally and share updates for aggregation instead of sharing raw data. However, such systems are vulnerable to poisoning attacks that may compromise the learning process. This threat becomes even more severe when combined with sybil attacks, where adversaries contribute numerous malicious updates with minimal effort, amplifying their impact. To overcome these challenges, particularly in the permissionless setup, we propose SyDeLP, a blockchain-enabled protocol for decentralized learning. SyDeLP integrates byzantine tolerant aggregation for poisoning mitigation with a novel Verifiable Delay Puzzle to counter sybil attacks requiring Proofs of Work to participate. Honest behavior is incentivized by dynamically reducing puzzle difficulty, decreasing the computational burden for honest nodes over time. Empirical evaluations conducted on two benchmark datasets across four types of poisoning attack demonstrate that SyDeLP consistently outperforms existing solutions in terms of poisoning resilience.</div
Partition Strategies for the Maker–Breaker Domination Game
No full textInternational audienceThe Maker-Breaker domination game is a positional game played on a graph by two players called Dominator and Staller. The players alternately select a vertex of the graph that has not yet been chosen. Dominator wins if at some point the vertices she has chosen form a dominating set of the graph. Staller wins if Dominator cannot form a dominating set. Deciding if Dominator has a winning strategy has been shown to be a PSPACE-complete problem even when restricted to chordal or bipartite graphs. In this paper, we consider strategies for Dominator based on partitions of the graph into basic subgraphs where Dominator wins as the second player. Using partitions into cycles and edges (also called perfect [1,2]factors), we show that Dominator always wins in regular graphs and that deciding whether Dominator has a winning strategy as a second player can be computed in polynomial time for outerplanar and block graphs. We then study partitions into subgraphs with two universal vertices, which is equivalent to considering the existence of pairing dominating sets with adjacent pairs. We show that in interval graphs, Dominator wins if and only if such a partition exists. In particular, this implies that deciding whether Dominator has a winning strategy playing second is in NP for interval graphs. We finally provide an algorithm in n k+3 for interval graphs with at most k nested intervals.</div
Converting low-grade heat into mechanical energy using a natural rubber elastocaloric device
No full textInternational audienceThis work leverages the thermomechanical properties of natural rubber, an abundant, lowcost and renewable material, for low-grade heat energy harvesting. Natural rubber tubes exhibited temperature-induced stress variations of 10 kPa.K-1 when pre-elongated to 5 times their original length, for a temperature range of 20 to 60°C. A prototype elastocaloric device was evaluated using a fluid to cyclically transfer heat from a hot heat exchanger to the natural rubber and then to a cold heat exchanger. Constitutive equations and simulations were used to assess the energy conversion capability. Finally, thermodynamic energy cycles were experimentally tested with temperature variations of 35 K and elongations of 4.5 to 5.5 times the original length, converting heat to mechanical work at 4 J per cycle, corresponding to 150 mJ.cm-3. The proposed design is scalable using a larger quantity of natural rubber, and after adding an electromagnetic system, low-grade heat energy may be converted into electricity
Short and local transformations between ()-colorings
No full textInternational audienceRecoloring a graph is about finding a sequence of proper colorings of this graph from an initial coloring to a target coloring . Each pair of consecutive colorings must differ on exactly one vertex. The question becomes: is there a sequence of colorings from to ? In this paper, we focus on -colorings of graphs of maximum degree . Feghali, Johnson and Paulusma proved that, if both colorings are non-frozen (i.e. we can change the color of a least one vertex), then a quadratic recoloring sequence always exists. We improve their result by proving that there actually exists a linear transformation. In addition, we prove that the core of our algorithm can be performed locally. Informally, if we start from a coloring where there is a set of well-spread non-frozen vertices, then we can reach any other such coloring by recoloring only independent sets one after another. Moreover, these independent sets can be computed efficiently in the LOCAL model of distributed computing
IMPO: Interpretable Memory-based Prototypical Pooling
No full textInternational audienceGraph Neural Networks (GNNs) have proven their effectiveness in various graph-structured data applications. However, one of the significant challenges in the realm of GNNs is representation learning, a critical concept that bridges graph pooling, aimed at creating compressed graph representations, and explainable artificial intelligence, which focuses on building models with transparent reasoning mechanisms. This research paper introduces a novel approach called Interpretable Memory-based Prototypical Pooling (IMPO) to address this challenge. IMPO is a graph pooling layer designed to enhance the interpretability of GNNs while maintaining high performance in graph classification tasks. It builds upon the MemPool algorithm and incorporates prototypical components to cluster nodes around class-aware centroids. This approach allows IMPO to selectively aggregate relevant substructures, paving the way for generating more interpretable graph representations. The experimental results in our study underscore the potential of pooling architectures in constructing inherently explainable GNNs. Notably, IMPO achieves state-of-the-art results in both classification and explanatory capacities across a diverse set of graph classification datasets
A suitable equivalent heat source for the transient simulation of thick FSW joining
No full textInternational audienceThe main objective of this paper is to propose a suitable equivalent heat source that allows simulating transient FSW welding of thick structures with commercial finite element codes. At first, a fully coupled thermo-fluid model is developed by using the Rigid-ALE formalism in the case of a but joint of 20mm thick. This quasi-stationary model is calibrated from experimental measurements in terms of Taylor-Quinney coefficient and heat exchange with the backing plate. This identification is carried out by means of an optimization algorithm with Python programming language. From these numerical results, different shapes of equivalent heat source are investigated for simulating the thermal field without accounting for the material stirring. To the authors knowledge, such a modeling strategy has never been reported before in the literature, especially for thick joints. Based on this results, a suitable version of the heat source proposed by Goldak et al. [Metall. Trans. B 15 (1984) 299-305] is proposed. It is shown that the new heat source has the ability modeling very satisfactory the thermal gradient in the thickness of the joint during welding