Laboratoire de Mathématiques de Bretagne Atlantique

1. Ordered Fields, Real Closed Fields.- 2. Semi-algebraic Sets.- 3. Real Algebraic Varieties.- 4. Real Algebra.- 5. The Tarski-Seidenberg Principle as a Transfer Tool.- 6. Hilbert's 17th Problem. Quadratic Forms.- 7. Real Spectrum.- 8. Nash Functions.- 9. Stratifications.- 10. Real Places.- 11. Topology of Real Algebraic Varieties.- 12. Algebraic Vector Bundles.- 13. Polynomial or Regular Mappings with Values in Spheres.- 14. Algebraic Models of C? Manifolds.- 15. Witt Rings in Real Algebraic Geometry.- Index of Notation.

We report that the oxytocin-mediated neuroprotective γ-aminobutyric acid (GABA) excitatory-inhibitory shift during delivery is abolished in the valproate and fragile X rodent models of autism. During delivery and subsequently, hippocampal neurons in these models have elevated intracellular chloride levels, increased excitatory GABA, enhanced glutamatergic activity, and elevated gamma oscillations. Maternal pretreatment with bumetanide restored in offspring control electrophysiological and behavioral phenotypes. Conversely, blocking oxytocin signaling in naïve mothers produced offspring having electrophysiological and behavioral autistic-like features. Our results suggest a chronic deficient chloride regulation in these rodent models of autism and stress the importance of oxytocin-mediated GABAergic inhibition during the delivery process. Our data validate the amelioration observed with bumetanide and oxytocin and point to common pathways in a drug-induced and a genetic rodent model of autism.

Autosomal dominant polycystic kidney disease (ADPKD) is heterogeneous with regard to genic and allelic heterogeneity, as well as phenotypic variability. The genotype-phenotype relationship in ADPKD is not completely understood. Here, we studied 741 patients with ADPKD from 519 pedigrees in the Genkyst cohort and confirmed that renal survival associated with PKD2 mutations was approximately 20 years longer than that associated with PKD1 mutations. The median age at onset of ESRD was 58 years for PKD1 carriers and 79 years for PKD2 carriers. Regarding the allelic effect on phenotype, in contrast to previous studies, we found that the type of PKD1 mutation, but not its position, correlated strongly with renal survival. The median age at onset of ESRD was 55 years for carriers of a truncating mutation and 67 years for carriers of a nontruncating mutation. This observation allows the integration of genic and allelic effects into a single scheme, which may have prognostic value.

International audience

The static and dynamic properties of Permalloy films with thicknesses varying from 200 nm to 920 nm have been investigated in detail. For these films possessing a perpendicular anisotropy, the critical thickness for stripe domain nucleation has been determined using magnetic force microscopy imaging and in-plane hysteresis loop measurements. The zero-field dynamic permeability spectra measured over the frequency range 0.1--6 GHz reveal one resonance line below the critical thickness, whereas multiple resonance peaks whose number increases with increasing film thickness are observed above the critical thickness. The two-dimensional dynamic micromagnetic simulations reproduce successfully the thickness evolution of the experimental permeability spectra and give access to the thickness-dependent spatial localization of the main modes. However, the experimental resonance linewidths exceed the computed ones for the thickest films. Extended dynamic micromagnetic simulations including the screening of the pumping field due to eddy currents account only partially for resonance line broadening. The existence of additional relaxation mechanisms due to magnetic inhomogeneities (micromagnetic and structural) is discussed on the basis of parallel ferromagnetic resonance measurements versus frequency.

In light of growing interest in data-driven methods for oceanic, atmospheric, and climate sciences, this work focuses on the field of data assimilation and presents the analog data assimilation (AnDA). The proposed framework produces a reconstruction of the system dynamics in a fully data-driven manner where no explicit knowledge of the dynamical model is required. Instead, a representative catalog of trajectories of the system is assumed to be available. Based on this catalog, the analog data assimilation combines the nonparametric sampling of the dynamics using analog forecasting methods with ensemble-based assimilation techniques. This study explores different analog forecasting strategies and derives both ensemble Kalman and particle filtering versions of the proposed analog data assimilation approach. Numerical experiments are examined for two chaotic dynamical systems: the Lorenz-63 and Lorenz-96 systems. The performance of the analog data assimilation is discussed with respect to classical model-driven assimilation. A Matlab toolbox and Python library of the AnDA are provided to help further research building upon the present findings.

From a ferromagnetic resonance experiment on a YIG|GGG|YIG crystal, phonons are shown to be able to provide coherent long-range coupling between two magnets, which act as ``microphones'' and ``speakers'' for acoustic waves to induce feedback interference over millimeter distances. In this ``phononic spin valve'' under perpendicular magnetization, the phonons, which are circularly polarized, carry efficiently angular momentum current through a nonmagnetic dielectric. The high acoustic quality of phonon transport in garnets and the strong coupling to the magnetic order may also be useful for quantum communication.

In this paper, we improve the parallel deep neural network (DNN) scheme BusterNet for image copy-move forgery localization with source/target region distinguishment. BusterNet is based on two branches, i.e., Simi-Det and Mani-Det, and suffers from two main drawbacks: (a) it should ensure that both branches correctly locate regions; (b) the Simi-Det branch only extracts single-level and low-resolution features using VGG16 with four pooling layers. To ensure the identification of the source and target regions, we introduce two subnetworks that are constructed serially: the copy-move similarity detection network (CMSDNet) and the source/target region distinguishment network (STRDNet). Regarding the second drawback, the CMSDNet subnetwork improves Simi-Det by removing the last pooling layer in VGG16 and by introducing atrous convolution into VGG16 to preserve field-of-views of filters after the removal of the fourth pooling layer; double-level self-correlation is also considered for matching hierarchical features. Moreover, atrous spatial pyramid pooling and attention mechanism allow the capture of multiscale features and provide evidence for important information. Finally, STRDNet is designed to determine the similar regions obtained from CMSDNet directly as tampered regions and untampered regions. It determines regions at the image-level rather than at the pixel-level as made by Mani-Det of BusterNet. Experimental results on four publicly available datasets (new synthetic dataset, CASIA, CoMoFoD, and COVERAGE) demonstrate that the proposed algorithm is superior to the state-of-the-art algorithms in terms of similarity detection ability and source/target distinguishment ability.

Some preliminary studies, dealing with the process optimisation of pre-cast building elements made of Lime and Hemp Concrete (LHC), have shown that compression during casting lead to significant improvements: better mechanical characteristics and facing. However, this compaction leads to an increase of the weight to volume ratio and to a decrease in porous volume. Thus, the amount of entrapped air inside material, which contributes to decrease the thermal conductivity, is lower. Our data actually show a slight increase in thermal conductivity when compactness increases. The goal of this study is to compare the effect of compaction during casting on both mechanical and thermal characteristics of hardened specimens in order to evaluate the relevance of such a process.

Abstract Data assimilation combines forecasts from a numerical model with observations. Most of the current data assimilation algorithms consider the model and observation error terms as additive Gaussian noise, specified by their covariance matrices and , respectively. These error covariances, and specifically their respective amplitudes, determine the weights given to the background (i.e., the model forecasts) and to the observations in the solution of data assimilation algorithms (i.e., the analysis). Consequently, and matrices significantly impact the accuracy of the analysis. This review aims to present and to discuss, with a unified framework, different methods to jointly estimate the and matrices using ensemble-based data assimilation techniques. Most of the methods developed to date use the innovations, defined as differences between the observations and the projection of the forecasts onto the observation space. These methods are based on two main statistical criteria: 1) the method of moments, in which the theoretical and empirical moments of the innovations are assumed to be equal, and 2) methods that use the likelihood of the observations, themselves contained in the innovations. The reviewed methods assume that innovations are Gaussian random variables, although extension to other distributions is possible for likelihood-based methods. The methods also show some differences in terms of levels of complexity and applicability to high-dimensional systems. The conclusion of the review discusses the key challenges to further develop estimation methods for and . These challenges include taking into account time-varying error covariances, using limited observational coverage, estimating additional deterministic error terms, or accounting for correlated noise.

In order to improve our understanding of human resource management (HRM) in small- and medium-sized enterprises (SMEs) and causality inside the black box of the firm's human management, this paper examines the relationship between high-performance work systems (HPWSs) and SMEs' performance. HPWSs are conceived here like a whole of complementary HRM practices and not in terms of isolated practices. Firm's performance is studied through profitability, degree of innovation and social climate. In the case of 275 French SMEs, we have found simultaneous and longitudinal correlations between HPWS and all performance indicators; and that controlling past performance does not eliminate the correlation of HPWS with future performance.

International audience

Abstract The aim of the special issue is to bring together important current international research on innovative teaching and learning practices in mathematics in engineering education, and to develop deeper understandings of the characteristics of current teaching and learning practices that can inform the design and implementation of future innovative practice. The focus of this review paper is to provide a state-of-the-art overview of this emerging field at the cross-roads between mathematics and engineering education, in addition to introducing the papers of this special issue. To guide this paper, we posed three review questions: (1) How can current (teaching/learning/study) practices of mathematics in engineering education be characterized with a view towards innovation?; (2) What are the ‘resources’ (cognitive, material, digital, social) used, and what are those that appear also well suited for innovative courses?; (3) What are promising innovative practices in mathematics in engineering education, and what are the implications for curriculum reform? Looking back across the studies we summarized in the review, we conclude that they are lagging behind the more fundamental changes that are happening in engineering education, whilst addressing selected aspects of innovative changes within the current system of engineering education. At the same time, the nine papers of this special issue contribute new perspectives for innovative practices in mathematics in engineering education, for a better understanding of current practices and for future research.

In the classical Monge–Kantorovich problem, the transportation cost depends only on the amount of mass sent from sources to destinations and not on the paths followed by this mass. Thus, it does not allow for congestion effects. Using the notion of traffic intensity, we propose a variant, taking into account congestion. This variant is a continuous version of a well-known traffic problem on networks that is studied both in economics and in operational research. The interest of this problem is in its relations with traffic equilibria of Wardrop type. What we prove in the paper is exactly the existence and the variational characterization of equilibria in a continuous space setting.

The aim of this study is to compare three oil extraction methods and to evaluate their efficiency for producing an oil rich in polyphenols. The three extraction methods are screw pressing, extraction by supercritical CO 2 percolation and the combination of these two processes (Gas-Assisted Mechanical Expression: GAME). Screw pressing is the most efficient process for producing grape seed oil with a high yield, but supercritical CO 2 process permits an increase of polyphenol co-extraction with oil. The GAME process allows extraction of more polyphenols than screw pressing and constitutes an interesting process considering oil yield.

This article proposes a fast and open-source implementation of the well-known Non-Local Means (NLM) denoising algorithm, in its original pixelwise formulation. The fast implementation is based on the computation of patch distances using sums of lines that are invariant under a patch shift. The optimal parameters of NLM (in the average peak signal to noise ratio - PSNR - sense) are computed from an image database, thereby leading to a parameter-free NLM implementation. Comparison is performed with the parameter-free blockwise NLM implementation already proposed in IPOL journal by Buades, Coll and Morel. As expected the blockwise implementation offers better PSNR, at least when the noise standard deviation is large enough, but there is no significant difference in quality when performing visual inspection. The highlight is that the proposed parameter-free pixelwise NLM implementation is faster than the patchwise one by a factor of 6 to 49.

International audience

Deep learning methods have become an integral part of computer vision and machine learning research by providing significant improvement performed in many tasks such as classification, regression, and detection. These gains have been also observed in the field of remote sensing for Earth observation where most of the state-of-the-art results are now achieved by deep neural networks. However, one downside of these methods is the need for large amounts of annotated data, requiring lots of labor-intensive and expensive human efforts, in particular for specific domains that require expert knowledge such as medical imaging or remote sensing. In order to limit the requirement on data annotations, several self-supervised representation learning methods have been proposed to learn unsupervised image representations that can consequently serve for downstream tasks such as image classification, object detection or semantic segmentation. As a result, self-supervised learning approaches have been considerably adopted in the remote sensing domain within the last few years. In this article, we review the underlying principles developed by various self-supervised methods with a focus on scene classification task. We highlight the main contributions and analyze the experiments, as well as summarize the key conclusions, from each study. We then conduct extensive experiments on two public scene classification datasets to benchmark and evaluate different self-supervised models. Based on comparative results, we investigate the impact of individual augmentations when applied to remote sensing data as well as the use of self-supervised pre-training to boost the classification performance with limited number of labeled samples. We finally underline the current trends and challenges, as well as perspectives of self-supervised scene classification.

International audience

We construct and study a new family of TQFTs based on nilpotent highest weight representations of quantum sl(2) at a root of unity indexed by generic complex numbers. This extends to cobordisms the non-semi-simple invariants defined in [9] including the Kashaev invariant of links. Our case is not covered by the modular category framework and so we give a new example of application of the so-called "universal construction". For each root of unity of order 2r where r≥2 is not divisible by 4, our TQFT provides a monoidal functor from a category of surfaces and their cobordisms into the category of graded finite dimensional vector spaces. The functor is always symmetric monoidal but for even values of r the braiding on GrVect has to be the super-symmetric one, thus our TQFT may be considered as a super-TQFT. In the special case r=2 our construction yields a TQFT for a canonical normalization of Reidemeister torsion and we re-prove the classification of Lens spaces via the non-semi-simple quantum invariants defined in [9]. We prove that the representations of mapping class groups and Torelli groups resulting from our constructions are potentially more sensitive than those obtained from the standard Reshetikhin–Turaev functors; in particular we prove that the action of the bounding pairs generators of the Torelli group has always infinite order.