Science & Mathematics

Bacteriophages participate in soil life by influencing bacterial community structure and function, biogeochemical cycling and horizontal gene transfer. Despite their great abundance, diversity, and importance in microbial processes, they remain little explored in environmental studies. The influence of abiotic factors on the persistence of bacteriophages is now recognized; however, it has been mainly studied under experimental conditions. This study aimed to determine whether the abiotic factors well-known to influence bacteriophage persistence also control the natural distribution of the known DNA bacteriophage populations. To this end, soil from eight study sites including forests and grasslands located in the Attert River basin (Grand Duchy of Luxembourg) were sampled, covering different soil and land cover characteristics. Shotgun metagenomics, reference-based bioinformatics and statistical analyses allowed characterising the diversity of known DNA bacteriophage and bacterial communities. After combining soil properties with the identified DNA bacteriophage populations, our in-situ study highlighted the influence of pH and calcium cations on the diversity of the known fraction of the soil DNA bacteriophages. More interestingly, significant relationships were established between bacteriophage and bacterial populations. This study provides new insights into the importance of abiotic and biotic factors in the distribution of DNA bacteriophages and the natural ecology of terrestrial bacteriophages.

The initial issue of complexity management in companies is the identification of the drivers of complexity. However, current literature lacks methods for assisting practitioners in the initial identification of such drivers. This paper, therefore, presents a novel framework for assisting practitioners with identifying complexity drivers in manufacturing companies. The framework uses a generic value chain and a generic product structure as its two dimensions. Multiple workshops are then conducted with company representatives across different value chain fields focusing on two main parts: First, surveys are used to assign complexity ratings to different generic product structure elements. Secondly, the complexity ratings are elaborated on by workshop participants. The process provides valuable insights into the perceived complexity drivers. The framework is then verified through a case study in the process industry. Based on the case study, multiple complexity drivers were identified across both value chain fields and product structure elements. The case study findings show that the framework facilitates practitioners in identifying organization-wide perceived complexity drivers. The framework contributes to both industry and research by addressing a neglected aspect of complexity management. It achieves this by providing a comprehensive and structured approach for the initial identification of complexity drivers across product elements and value chain fields.

En dépit de leur absence notable sur la plupart des plans de ville, les ruelles (hẻm) forment le cœur de l’armature urbaine de HCMV et constituent encore le cadre de vie de près de 85 % de la population citadine. Ces quartiers accueillent de très importantes densités de population, allant jusqu’à plus de 80 000 hab. / km2, ce qui rend les conditions de leur modernisation et de leur élargissement très complexes. L’objectif de ce papier est précisément de rentre compte des mutations urbaines contemporaines de l’« envers métropolitain » de HCMV, que sont les quar- tiers de ruelles, généralement peu documentés dans les travaux de recherche et d’expertise. Il s’agit d’interroger les différentes trajectoires de modernisation contemporaine de la ville ancienne, ayant connu une croissance organique, souvent à l’écart des projets planification. En termes de gouvernance urbaine, cette ques- tion permet de souligner le rôle de plus en plus important de l’échelle de gestion locale des districts, en charge de l’entretien – et aujourd’hui de la modernisation – du réseau des ruelles. En se fondant sur une étude comparée de six sous-districts de HCMV, ce working paper témoigne de la grande diversité socio-spatiale des quartiers de ruelles et donc de leurs trajectoires différenciées dans le contexte de modernisation urbaine contemporain.

Action Unit (AU) detection aims at automatically caracterizing facial expressions with the muscular activations they involve. Its main interest is to provide a low-level face representation that can be used to assist higher level affective computing tasks learning. Yet, it is a challenging task. Indeed, the available annotated databases display limited face variability and are imbalanced toward neutral expressions. Furthermore, as AU involve subtle face movements they are difficult to annotate so that some of the few provided datapoints may be mislabeled. Therefore, a deep neural network trained on such datasets is likely to overfit and to consequently display poor generalisation performance. In this thesis, we explore three different ways to improve the generalisation performance of an AU detector.First, we aim at exploiting known inter-dependencies between Action Units to better structure the prediction of the network. To do so, we leverage multi-task recurrent neural networks that process tasks sequentially and help each task prediction with the results of the previous ones. However, such method requires imposing an arbitrary order into an unordered task set. Crucially, it has been shown that this ordering choice impacts predictive performance. For that purpose, we propose Multi-Order Networks (MONET) that jointly learns multiple tasks along with the order in which they should be predicted. Experimentally, we first show that MONET is able to learn the optimal prediction order in a controlled environment. Second, we validate MONET architecture by showing that MONET outperform existing multi-task baselines on multiple attribute detection problems chosen for their wide range of dependency settings. More importantly, we demonstrate that MONET significantly extends state-of-the-art AU detection performance on both DISFA and BP4D. Second, we build upon the observation that each Action Unit is locally constrained to a known region of the face. Consequently we attempt to guide our deep Action Unit detector towards paying attention to AU-relevant zones. For that purpose, we draw inspiration from the recent success of vision transformer architectures and their multi-head attention mechanism. More precisely we benchmark several strategies to guide the multi-head attention of vision transformers with either landmarks-based and face parsing-based priors. The main outcome of our study is that, whatever the nature of the prior, forcing different attention heads to pay attention to different zones is an useful strategy as it improves AU detection performance on both DISFA and BP4D. Finally, we tackle the noisy annotation problem of AU detection. To this end, we first attempt to use label smoothing to reduce the network confidence and consequently mitigate the influence of noisy examples on the training. However, we experimentally observe that label smoothing degrades AU detection performance. To explain this drop in performance, we hypothetize that applying label smoothing in imbalanced scenarios worsens the pre-existing under-confidence in the minority class. To circumvent that issue, we propose a Vanilla Asymetric Label Smoothing (VALS) that use separate smoothing coefficients for positive and negative examples. Consequently VALS provides the flexibility to reduce over-confidence in the majority class while leaving the minority class untouched. We further refine this strategy by proposing Robin Hood Label Smoothing (RHLS) that smoothes only the majority class with a smoothing coefficient that is scaled based on AU empirical frequencies. Experimentally, we show that both VALS and RHLS display promising performance on BP4D and outperforms state-of-the-art results on DISFA.

In Mwotlap, an Oceanic language of Vanuatu, the principal device for referring to space is a paradigm of six directionals. Organized in pairs, these morphemes define three ways to draw a vector in space: by reference to a salient participant (hither-thither); by reference to an asymmetry perceptible within the immediate, local setting (up-down; in-out); or by reference to a fixed, absolute system of four horizontal quadrants (also lexified as up-down; in-out). These three "coordinate sets" can be shown to obey a strict hierarchy, determining which one the speaker should activate in a given situation. After providing an overview of this directional system, this paper investigates in more detail the mechanics of geocentric reference in Mwotlap, whereby a land/sea axis (in-out) is crossed by a second axis, running from [south]east (up) to [north]west (down). In order to account for this use of the vertical directionals up--down on the horizontal plane, a semantic hypothesis is proposed that is related to the seafaring history of Mwotlap's population.

On décrit les dispositifs d'irradiation et de protection installés pour recevoir 1 kilocurie de radiocobalt, destinés à des recherches sur les effets chimiques des rayonnements γ et sur les réactions photonucléaires. Les échantillons de cobalt radioactifs sont placés dans six petits cylindres fixes disposés selon un anneau enrobé dans un tambour de plomb. Les diverses possibilités d'irradiation sont décrites et on indique les doses intégrales par heure en diverses positions.

In this study, an Analytic Hierarchy Process based approach is proposed for the selection of heat stress prevention measures. This approach provides decision makers with an easy way to evaluate the criteria that help the decision-making process in choosing appropriate measures for hotspots. The study focuses on the key pre-criteria before implementing an intervention. The proposed scheme is based on a comparison between the interventions to show the importance of the alternatives. The evaluation of the measures is obtained from a questionnaire where human judgment is used for a comparison, based on their perception and priorities. The final results showed that green roof is most beneficial option for heat stress mitigation with respect to criteria taken into account.

In the present work, we consider 2-person zero-sum stochastic differential games with a nonlinear pay-off functional which is defined through a backward stochastic differential equation. Our main objective is to study for such a game the problem of the existence of a value without Isaacs condition. Not surprising, this requires a suitable concept of mixed strategies which, to the authors' best knowledge, was not known in the context of stochastic differential games. For this, we consider nonanticipative strategies with a delay defined through a partition $\pi$ of the time interval $[0,T]$. The underlying stochastic controls for the both players are randomized along $\pi$ by a hazard which is independent of the governing Brownian motion, and knowing the information available at the left time point $t_{j-1}$ of the subintervals generated by $\pi$, the controls of Players 1 and 2 are conditionally independent over $[t_{j-1},t_j)$. It is shown that the associated lower and upper value functions $W^{\pi}$ and $U^{\pi}$ converge uniformly on compacts to a function $V$, the so-called value in mixed strategies, as the mesh of $\pi$ tends to zero. This function $V$ is characterized as the unique viscosity solution of the associated Hamilton-Jacobi-Bellman-Isaacs equation.

Cette contribution présente une figurine appartenant à la collection Berne-Malibu-Copenhague, issue de pillages réalisés en Italie du Sud dans les années 1970-1980 et conservée désormais au Museo nazionale archeologico della Sibaritide. Examinée dans le cadre de la publication d'ensemble de cette collection, en cours de préparation, la pièce se caractérise par une grande originalité morphologique. Le type moulé a été créé à Métaponte à la fin du VII e siècle av. J.-C., mais la technique de fabrication combinant un avers moulé et un revers modelé ne trouve pas de parallèle exact dans cette cité, ni dans les établissements voisins. Sa provenance ne peut pas être assurée actuellement mais elle constitue un témoin original de la diversité des procédés de fabrication des premiers plasticiens mouleurs de Grande Grèce.

Steady buckling (coiling) of thin falling liquid jets is sensitive to surface tension, yet an understanding of these capillary effects lags far behind what is known about surface tension-free coiling. In experiments with sub-millimetric jets and ultra-low flow rates, we find that the critical dispensing height Hc for coiling decreases with increasing flow rate, a trend opposite to that found previously for inertia-free coiling. We resolve the apparent contradiction using nonlinear numerical simulations based on slender-jet theory, which show that the trend reversal is due to the strong effect of surface tension in our experiments. We use our experiments to construct a regime diagram (coiling vs. stagnation flow) in the space of capillary number Ca and jet slenderness , and find that it agrees well with fully nonlinear numerical simulations. However, it differs substantially from the analogous regime diagram determined experimentally by Le Merrer et al. (2012) for the unsteady buckling of a compressed liquid bridge. Using linear stability analysis, we show that the differences between the two regime diagrams can be explained by a combination of shape nonuniformity and the influence of gravity.

Pandemoniumpanyard, 3 Norfolk street, Belmont

La culture du riz, une céréale largement cultivée dans le monde et particulièrement en Haïti où une très forte partie de terre est consacrés à cette culture. Par rapport à la moyenne mondiale le rendement est très faible en Haïti. Les résultats montrent que la variété Proce4 a le meilleur rendement avec 5.72 tonnes /hectare et les variétés Juma 69 et M1 ont enregistrées rendements de 4.7 tonnes/ hectare. La variété V08 présente le pour le plus élevé de talles fertiles avec une valeur de 8.05 % en moyenne. La variété Proce4 s’est révélée la plus performante pour une augmentation de rendement dans la plaine de Torbeck.

In this paper we present a family of gradient-enhanced continuum damage models which can be viewed as a regularization of the variational approach to fracture capable of predicting in a unified framework the onset and space-time dynamic propagation (growth, kinking, branching, arrest) of complex cracks in quasi-brittle materials under severe dynamic loading. The dynamic evolution problem for a general class of such damage models is formulated as a variational inequality involving the action integral of a generalized Lagrangian and its physical interpretation is given. Finite-element based implementation is then detailed and mathematical optimization methods are directly used at the structural scale exploiting fully the variational nature of the formulation. Finally, the link with the classical dynamic Griffith theory and with the original quasi-static model as well as various dynamic fracture phenomena are illustrated by representative numerical examples in quantitative accordance with theoretical or experimental results.