Last 7 days
Bookmark and Share    
Full Text
See detailL'agriculture de subsistance et de semi-subsistance en Roumanie
Dona, Ion; Burny, Philippe ULg

Conference (2014, August 30)

Detailed reference viewed: 7 (1 ULg)
Full Text
See detailAspects économiques de la filière du vermicelle de tolomane au Vietnam
Le Thi Long, Vy; Vu Dinh, Ton; Lebailly, Philippe ULg et al

Conference (2014, August 30)

Detailed reference viewed: 6 (0 ULg)
Full Text
See detailProblématique foncière chez les exploitants agricoles de la Province de Kirundo au nord du Burundi
Minani, Bonaventure; Rurema, Déo-Guide; Burny, Philippe ULg et al

Conference (2014, August 30)

Detailed reference viewed: 7 (3 ULg)
Full Text
Peer Reviewed
See detailAccelerating Random Forests in Scikit-Learn
Louppe, Gilles ULg

Conference (2014, August 29)

Random Forests are without contest one of the most robust, accurate and versatile tools for solving machine learning tasks. Implementing this algorithm properly and efficiently remains however a ... [more ▼]

Random Forests are without contest one of the most robust, accurate and versatile tools for solving machine learning tasks. Implementing this algorithm properly and efficiently remains however a challenging task involving issues that are easily overlooked if not considered with care. In this talk, we present the Random Forests implementation developed within the Scikit-Learn machine learning library. In particular, we describe the iterative team efforts that led us to gradually improve our codebase and eventually make Scikit-Learn's Random Forests one of the most efficient implementations in the scientific ecosystem, across all libraries and programming languages. Algorithmic and technical optimizations that have made this possible include: - An efficient formulation of the decision tree algorithm, tailored for Random Forests; - Cythonization of the tree induction algorithm; - CPU cache optimizations, through low-level organization of data into contiguous memory blocks; - Efficient multi-threading through GIL-free routines; - A dedicated sorting procedure, taking into account the properties of data; - Shared pre-computations whenever critical. Overall, we believe that lessons learned from this case study extend to a broad range of scientific applications and may be of interest to anybody doing data analysis in Python. [less ▲]

Detailed reference viewed: 12 (3 ULg)
Full Text
See detailComputational homogenization of cellular materials with propagation of instabilities through the scales
Nguyen, Van Dung ULg; Noels, Ludovic ULg

Conference (2014, August 29)

The aim of this work is to develop an efficient multi–scale finite element framework to capture the buckling instabilities in cellular materials. As a classical multi–scale computational homogenization ... [more ▼]

The aim of this work is to develop an efficient multi–scale finite element framework to capture the buckling instabilities in cellular materials. As a classical multi–scale computational homogenization scheme looses accuracy with the apparition of the macroscopic localizations resulting from the micro–buckling, the second–order multi–scale computational homogenization scheme1 is considered. This second–order computational framework is herein enhanced with the following novelties so that it can be used for cellular materials. First, at the microscopic scale, the periodic boundary condition is used because of its efficiency. As the meshes generated from cellular materials exhibit a large void part on the boundaries and are not conforming in general, the classical enforcement based on the matching nodes cannot be applied. A new method based on the polynomial interpolation2 without the requirement of the matching mesh condition on opposite boundaries of the representative volume element (RVE) is developed. Next, in order to solve the underlying macroscopic Mindlin strain gradient continuum of this second–order scheme by the displacement–based finite element framework, the treatment of high order terms is based on the discontinuous Galerkin (DG) method to weakly impose the C1-continuity3. Finally, as the instability phenomena are considered at both scales of the cellular materials, the path following technique is adopted to solve both the macroscopic and microscopic problems4. The micro–buckling leading to the macroscopic localization and the size effect phenomena can be captured within the proposed framework. In particular it is shown that results are not dependent on the mesh size at the macroscopic scale during the softening response, and that they agree well with the direct numerical simulations. [less ▲]

Detailed reference viewed: 20 (6 ULg)
Full Text
Peer Reviewed
See detailLower critical field and SNS-Andreev spectroscopy of 122-arsenides: Evidence of nodeless superconducting gap
Abdel-Hafiez, M.; Pereira, P.J.; Kuzmichev, S.A. et al

in Physical Review. B : Condensed Matter (2014), 90

Using two experimental techniques, we studied single crystals of the 122-FeAs family with almost the same critical temperature, Tc. We investigated the temperature dependence of the lower critical field ... [more ▼]

Using two experimental techniques, we studied single crystals of the 122-FeAs family with almost the same critical temperature, Tc. We investigated the temperature dependence of the lower critical field Hc1(T) of a Ca0.32Na0.68Fe2As2 (Tc ≈ 34 K) single crystal under static magnetic fields H parallel to the c axis. The temperature dependence of the London penetration depth can be described equally well either by a single anisotropic s-wave-like gap or by a two-gap model, while a d-wave approach cannot be used to fit the London penetration depth data. Intrinsic multiple Andreev reflection effect spectroscopy was used to detect bulk gap values in single crystals of the intimate compound Ba0.65K0.35Fe2As2, with the same Tc. We estimated the range of the large gap value L = 6–8 meV (depending on small variation of Tc) and its a k space anisotropy of about 30%, and the small gap Delta ≈ 1.7 ± 0.3 meV. This clearly indicates that the gap structure of our investigated systems more likely corresponds to a nodeless s-wave two gaps. [less ▲]

Detailed reference viewed: 16 (1 ULg)
Full Text
Peer Reviewed
See detailLa place de la communication dans la gestion du risque sous l'angle de la responsabilité sociale: le cas des entreprises seveso
Robert, Jocelyne ULg; Frau, Maité; al

in Boudrandi, Stéphane; Delaye, Richard; Peretti, Jean-Marie (Eds.) L'audit social au service du progrès du management et de la bonne gouvernance-Proceedings (2014, August 29)

Detailed reference viewed: 6 (0 ULg)
Full Text
Peer Reviewed
See detailA post-mortem study of physiopathological characteristics of genital tracts of culled dairy cows in South Vietnam.
Nguyen Kien, Cuong; Okouyi, Marcel; Hanzen, Christian ULg

Poster (2014, August 28)

Detailed reference viewed: 6 (2 ULg)
Full Text
See detailL'odeur : question de sens ou de chimie
Romain, Anne-Claude ULg

Scientific conference (2014, August 28)

Afin de mesurer au mieux un paramètre, il faut préalablement connaître son identité. Concernant la métrologie de la pollution olfactive, toute la difficulté réside dans la particularité de cette pollution ... [more ▼]

Afin de mesurer au mieux un paramètre, il faut préalablement connaître son identité. Concernant la métrologie de la pollution olfactive, toute la difficulté réside dans la particularité de cette pollution. En effet, bien qu’associée à la pollution de l’air par les molécules volatiles qui la génèrent, elle en diffère par son effet sensoriel. Il va de soi que la pollution olfactive n’a de sens que par son odeur. Chaque individu a son propre odorat et dès lors sa propre interprétation de l’odeur. C’est cette subjectivité qui rend cette pollution si délicate à qualifier et à quantifier et qui la distingue de la pollution de l’air. De plus, contrairement à d’autres pollutions sensorielles de type sonore ou visuel, l’élément porteur ’est pas dans ce cas une grandeur physique telle que la longueur d’onde électromagnétique ou la variation de pression mais une grandeur physico-chimique particulière dont la mesure est beaucoup plus complexe. Un dictionnaire de langue française définit l’odeur comme « une émanation volatile, caractéristique de certains corps et susceptible de provoquer chez l’homme ou chez un animal des sensations dues à l’excitation d’organes spécialisés ». Cette définition de l’odeur met notamment en jeu deux disciplines : la chimie et la physiologie. En effet un chimiste pourrait présenter l’odeur comme un ensemble de molécules aux propriétés spécifiques qui inhalées par un individu génère une réaction, il s’intéresserait dès lors aux odorants alors qu’un physiologiste parlerait plutôt du stimulus induit par ces odorants et des mécanismes qui le provoquent au niveau des organes. Le traitement de l’information olfactive par le cerveau, la perception olfactive, conduit à plusieurs types de sensation. En effet, l’odeur perçue a trois grandes particularités : l’intensité, la qualité (ou la nature de l’odeur) et la tonalité hédonique ou affective (agrément/désagrément). A ces trois dimensions, il faut ajouter le caractère temporel du stimulus, dimension très utilisée dans le cadre de la pollution olfactive. [less ▲]

Detailed reference viewed: 10 (2 ULg)
See detailBayesian approach integrating correlated foreign information into a multivariate genetic evaluation
Vandenplas, Jérémie ULg; Gengler, Nicolas ULg

in Book of Abstracts of the 65th Annual Meeting of the European Federation of Animal Science (2014, August 28)

High-perfomance computing facilities proposing shared-memory and distributed-memory multiprocessors are becoming available. With those clusters, parallel computing could lead to increased performances and ... [more ▼]

High-perfomance computing facilities proposing shared-memory and distributed-memory multiprocessors are becoming available. With those clusters, parallel computing could lead to increased performances and problem sizes. However, to our knowledge and especially for variance components estimations, most software available in animal breeding, based on sparse matrices computations, do not allow parallel computing and are limited by memory accessible by the central processing unit, or allow parallel computing only for options with dense matrices computations, which limits anyway problem sizes due to storage of dense matrices. The aim was to propose simple and effective modifications for the BLUPF90 family of programs to reduce computing time with consideration of required memory. Modifications were based on academic free packages proposing solver and sparse inversion for sparse symmetric indefinite linear systems. First, modifications concerned the sparse inversion subroutine implemented in the package FSPAK. Rearrangements of 'do' loops to allow optimizations of computer operations by some compilers and addition of OpenMP directives were performed. The ordering operation was modified to more easily compare a multiple minimum degree algorithm (MMD; implemented in FSPAK) and a multilevel nested dissection algorithm (implemented in METIS 4.0.3). Second, the package PARDISO Version 5.0.0 was used instead of FSPAK. This package proposes in particular a parallel solver and sparse inversion on shared-memory multiprocessors. Modified FSPAK and PARDISO were compared to original FSPAK using MMD through REMLF90. Different models, such as univariate or bivariate (random regressions) test-day animal and single-step genomic models, were tested. All jobs were run 5 times. With an appropriate ordering algorithm, speedup for each REMLF90 iteration were up to 7.5 for modified FSPAK and up to 22.8 for PARDISO with 2 threads. With 4 threads, speedup increased to 8.3 and 32.5, respectively. [less ▲]

Detailed reference viewed: 5 (2 ULg)
Full Text
See detailUse of high performance computing in animal breeding
Vandenplas, Jérémie ULg; Gengler, Nicolas ULg

Poster (2014, August 28)

High-perfomance computing facilities proposing shared-memory and distributed-memory multiprocessors are becoming available. With those clusters, parallel computing could lead to increased performances and ... [more ▼]

High-perfomance computing facilities proposing shared-memory and distributed-memory multiprocessors are becoming available. With those clusters, parallel computing could lead to increased performances and problem sizes. However, to our knowledge and especially for variance components estimations, most software available in animal breeding, based on sparse matrices computations, do not allow parallel computing and are limited by memory accessible by the central processing unit, or allow parallel computing only for options with dense matrices computations, which limits anyway problem sizes due to storage of dense matrices. The aim was to propose simple and effective modifications for the BLUPF90 family of programs to reduce computing time with consideration of required memory. Modifications were based on academic free packages proposing solver and sparse inversion for sparse symmetric indefinite linear systems. First, modifications concerned the sparse inversion subroutine implemented in the package FSPAK. Rearrangements of 'do' loops to allow optimizations of computer operations by some compilers and addition of OpenMP directives were performed. The ordering operation was modified to more easily compare a multiple minimum degree algorithm (MMD; implemented in FSPAK) and a multilevel nested dissection algorithm (implemented in METIS 4.0.3). Second, the package PARDISO Version 5.0.0 was used instead of FSPAK. This package proposes in particular a parallel solver and sparse inversion on shared-memory multiprocessors. Modified FSPAK and PARDISO were compared to original FSPAK using MMD through REMLF90. Different models, such as univariate or bivariate (random regressions) test-day animal and single-step genomic models, were tested. All jobs were run 5 times. With an appropriate ordering algorithm, speedup for each REMLF90 iteration were up to 7.5 for modified FSPAK and up to 22.8 for PARDISO with 2 threads. With 4 threads, speedup increased to 8.3 and 32.5, respectively. [less ▲]

Detailed reference viewed: 5 (0 ULg)
Full Text
See detailInnovative lactation stage specific prediction of CH4 from milk MIR spectra
Vanlierde, Amélie; Vanrobays, Marie-Laure ULg; Dehareng, Frédéric et al

Conference (2014, August 28)

Detailed reference viewed: 11 (2 ULg)
Full Text
See detailInteractions in massive binaries: spectral modelling – CoMBiSpeC – and observational analyses
Palate, Matthieu ULg

Doctoral thesis (2014)

Massive stars are very important to the life of the galaxiesbut they are rare and many open questions on their formation, evolution and characteristics remain. One of the best ways to improve our ... [more ▼]

Massive stars are very important to the life of the galaxiesbut they are rare and many open questions on their formation, evolution and characteristics remain. One of the best ways to improve our knowledge on massive stars is to study massive binaries (that represent at least 50% of massive stars see Mahy et al. 2009, and Sana & Evans 2011) and their spectra. Spectral and atmosphere modelling of stars are very interesting tools that are now often used in spectral analyses. However, these models are designed for single spherical stars. In this context, we have developed a numerical code of spectral modelling of massive binary systems: CoMBiSpeC (code of massive binary spectral computation) that computes the spectra of massive binary systems and accounts for various effects that affect these systems. Our models, and the observational analyses of the four binary systems presented in the second part of this work, have allowed to explain several effects linked to binarity: luminosity class discrepancy, temperature distribution, Struve-Sahade effect, and some kind of line profile variations. In summary, CoMBiSpeC is a first step in the spectral modelling of massive binary systems and several improvements are still needed to be able to study all massive binaries. Nevertheless, this model can explain and reproduce various observed phenomena which was the main goal of this thesis. [less ▲]

Detailed reference viewed: 23 (10 ULg)
Full Text
See detailIntroduction à la Théorie générale de l'Etat - Manuel (3e édition)
Behrendt, Christian ULg; Bouhon, Frédéric ULg

Book published by Larcier (2014)

Detailed reference viewed: 114 (35 ULg)
Full Text
Peer Reviewed
See detailNeutrophil contribution to spinal cord injury and repair
Neirinckx, Virginie ULg; Coste, Cécile ULg; Franzen, Rachelle ULg et al

in Journal of Neuroinflammation (2014), 11(1),

Spinal cord injuries remain a critical issue in experimental and clinical research nowadays, and it is now well accepted that the immune response and subsequent inflammatory reactions are of significant ... [more ▼]

Spinal cord injuries remain a critical issue in experimental and clinical research nowadays, and it is now well accepted that the immune response and subsequent inflammatory reactions are of significant importance in regulating the damage/repair balance after injury. The role of macrophages in such nervous system lesions now becomes clearer and their contribution in the wound healing process has been largely described in the last few years. Conversely, the contribution of neutrophils has traditionally been considered as detrimental and unfavorable to proper tissue regeneration, even if there are very few studies available on their precise impact in spinal cord lesions. Indeed, recent data show that neutrophils are required for promoting functional recovery after spinal cord trauma. In this review, we gathered recent evidence concerning the role of neutrophils in spinal cord injuries but also in some other neurological diseases, highlighting the need for further understanding the different mechanisms involved in spinal cord injury and repair. [less ▲]

Detailed reference viewed: 5 (0 ULg)
Peer Reviewed
See detailMeasuring consciousness in coma and related states
Di Perri, Carol; Thibaut, Aurore ULg; Soddu, Andrea ULg et al

in World Journal of Radiology (2014), 6(8),

Detailed reference viewed: 7 (3 ULg)
Full Text
See detailPrediction of meso-scale mechanical properties of poly-silicon materials
Lucas, Vincent ULg; Wu, Ling ULg; Arnst, Maarten ULg et al

Conference (2014, August 27)

The miniature sizes of micro–electro–mechanical systems (MEMS) as well as the nature of their manufacturing processes, such as etching, material layer deposition, or embossing, are responsible for the ... [more ▼]

The miniature sizes of micro–electro–mechanical systems (MEMS) as well as the nature of their manufacturing processes, such as etching, material layer deposition, or embossing, are responsible for the existence of a scatter in the final dimensions, material properties ... of manufactured micro–sensors. This scatter is potentially threatening the behavior and reliability of samples from a batch fabrication process, motivating the development of non-deterministic computational approaches to predict the MEMS properties. In this work we extract the meso-scale properties of the poly-silicon material under the form of a probabilistic distribution. To this end, Statistical Volume Elements (SVE) of the micro-structure are generated under the form of a Voronoï tessellation with a random orientation for each silicon grain. Hence, a Monte-Carlo procedure combined with a homogenization technique allows a distribution of the material tensor at the meso-scale to be estimated. As the finite element method is used to discretize the SVE and to solve the micro-scale boundary value problem, the homogenization technique used to extract the material tensor relies on the computational homogenization theory. In a future work, we will investigate, in the context of MEMS vibrometers, the propagation to the macro–scale of the meso-scale distribution of the homogenized elasticity tensor, with the final aim of predicting the uncertainty on their resonance frequencies. [less ▲]

Detailed reference viewed: 16 (7 ULg)
Full Text
Peer Reviewed
See detailAntibiotic resistance trend of Staphylococcus aureus isolated between 2010 and 2012 from mastitis cases in Azawak zebu in Niger
Issa Ibrahim, Abdoulkarim ULg; Duprez, Jean-Noël ULg; Bada-Alambedji, Rianatou et al

in African Journal of Microbiology Research (2014), 8((35)), 3271-3275

Detailed reference viewed: 11 (4 ULg)
Full Text
See detailDiagnostic de sites et méthode d’implantation de tours pour une insertion urbaine de qualité
Saint-Pierre, Claire ULg

Doctoral thesis (2014)

The matter of towers construction in urban areas has been coming back among topical subjects, taking part in the questioning of sustainable urban development. While the debate still remains as regards ... [more ▼]

The matter of towers construction in urban areas has been coming back among topical subjects, taking part in the questioning of sustainable urban development. While the debate still remains as regards towers legitimacy, location choices are primarily driven by economic considerations (funding, real estate availability) and the issue of urban insertion is not always tackled at the right time. The decision-making process of tower construction is linear and takes into account some criteria relatively late even though if it was done earlier they could anticipate the impact of the tower on its urban environment. Using several comparative analysis (benchmarking), we identified the inherent criteria in this decision-making process. Their analysis and classification allowed us to highlight the preponderance of six criteria and the presence of significant interaction between them: the functionality, the shape, the movement, the public space, the environment and the townscape. The relations which exist between criteria reflect the complexity of the tower location process for which the most appropriate study approach seems to be the systemic. Between the multi-criteria analysis and the systemic approach, our implantation method of towers offers to consider these criteria at the very beginning of the project. Thus, it allows to draw up a list of indicators of urban integration which evaluation will indicate the quality of tower insertion in the urban environment. By considering upstream the identified criteria and also in a regular and diffuse way, the decision-making process of tower construction will be more efficient and effective to help finding the right location for a future tall building project. This implantation method is also designed to give rise discussions and reflections on the topics tackled, so that the dialogue between the actors turns towards a better location of towers in urban areas. [less ▲]

Detailed reference viewed: 26 (6 ULg)