Bordeaux INP

Cours-la-Ville, France

Bordeaux INP

Cours-la-Ville, France
SEARCH FILTERS
Time filter
Source Type

Lemoine A.,University of Bordeaux 1 | Glockner S.,Bordeaux INP | Breil J.,CEA Cesta
Journal of Computational Physics | Year: 2017

Moment-of-Fluid (MoF) is a piecewise linear interface reconstruction method that tracks fluid through its volume fraction and centroid, which are deduced from the zeroth and first moments. We present a method that replaces the original minimization stage by an analytic reconstruction algorithm on bi-dimensional Cartesian grids. This algorithm provides accurate results for a lower computational cost than the original minimization algorithm. When more than two fluids are involved, this algorithm can be used coupled with the minimization algorithm. Although this paper deals with Cartesian grids, everything remains valid for any meshes that are made of rectangular cells. © 2016 Elsevier Inc.


Re B.,Polytechnic of Milan | Dobrzynski C.,Bordeaux INP | Dobrzynski C.,French Institute for Research in Computer Science and Automation | Guardone A.,Polytechnic of Milan
Journal of Computational Physics | Year: 2017

A novel strategy to solve the finite volume discretization of the unsteady Euler equations within the Arbitrary Lagrangian–Eulerian framework over tetrahedral adaptive grids is proposed. The volume changes due to local mesh adaptation are treated as continuous deformations of the finite volumes and they are taken into account by adding fictitious numerical fluxes to the governing equation. This peculiar interpretation enables to avoid any explicit interpolation of the solution between different grids and to compute grid velocities so that the Geometric Conservation Law is automatically fulfilled also for connectivity changes. The solution on the new grid is obtained through standard ALE techniques, thus preserving the underlying scheme properties, such as conservativeness, stability and monotonicity. The adaptation procedure includes node insertion, node deletion, edge swapping and points relocation and it is exploited both to enhance grid quality after the boundary movement and to modify the grid spacing to increase solution accuracy. The presented approach is assessed by three-dimensional simulations of steady and unsteady flow fields. The capability of dealing with large boundary displacements is demonstrated by computing the flow around the translating infinite- and finite-span NACA 0012 wing moving through the domain at the flight speed. The proposed adaptive scheme is applied also to the simulation of a pitching infinite-span wing, where the bi-dimensional character of the flow is well reproduced despite the three-dimensional unstructured grid. Finally, the scheme is exploited in a piston-induced shock-tube problem to take into account simultaneously the large deformation of the domain and the shock wave. In all tests, mesh adaptation plays a crucial role. © 2017 The Authors


Jeannot E.,French Institute for Research in Computer Science and Automation | Mercier G.,Bordeaux INP | Tessier F.,French Institute for Research in Computer Science and Automation
Proceedings of COM-HPC 2016: 1st Workshop on Optimization of Communication in HPC Runtime Systems - Held in conjunction with SC 2016: The International Conference for High Performance Computing, Networking, Storage and Analysis | Year: 2016

The evolution of massively parallel supercomputers make palpable two issues in particular: the load imbalance and the poor management of data locality in applications. Thus, with the increase of the number of cores and the drastic decrease of amount of memory per core, the large performance needs imply to particularly take care of the load-balancing and as much as possible of the locality of data. One mean to take into account this locality issue relies on the placement of the processing entities and load balancing techniques are relevant in order to improve application performance. With large-scale platforms in mind, we developed a hierarchical and distributed algorithm which aim is to perform a topology-aware load balancing tailored for Charm++ applications. This algorithm is based on both LibTopoMap for the network awareness aspects and on TREEMATCH to determine a relevant placement of the processing entities. We show that the proposed algorithm improves the overall execution time in both the cases of real applications and a synthetic benchmark as well. For this last experiment, we show a scalability up to one millions processing entities. © 2016 IEEE.


Georgiou Y.,ATOS Bull | Jeannot E.,French Institute for Research in Computer Science and Automation | Mercier G.,Bordeaux INP | Villiermet A.,French Institute for Research in Computer Science and Automation
ACM International Conference Proceeding Series | Year: 2017

The Resource and JobManagement System (RJMS) is a crucial system software part of the HPC stack. It is responsible for efficiently delivering computing power to applications in supercomputing environments. Its main intelligence relies on resource selection techniques to find the most adapted resources to schedule the users' jobs. Improper resource selection operations may lead to poor performance executions and global system utilization along with an increase of the system fragmentation and jobs starvation. These phenomena play a role in the increase of the platforms' total cost of ownership and should be minimized. This paper introduces a new method that takes into account the topology of the machine and the application characteristics to determine the best choice among the available nodes of the platform based upon their position within the network and taking into account the applications communication pattern. To validate our approach, we integrate this algorithm as a plugin for Slurm, a popular and widespread HPC resource and job management system (RJMS). We assess our plugin with different optimization schemes by comparing with the default topology-aware Slurm algorithm using both emulation and simulation of a large-scale platform, and by carrying out experiments in a real cluster. We show that transparently taking into account the job communication pattern and the topology allows for relevant performance gains.


Zbrzeski A.,Bordeaux INP | Zbrzeski A.,University of Bordeaux 1 | Bornat Y.,Bordeaux INP | Bornat Y.,University of Bordeaux 1 | And 6 more authors.
Frontiers in Neuroscience | Year: 2016

Cervical spinal cord injury can disrupt connections between the brain respiratory network and the respiratory muscles which can lead to partial or complete loss of ventilatory control and require ventilatory assistance. Unlike current open-loop technology, a closed-loop diaphragmatic pacing system could overcome the drawbacks of manual titration as well as respond to changing ventilation requirements. We present an original bio-inspired assistive technology for real-time ventilation assistance, implemented in a digital configurable Field Programmable Gate Array (FPGA). The bio-inspired controller, which is a spiking neural network (SNN) inspired by the medullary respiratory network, is as robust as a classic controller while having a flexible, low-power and low-cost hardware design. The system was simulated in MATLAB with FPGA-specific constraints and tested with a computational model of rat breathing; the model reproduced experimentally collected respiratory data in eupneic animals. The open-loop version of the bio-inspired controller was implemented on the FPGA. Electrical test bench characterizations confirmed the system functionality. Open and closed-loop paradigm simulations were simulated to test the FPGA system real-time behavior using the rat computational model. The closed-loop system monitors breathing and changes in respiratory demands to drive diaphragmatic stimulation. The simulated results inform future acute animal experiments and constitute the first step toward the development of a neuromorphic, adaptive, compact, low-power, implantable device. The bio-inspired hardware design optimizes the FPGA resource and time costs while harnessing the computational power of spike-based neuromorphic hardware. Its real-time feature makes it suitable for in vivo applications. © 2016 Zbrzeski, Bornat, Hillen, Siu, Abbas, Jung and Renaud.


Ronnow D.,University of Gävle | Bjorsell N.,University of Gävle | Laporte-Fauret B.,Bordeaux INP
I2MTC 2017 - 2017 IEEE International Instrumentation and Measurement Technology Conference, Proceedings | Year: 2017

Short range synthetic aperture radar (SAR) was used to study electrically small objects. A metallic rod and a metallic sphere could not be separated in the SAR images. Polarimetric SAR images were analyzed and images corresponding to different antenna orientations were obtained by applying rotation matrices to radar data. The target intensity varied with the rotation angle. Elongation and orientation of the objects could be determined from the ratio of minimum and maximum intensity. Upper and lower limits for measurable elongation depend on measurement errors. © 2017 IEEE.


Mellor A.,RMIT University | Boukir S.,Bordeaux INP
ISPRS Journal of Photogrammetry and Remote Sensing | Year: 2017

Ensemble classifiers, such as random forests, are now commonly applied in the field of remote sensing, and have been shown to perform better than single classifier systems, resulting in reduced generalisation error. Diversity across the members of ensemble classifiers is known to have a strong influence on classification performance - whereby classifier errors are uncorrelated and more uniformly distributed across ensemble members. The relationship between ensemble diversity and classification performance has not yet been fully explored in the fields of information science and machine learning and has never been examined in the field of remote sensing. This study is a novel exploration of ensemble diversity and its link to classification performance, applied to a multi-class canopy cover classification problem using random forests and multisource remote sensing and ancillary GIS data, across seven million hectares of diverse dry-sclerophyll dominated public forests in Victoria Australia. A particular emphasis is placed on analysing the relationship between ensemble diversity and ensemble margin - two key concepts in ensemble learning. The main novelty of our work is on boosting diversity by emphasizing the contribution of lower margin instances used in the learning process. Exploring the influence of tree pruning on diversity is also a new empirical analysis that contributes to a better understanding of ensemble performance. Results reveal insights into the trade-off between ensemble classification accuracy and diversity, and through the ensemble margin, demonstrate how inducing diversity by targeting lower margin training samples is a means of achieving better classifier performance for more difficult or rarer classes and reducing information redundancy in classification problems. Our findings inform strategies for collecting training data and designing and parameterising ensemble classifiers, such as random forests. This is particularly important in large area remote sensing applications, for which training data is costly and resource intensive to collect. © 2017 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS)


News Article | December 22, 2016
Site: phys.org

Confocal microscope observations of three types of vesicles. In red, methylene blue-loaded vesicles; in green, calcein-loaded vesicles. Sucrose-loaded vesicles are indicated by an arrow. Laser irradiation of 633 nm then 488 nm ruptures the red and green vesicles in succession. The other vesicle remains intact. Credit: Peyret et al. 2016. Cells are the site of a multitude of chemical reactions, the precision of which is envied by scientists. A team of researchers from the CNRS and Bordeaux INP have neared this level of control by controlling the explosion of polymersomes through laser irradiation. These hollow polymer spheres, which can mimic certain cellular functions, react to a specific wavelength and thus release their content on demand. This research has been published in Angewandte Chemie International Edition. Polymersomes are artificial vesicles that can mimic organelles, compartments naturally found in nucleic cells. Here researchers encapsulated fluorescent molecules in "giant" polymersomes with a diameter of a dozen micrometers. These fluorescent groups have the characteristic of decomposing under the action of light, but only at a specific wavelength. A suitable level of irradiation degrades these molecules and thus increases the solute concentration within the polymersomes. This entails an imbalance, which, because the polymersomes are relatively impermeable, cannot be compensated sufficiently rapidly. The vesicles are thus forced to rupture. The team designed three types of polymersomes, each containing a specific fluorescent group, meant to react to different wavelengths. The researchers attained such a level of control that they were able to observe the vesicles and target them individually using a confocal microscope equipped with adequate lasers. In addition to light, other methods of control are currently being developed by researchers to rupture microvesicles: temperature, pH, magnetic fields and so on. This research could have medical applications in the long term, but for the time being researchers are studying the possibility of releasing substances in a controlled manner within artificial polymer cells, in order to be able to reproduce and better understand some of the metabolic reactions of the biological cell. More information: Ariane Peyret et al. Polymersome Popping by Light-Induced Osmotic Shock under Temporal, Spatial, and Spectral Control, Angewandte Chemie (2016). DOI: 10.1002/ange.201609231


Falleri J.-R.,Bordeaux INP | Reveillere E.G.E.L.,Bordeaux INP
2015 23rd European Signal Processing Conference, EUSIPCO 2015 | Year: 2015

This paper deals with our positive experience about project-based pedagogy with the help of industrial partners to teach signal and image processing. During one semester, students are working in small groups of 6 to 8 students, supervised by two teachers or engineers working in a small or a big com pany. Various topics are proposed each year such as radar pro cessing, mobile communication system or image processing. The role played by the industrial partners is crucial: they give seminars about program management, they evaluate the tech nical quality of the projects and the clarity of the oral presen tation. An award ceremony is also organized at school during which the activities of the companies are presented. There are also some discussions about the activities of a young engineer and several awards in various categories are given. A cocktail party ends up the day. Anonymous online surveys completed by our students as well as discussions with our partners con firm the relevance of these projects. © 2015 EURASIP.


Feng W.,Bordeaux INP | Boukir S.,Bordeaux INP
Proceedings - International Conference on Image Processing, ICIP | Year: 2015

Mislabeled training data is a challenge to face in order to build a robust classifier whether it is an ensemble or not. This work handles the mislabeling problem by exploiting four different ensemble margins for identifying, then eliminating or correcting the mislabeled training data. Our approach is based on class noise ordering and relies on the margin values of misclassified data. The effectiveness of our ordering-based class noise removal and correction methods is demonstrated in performing image classification. A comparative analysis is conducted with respect to the majority vote filter, a reference ensemble-based class noise filter. © 2015 IEEE.

Loading Bordeaux INP collaborators
Loading Bordeaux INP collaborators