Éditeur(s) : HAL CCSD
Résumé : Climate change is one of the biggest public health threats in the 21th century. An increase in temperatures will lead to an increase in mortality attributable to temperature. In addition, some populations and territories are particularly vulnerable to the impact of increases in heat. It is thus necessary to identify these populations and territories as well as examine future heat-Related health impacts in order to recommend equity-Oriented policies today and in the future. The general objective of this thesis is to document current and future heat-Related vulnerability factors in the context of climate change. In order to address this general objective, the thesis involved four components: a) to conduct a systematic review and a meta-Analysis to assess the heterogeneity in the heat-Mortality associations with respect to individual and contextual population characteristics; b) to identify whether and how the magnitude of mean temperature effects on all-Cause mortality were modified by chronic air pollution exposure, social deprivation, and a combination of these two dimensions; c) to develop a method to quantify the climate change impacts on heat-Related mortality using climate modeling; d) to assess historical and future social disparities in years of life lost caused by ambient temperature in Montreal and Paris, and compare these estimates as well as the impact of climate change on social disparities between the two cities. This thesis highlights which populations are more vulnerable to heat and shows that several differences exist with regard to guidelines from international public health institutions for the identification of vulnerable populations. This thesis also identified chronic air pollution exposure as a new vulnerability factor in heat-Related mortality and that it has a double interaction with social deprivation. Furthermore, in this thesis a novel method to quantify future heat-Related mortality was developed which emphasized the strong evidence of an increase in heat-Related mortality under climate change. This method was then applied to estimate the increase in daily years of life lost social disparities in both Montreal and Paris under climate change which showed that this increase would be greater in Montreal compared to Paris in the future. Thus, this thesis which used a variety of epidemiologic methods has clarified which populations are particularly vulnerable to heat impacts and challenges guidelines for the identification of vulnerable populations from international public health institutions. It has also highlighted the climate change impacts on health inequalities and aims to reorient equity-Focused policies.
Les changements climatiques constituent l’un des enjeux les plus importants en santé publique du XXIème siècle. En effet, une hausse des températures provoquera une hausse de la mortalité attribuable à la chaleur. En outre, certaines populations et territoires sont particulièrement vulnérables aux effets de la chaleur. Il est donc nécessaire de les identifier clairement ainsi que les impacts futurs pour orienter avec équité les politiques publiques aujourd’hui et à l’avenir. L’objectif de cette thèse est de documenter les facteurs de vulnérabilité à la chaleur dans le présent et les éléments permettant leur prise en compte pour l’avenir dans le contexte des changements climatiques. Quatre étapes ont été menées pour répondre à cet objectif : a) Mener une revue systématique et une méta-Analyse des facteurs de vulnérabilité face aux risques de mortalité en lien avec la chaleur ; b) Analyser si l’exposition chronique à la pollution atmosphérique modifie la relation entre la chaleur et la mortalité dans un contexte urbain (Paris) et explorer la double interaction avec la défaveur sociale ; c) Développer une méthode de quantification des impacts liés à la chaleur en lien avec les changements climatiques en intégrant une grande diversité de simulations climatiques ; d) Estimer les inégalités d’années de vies perdues attribuables à la chaleur et leurs projections futures dans deux contextes distincts (Montréal et Paris) et comparer ces inégalités. Cette thèse a permis de mettre en évidence sur la base de la littérature épidémiologique les groupes de population qui sont les plus vulnérables face aux effets de la chaleur et de montrer qu’il y a plusieurs divergences par rapport aux recommandations émises par les institutions de santé publique vis-À-Vis de l’identification de populations vulnérables. Cette thèse a permis également d’identifier l’exposition chronique à la pollution atmosphérique comme nouveau facteur de vulnérabilité à la chaleur et que cette vulnérabilité était encore plus prononcée lorsqu’il s’agissait de populations défavorisées socialement. Puis, une méthode permettant de quantifier l’impact des changements climatiques sur les décès attribuables à la chaleur et ses sources d’incertitudes a été développée, et a permis de mettre en évidence que la probabilité que les changements climatiques conduisent à une augmentation de la mortalité en lien avec la chaleur est très forte. Cette méthode a ensuite été utilisée pour estimer que l’augmentation de la température conduira à une augmentation des inégalités sociales d’années de vie perdues à la fois à Montréal et à Paris, l’effet des changements climatiques sur l’accroissement de ces inégalités étant plus fort à Montréal qu’à Paris. Cette thèse, en se basant sur diverses méthodes épidémiologiques, a permis dans l’ensemble de clarifier quelles populations étaient particulièrement à risque face aux effets de la chaleur et de questionner les recommandations émises par les organismes tels que l’OMS. Elle a également permis de montrer l’effet des changements climatiques sur l’évolution de vulnérabilités face à la chaleur pour inciter dès aujourd’hui la mise en place de politiques publiques équitables et limiter l’impact des changements climatiques sur l’accroissement des inégalités de santé.
https://tel.archives-ouvertes.fr/tel-01456772

Droits : info:eu-repo/semantics/OpenAccess
NNT : 2015PA11T003
tel-01456772
https://tel.archives-ouvertes.fr/tel-01456772
https://tel.archives-ouvertes.fr/tel-01456772/document
https://tel.archives-ouvertes.fr/tel-01456772/file/VA_BENMARHNIA_TARIK_05022015.pdf

Éditeur(s) : HAL CCSD
Résumé : International audience
Dans ce papier, nous proposons une solution qui exploite à la fois des informations sur la structure du réseau et sur les attributs des noeuds, pour extraire un nouveau type de motifs que nous appelons "liens conceptuels". Les liens conceptuels sont des représentations synthétiques d'ensembles de liens entre des groupes de noeuds, au sein desquels les noeuds partagent des attributs communs. Nous montrons que l'ensemble des liens conceptuels extraits à partir d'un réseau social apporte une connaissance pertinente sur les types de noeuds les plus connectés, et fournit une "vue conceptuelle" du réseau. L'efficacité de notre solution est démontrée à travers des résultats expérimentaux.
Dans ce papier, nous proposons une solution qui exploite à la fois des informations sur la structure du réseau et sur les attributs des noeuds, pour extraire un nouveau type de motifs que nous appelons "liens conceptuels". Les liens conceptuels sont des r
Paris, France

hal-00786167
https://hal.archives-ouvertes.fr/hal-00786167

Éditeur(s) : HAL CCSD Wiley
Résumé : International audience
This paper describes how to tackle new challenging coastal engineering problems related to beach erosion with a shape optimization approach. The method modifies the shape of the sea bottom in order to reduce beach erosion effects. Global optimization is shown to be necessary as the related functionals have several local minima. We describe the physical model used, the proposed protection devices against beach erosion and real case applications.
ISSN: 0029-5981

hal-00411912
https://hal.archives-ouvertes.fr/hal-00411912
DOI : 10.1002/nme.2209

Éditeur(s) : HAL CCSD American Geophysical Union
Résumé : International audience
The northern Canadian Cordillera (NCC) is an active orogenic belt in northwestern Canada characterized by deformed autochtonous and allochtonous structures that were emplaced in successive episodes of convergence since the Late Cretaceous. Seismicity and crustal deformation are concentrated along corridors located far (>200 to ~800 km) from the convergent plate margin. Proposed geodynamic models require information on crust and mantle structure and strain history, which are poorly constrained. We calculate receiver functions using 66 broadband seismic stations within and around the NCC and process them to estimate Moho depth and P-to-S velocity ratio (Vp/Vs) of the Cordilleran crust. We also perform a harmonic decomposition to determine the anisotropy of the subsurface layers. From these results, we construct simple seismic velocity models at selected stations and simulate receiver function data to constrain crust and uppermost mantle structure and anisotropy. Our results indicate a relatively flat and sharp Moho at 32 ± 2 km depth and crustal Vp/Vs of 1.75 ± 0.05. Seismic anisotropy is pervasive in the upper crust and within a thin (~10–15 km thick) sub-Moho layer. The modeled plunging slow axis of hexagonal symmetry of the upper crustal anisotropic layer may reflect the presence of fractures or mica-rich mylonites. The subhorizontal fast axis of hexagonal anisotropy within the sub-Moho layer is generally consistent with the SE-NW orientation of large-scale tectonic structures. These results allow us to revise the geodynamic models proposed to explain active deformation within the NCC.
ISSN: 2169-9313

Droits : info:eu-repo/semantics/OpenAccess
hal-01685550
https://hal.archives-ouvertes.fr/hal-01685550
https://hal.archives-ouvertes.fr/hal-01685550/document
https://hal.archives-ouvertes.fr/hal-01685550/file/Tarayoun_et_al-2017-Journalsur Geophysical_Research__Solid_Earth.pdf
DOI : 10.1002/2017JB014284

Éditeur(s) : HAL CCSD
Résumé : International audience
This paper proposes a new knowledge discovery method called FLMin to discover frequent patterns in a social network. The algorithm works without previous knowledge on the network and exploits both the structure and the attributes of nodes to extract regularities called Frequent Links. Unlike traditional works in this area that solely exploit structural regularities of the network, the originality of FLMin is its ability to gather these two kinds of information in the search for patterns. In this paper, we detail the method proposed for extracting frequent links and discuss its complexity and its flexibility. The efficiency of our solution is evaluated by conducting qualitative and quantitative studies for understanding how behaves FLMin according to different parameters.
Networked Digital Technologies
Dubai, United Arab Emirates

hal-00768431
https://hal.archives-ouvertes.fr/hal-00768431
DOI : 10.1007/978-3-642-30567-2_38

Éditeur(s) : HAL CCSD Wiley
Résumé :  
Animals, especially insects, are principal pollen vectors of tropical trees and have behavior patterns that affect gene dispersal. Here, we explore complex pollination systems using a new simulation model Eco-Gene and considering, among other factors, flowering synchrony, spatial distribution of trees, degree of selfing, population densities, pollinator flight distances, pollen deposition, and pollinator response to floral display size. Sensitivity analyses using two contrasting tree data sets (Jacaranda copaia and Dipteryx odorata) determined the importance of each parameter on three response variables: the proportion of seeds from self-pollination, effective population size, and pollen dispersal. Spatial considerations and attractiveness of floral displays were prominent features determining the population genetic result of pollinators, and some biological implications of the results are discussed.
ISSN: 0006-3606

hal-01032091
https://hal.archives-ouvertes.fr/hal-01032091
DOI : 10.1111/j.1744-7429.2004.tb00309.x

Éditeur(s) : HAL CCSD Presses polytechniques et universitaires romandes, Publisher of the EPFL Press
Résumé : Chacun peut observer dans le monde réel l’omniprésence de systèmes complexes constitués de nombreux éléments en interaction et dont les caractéristiques globales ne peuvent se réduire à celles de leurs composants : ce sont par exemple les cellules d’un organisme vivant, les colonies de fourmis, les écosystèmes, ou encore les systèmes économiques, l’Internet et les réseaux sociaux. Dans le futur, nos sociétés devront relever l’immense défi scientifique et technologique qui consiste à comprendre ces systèmes pour les adapter, les contrôler et les modéliser. Au travers d’exemples choisis sans complication inutile, les auteurs proposent une introduction par la pratique aux systèmes complexes. Leur méthode repose sur la conviction que toute compréhension suppose un acte d’expérience; l’ouvrage est composé d’ateliers où il s’agit de simuler une dynamique collective à partir de sa modélisation orientée-agent. Les étudiants, les enseignants, les créateurs et plus largement toute personne avide de développer sa culture et d’actualiser ses connaissances tireront parti de cet ouvrage pluridisciplinaire accessible à un large public.
https://hal.archives-ouvertes.fr/hal-01284628

ISBN : 978-2-88074-991-0
hal-01284628
https://hal.archives-ouvertes.fr/hal-01284628

Éditeur(s) : HAL CCSD
Résumé : International audience
Numerous social network mining methods have been proposed until now for addressing social mining tasks and specially searching for communities or frequent social patterns. However, the degree of complementarity of these methods has been very little studied. In this paper, we focus on two knowledge extraction processes in social networks: a link-based clustering that may extract social communities and a recent approach that searches for frequent conceptual link involving both clustering and search for frequent social patterns. We explore how the models extracted by each method may match and which potential useful knowledge they may provide. Our objective is to evaluate the potential relationships between communities and frequent conceptual links. For this purpose, we propose a set of measures for evaluating the degree to which these patterns extracted from the same dataset are matching. Our approach is applied on two datasets and demonstrates the importance of considering simultaneously various kinds of knowledge and their complementarity.
IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining
Niagara falls, Canada

hal-00874351
https://hal.archives-ouvertes.fr/hal-00874351

Éditeur(s) : HAL CCSD IGI Global
Résumé : International audience
The social behavior of individuals is an important factor of the transmission and the evolution of many diseases. As such, epidemic studies have attempted to integrate social aspects in dissemination modeling. Since the pioneering works of Klovdahl on AIDS in 1985, epidemiological investigations and interventions increasingly focus on social networks. Significant factors of the transmission and outbreak of many infectious diseases are the structure and nature of human interactions. Network-based modeling approaches have found various applications in epidemiology as a simple yet efficient way to represent the complexity of human relationships implicated in dissemination processes. However, most results have been obtained by considering social networks as steady stage structures. Evolving networks have not been explored. The objective is first to give an overview of network-based modeling attempts in epidemiology to analyze and understand the dissemination processes with an emphasis on dynamic networks. The authors approach is designed to understand the impact of social links dynamics on epidemic spread. The authors present the results obtained by combining network evolution patterns (link creation and deletion) and a typical epidemic model. The speed of link dynamics and the infection time strongly influence the occurrence and value of the epidemic peak.
ISSN: 1947-8186

hal-00700105
https://hal.archives-ouvertes.fr/hal-00700105
DOI : 10.4018/jismd.2012070103

Éditeur(s) : HAL CCSD
Résumé : International audience
Although earthquakes and tsunamis are less frequent in the Dover Straitthan over active subduction zones, a plausible potential exists for intraplate earthquakes of magnitude Mw=6.9 generating a tsunami with damaging consequences. In April 1580, an earthquake shook the region violently and destructions were reported as far as London in the north andRouen in the south. Despite fair weather conditions, a series of abnormal sea waves was reported in several harbours (Calais, Boulogne and Dover) on the same day. A first step was to produce a range of numerical coseismic tsunami simulations and to compare them with historical witness accounts. Results raise the question of whether such earthquakes could also triggerchalk flow-generated tsunamis along cliff lines on both sides of the Strait. Gravity-driven collapses affect thechalk cliffs periodically, but local tsunami waves caused by very large mass movements could reach heights of several meters and, for example, strike Dover.
Proceedings of the 15th World Conference on Earthquake Engineering
Lisbonne, Portugal

halshs-01291496
https://halshs.archives-ouvertes.fr/halshs-01291496

Éditeur(s) : HAL CCSD
Résumé : Nous proposons dans cet article de mettre en avant quelques idées pour la simulation numérique de plusieurs classes d'équations d'ondes dispersives (KdV, Benjamin-Ono) ou non (Kuramoto-Sivashinski) par des méthodes multiniveaux. Nous partons des idées initialement développés pour les équations dissipatives et en proposons une extension. Il en ressort que la présence d'une régularisation en temps est nécessaire pour pouvoir traiter de manière stable, et avec des schémas appropriés, les différentes échelles de la solution, préalablement générées à l'aide de méthodes hierarchiques (inconnues incrémentales, ondelettes, Fourier)
https://hal.inria.fr/inria-00271154

inria-00271154
https://hal.inria.fr/inria-00271154
https://hal.inria.fr/inria-00271154/document
https://hal.inria.fr/inria-00271154/file/cclz.pdf

Éditeur(s) : HAL CCSD Springer Verlag
Résumé : Some major challenges for geophysicists and structural geologists using three-dimensional boundary element method codes (3D-BEM) are: (1) reducing the amount of memory required to solve large and dense systems and (2) incorporation of inequality constraints such as traction inequality constraints (TIC) and displacement inequality constraints (DIC). The latter serves two purposes. First, for example, inequality constraints can be used to simulate frictional slip (using TIC). Second, these constraints can prevent element interpenetration while allowing opening mode (using DIC). We have developed a method that simultaneously incorporates both types of functionality of the inequality constraints. We show that the use of an appropriate iterative solver not only avoids the allocation of significant memory for solving the system (allowing very large model computation and simplifying parallelization on multi-core processors), but also admits interesting features such as natural incorporation of TICs and DICs. Compared to other techniques of contact management (e.g., Lagrange multipliers, penalty method, or complementarity problem), this new simple methodology, which does not use any incremental trial-and-error procedures, brings more flexibility, while making the system more stable and less subject to round-off errors without any computational overhead. We provide validations and comparisons of the inequality constraints implementation using 2D analytical and numerical solutions.
ISSN: 1420-0597

hal-00561035
https://hal.archives-ouvertes.fr/hal-00561035
DOI : 10.1007/s10596-009-9170-x

Éditeur(s) : HAL CCSD Wiley
Résumé : 1. Functional traits provide insight into a variety of ecological questions, yet the optimal sampling method to estimate the community-level distribution of plant functional trait values remains a subject of debate, especially in species-rich forests. 2. We present a simulation analysis of the trait distribution of a set of nine completely sampled permanent plots in the lowland rain forests of French Guiana. 3. Increased sampling intensity consistently improved accuracy in estimating community-weighted means and variances of functional trait values, whereas there was substantial variation among functional traits and minor differences among sampling strategies. 4. Thus, investment in intensified sampling yields a greater improvement in the accuracy of estimation than does an equivalent investment in sampling design complication. 5. Notably, 'taxon-free' strategies frequently had greater accuracy than did abundance-based strategies, which had the additional cost of requiring botanical surveys. 6. We conclude that there is no substitute for extensive field sampling to accurately characterize the distribution of functional trait values in species-rich forests.
ISSN: 0269-8463

hal-01536541
https://hal.archives-ouvertes.fr/hal-01536541
PRODINRA : 343202
DOI : 10.1111/1365-2435.12433

Éditeur(s) : HAL CCSD American Geophysical Union
Résumé : International audience
The anisotropic single crystal seismic properties are reviewed in the light of recent experimental and theoretical determinations. Although considerable progress has been made on the determination of single crystal properties, data are still lacking, particularly for the temperature derivatives of transition zone and lower mantle phases. The common types of LPO of olivine, opx and cpx are presented together with their associated seismic properties. It is emphasized that simple seismic symmetry pattern of upper mantle rocks are a direct result of the interaction of olivine, opx and cpx. Using the standard structural frame (X lineation, Z pole to foliation) for typical LPOs of olivine, opx and cpx have the maximum Vp parallel to X, in the XZ plane and parallel to Y respectively. Destructive interference occurs between these minerals and hence P-wave anisotropy should be sensitive to the aggregate composition. For shear wave splitting (dVs) typical olivine and opx LPOs result in similar patterns with the maximum dVs in the YZ plane and the fast split shear wave (Vs1) polarized parallel to the foliation. A typical cpx LPO on the other hand produces destructive interference as the max dVs is close to X. By comparison with experiments and numerical simulations, it is estimated that upper mantle samples have an olivine LPO strength which recorded shear strain gamma of between 0.25 and 2.0. Pyrolite and piclogite models are compared with global transverse isotropic models. The slowly reducing P-wave anisotropy in the first 200 km can be explained by a model with constant composition and LPO strength. The sharp decrease in the observed anisotropy in the global models cannot be explained by the transformation of opx to cpx at 300 km, it is proposed that this decrease is due to a reduction in LPO strength from 200 to 350 km at the base of the lithosphere.
Earth's Deep Interior: Mineral Physics and Tomography From the Atomic Scale to the Global Scale

ISBN : 9780875909752
hal-01391524
http://hal.univ-reunion.fr/hal-01391524
http://hal.univ-reunion.fr/hal-01391524/document
http://hal.univ-reunion.fr/hal-01391524/file/Mainprice_AGUmonog_DM_GB_WBI_2000.pdf

Éditeur(s) : HAL CCSD Wiley
Résumé : International audience
We describe the simulation of an exterior problem using a magnetic field deriving from magnetostatics, with a numerical method mixing the approaches of C. I. Goldstein and L.-A. Ying. This method is based on the use of a graded mesh obtained by gluing homothetic layers in the exterior domain. On this mesh, we use an edge elements discretization and a recently proposed mixed formulation. In this paper, we provide both a theoretical and numerical study of the method. We establish an error estimate, describe the implementation, propose some preconditioning techniques and show the numerical results. We also compare these results with those obtained from an equivalent boundary elements approach. In this way, we retain that our method leads to a practical numerical implementation, a saving of storage, and turns out to be an alternative to the classical boundary elements method.
ISSN: 0749-159X

hal-00730196
https://hal.archives-ouvertes.fr/hal-00730196
DOI : 10.1002/num.10064

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
The western margin of the Gulf of Mexico (Veracruz State, Mexico) displays an extensive Neogene gravitational system, whereby the Neogene siliciclastic sediments are detached from underlying Mesozoic carbonates along decollement surface in Oligocene underpressured clays. Rapidly subsiding half-grabens develop above the footwall associated with major listric faults, whereas mini-basins develop in the deepest parts of the slope, in conjunction with the growth of west-verging compressional features. Between the high-angle normal faults in the west and the thrust anticlines in the east, a wide roll-over structure has grown progressively, resulting in a major topographic break in the morphology of the slope profile. Coupled forward kinematic modelling (Thrustpack) and stratigraphic modelling (Dionisos), calibrated against seismic profiles and two key nearshore wells, have been applied to representative regional transects across the margin, in order to quantify the vertical (subsidence) and horizontal (gravitational) deformations, and to discuss the impact of various key parameters on the distribution of debris flow depositions. These simulations are compared with data from distant ODP wells and discussed in the scope of current exploration strategies in both the US and Mexican portions of the Gulf of Mexico basin.
ISSN: 0040-1951

hal-00424595
https://hal.archives-ouvertes.fr/hal-00424595
DOI : 10.1016/j.tecto.2008.06.017

Éditeur(s) : HAL CCSD Springer
Résumé : International audience
Traditional network generation models aim to replicate global structural properties observed on various real-world networks through synthetic link formation mechanisms such as triadic closure or preferential attachment. Nowadays, the large amount of data available allow to study more precisely the link formation processes and to compare models with real situations. In this work, we analyze the network formed by the communication activities of the users of a telephony operator for studying the link formation process. Our goal is to identify the underlying formation mechanisms and checking if they match those proposed in traditional models. Indeed, the communications emitted and received by users are strong indicators for understanding the underlying patterns of the link formation process and highlighting how some individual properties induce the formation process in such a network. In a first study conducted at a global level, we show that the traditional mechanisms commonly used in network generation models cannot reproduce alone the link formation in this network. In a second study, we adopt a new point of view and analyze locally the link formation process by searching for correlations between some user attributes and the formation of the new links. The results obtained show that a very strong percentage of new links are formed between individuals with strong similarity.
ISSN: 1869-5450

hal-01312310
https://hal.univ-antilles.fr/hal-01312310
DOI : 10.1007/s13278-015-0304-y

Éditeur(s) : HAL CCSD
Résumé : International audience
Proceedings of the Fourth IEEE International Conference on Research Challenges in Information Science (RCIS 2010)

hal-00601642
https://hal.archives-ouvertes.fr/hal-00601642

Éditeur(s) : HAL CCSD Hermann-Éditions
Résumé : International audience
Cet article présente FLMin, une nouvelle méthode d'extraction de motifs fréquents dans les réseaux sociaux. Contrairement aux méthodes traditionnelles qui s'intéressent uniquement aux régularités structurelles, l'originalité de notre approche réside dans sa capacité à exploiter la structure et les attributs des noeuds pour extraire des régularités, que nous appelons "liens fréquents", dans les liens entre des noeuds partageant des caractéristiques communes.
Extraction et gestion des connaissances
Bordeaux, France

hal-00768408
https://hal.archives-ouvertes.fr/hal-00768408

Éditeur(s) : HAL CCSD INRIA
Résumé : International audience
Markov chain Monte Carlo (MCMC) methods together with hidden Markov models are extensively used in the Bayesian inference for many scientific fields like environment and ecology. Through simulated examples we show that the speed of convergence of these methods can be very low. In order to improve the convergence properties, we propose a method to make parallel chains interact. We apply this method to a biomass evolution model for fisheries.
Les méthodes de Monte Carlo par chaînes de Markov (MCMC) couplées à des modèles de Markov cachés sont utilisées dans de nombreux domaines, notamment en environnement et en écologie. Sur des exemples simples, nous montrons que la vitesse de convergence de ces méthodes peut être très faible. Nous proposons de mettre en interaction plusieurs algorithmes MCMC pour accélérer cette convergence. Nous appliquons ces méthodes à un modèle d'évolution de la biomasse d'une pêcherie.
ISSN: 1638-5713

inria-00506386
https://hal.inria.fr/inria-00506386
https://hal.inria.fr/inria-00506386/document
https://hal.inria.fr/inria-00506386/file/campillo2008d.pdf