Éditeur(s) : HAL CCSD
Résumé : International audience
We consider a second-order differential system with Hessian-driven damping . An interesting property of this system is that, after introduction of an auxiliary variable y , it can be equivalently written as a first-order system in time and space. This allows us to extend the analysis to the case of a convex lower semicontinuous potential and so to introduce constraints in the model. When considering the indicator function of a closed convex set, the subdifferential operator takes account of the contact forces. In this setting, by playing with the geometrical damping parameter, we can describe nonelastic shock laws with restitution coefficient. Taking advantage of the infinite dimensional framework, we introduce a nonlinear hyperbolic PDE describing a damped oscillating system with obstacle. The system is dissipative; in the convex case each trajectory weakly converges to a minimizer of the global potential energy function. Exponential stabilization is obtained under strong convexity assumptions.
Differential Equations & Applications

hal-00803214
https://hal.archives-ouvertes.fr/hal-00803214
https://hal.archives-ouvertes.fr/hal-00803214/document
https://hal.archives-ouvertes.fr/hal-00803214/file/Attouch-Mainge-Redont-30August2011.pdf

Résumé : Cette thèse est consacrée à la résolution d'un problème fondamental de l'analyse variationnelle qu'est la recherchede zéros d'opérateurs maximaux monotones dans un espace de Hilbert. Nous nous sommes tout d'abord intéressés au cas de l'opérateur somme étendue de deux opérateurs maximaux monotones; la recherche d'un zéro de cet opérateur est un problème dont la bibliographie est peu fournie: nous proposons une version modifiée de l'algorithme d'éclatement forward-backward utilisant à chaque itération, l'epsilon-élargissement d'un opérateur maximal monotone,afin de construire une solution. Nous avons ensuite étudié la convergence d'un nouvel algorithme de faisceaux pour construire ID zéro d'un opérateur maximal monotone quelconque en dimension finie. Cet algorithme fait intervenir une double approximation polyédrale de l'epsilon-élargissement de l'opérateur considéré
This thesis is devoted to solving a basic problem of variational analysis which is the search of zeros of maximal monotone operators in a Hilbert space. First of aIl, we concentrate on the case of the extended som of two maximal monotone operators; the search of a zero of this operator is a problem for which the bibliography is not abondant: we purpose a modified version of the forward-backward splitting algorithm using at each iteration, the epsilon-enlargement of a maximal monotone operator, in order to construet a solution. Secondly, we study the convergence of a new bondie algorithm to construet a zero of an arbitrary maximal monotone operator in a finite dimensional space. In this algorithm, intervenes a double polyhedral approximation of the epsilon-enlargement of the considered operator

http://www.theses.fr/2012AGUY0548/document

Éditeur(s) : HAL CCSD Springer Verlag
Résumé : International audience
We provide a first-order necessary and sufficient condition for optimality of lower semicontinuous functions on Banach spaces using the concept of subdifferential. From the sufficient condition we derive that any subdifferential operator is monotone absorbing, hence maximal monotone when the function is convex.
ISSN: 0925-5001

hal-00761658
https://hal.archives-ouvertes.fr/hal-00761658
ARXIV : 1212.0532
DOI : 10.1007/s10898-013-0078-6

Éditeur(s) : HAL CCSD Taylor & Francis
Résumé : International audience
ISSN: 0233-1934

hal-00816722
https://hal.univ-antilles.fr/hal-00816722
DOI : 10.1080/02331939408843968

Éditeur(s) : HAL CCSD
Résumé : International audience
We show that the properties of dense subdifferentiability and of trustworthiness are equivalent for any subdifferential satisfying a small set of natural axioms. The proof relies on a remarkable property of the subdifferential of the inf-convolution of two (non necessarily convex) functions. We also show the equivalence of the dense subdifferentiability property with other basic properties of subdifferentials such as a weak∗ Lipschitz Separation property, a strong Compact Separation property and a Minimal property for the analytic approximate subdifferential of Ioffe.
Serdica. Mathematical Journal

hal-00694603
https://hal.archives-ouvertes.fr/hal-00694603

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
The growth of relief in active tectonic areas is mainly controlled by the interactions between tectonics and surface processes (erosion and sedimentation). The study of long-lived morphologic markers formed by these interactions can help in quantifying the competing effects of tectonics, erosion and sedimentation. In regions experiencing active extension, river-long profiles and faceted spurs (triangular facets) can help in un- derstanding the development of mountainous topography along normal fault scarps. In this study, we devel- oped analogue experiments that simulate the morphologic evolution of a mountain range bounded by a normal fault. This paper focuses on the effect of the fault slip rate on the morphologic evolution of the foot- wall by performing three analogue experiments with different fault slip rates under a constant rainfall rate. A morphometric analysis of the modelled catchments allows comparing with a natural case (Tunka half- graben, Siberia). After a certain amount of fault slip, the modelled footwall topographies of our models reaches a dynamic equilibrium (i.e., erosion balances tectonic uplift relative to the base level) close to the fault, whereas the topography farther from the fault is still being dissected due to regressive erosion. We show that the rates of vertical erosion in the area where dynamic equilibrium is reached and the rate of re- gressive erosion are linearly correlated with the fault throw rate. Facet morphology seems to depend on the fault slip rate except for the fastest experiment where faceted spurs are degraded due to mass wasting. A stream-power law is computed for the area wherein rivers reach a topographic equilibrium. We show that the erosional capacity of the system depends on the fault slip rate. Finally, our results demonstrate the pos- sibility of preserving convex river-long profiles on the long-term under steady external (tectonic uplift and rainfall) conditions.
ISSN: 0040-1951

hal-00646966
https://hal.archives-ouvertes.fr/hal-00646966
https://hal.archives-ouvertes.fr/hal-00646966/document
https://hal.archives-ouvertes.fr/hal-00646966/file/preprint_Strak_2011.pdf
DOI : 10.1016/j.tecto.2011.10.005

Éditeur(s) : HAL CCSD Springer Verlag
Résumé : International audience
Starting from a locally convex metrisable topological space and from any asymptotic scale, we construct a generalized extension of this space. To those extensions, we associate Hausdorff topologies. We introduce the notion of a temperate map, with respect to a given asymptotic scale, between two locally convex metrisable semi-normed spaces. We show that such mappings extend in a canonical way to mappings between the respective generalized extensions. We give an application to nonlinear Dirichlet boundary value problems with singular data in the framework of generalized extensions
ISSN: 0026-9255

hal-01529549
https://hal.univ-antilles.fr/hal-01529549

Éditeur(s) : HAL CCSD EDP Sciences
Résumé : International audience
In the setting of a real Hilbert space H, we investigate the asymptotic behavior, as time t goes to infinity, of trajectories of second-order damped evolution equations, which are governed by the sum of the gradient operator of a convex differentiable potential function , and a maximal monotone operator which is assumed to be cocoercive. Under a sharp condition involving the viscous and the cocoercive parameters, it is proved that each trajectory weakly converges to an equilibrium. Passing from weak to strong convergence is obtained by introducing an asymptotically vanishing Tikhonov-like regularizing term. As special cases, we recover the asymptotic analysis of the heavy ball with friction dynamic attached to a convex potential, the second-order gradient-projection dynamic, and the second-order dynamic governed by the Yosida approximation of a general maximal monotone operator. The breadth and flexibility of the proposed framework is illustrated through applications in the areas of constrained optimization, dynamical approach to Nash equilibria for noncooperative games, and asymptotic stabilization in the case of a continuum of equilibria.
ISSN: 1292-8119

hal-00803774
https://hal.archives-ouvertes.fr/hal-00803774
https://hal.archives-ouvertes.fr/hal-00803774/document
https://hal.archives-ouvertes.fr/hal-00803774/file/ATTOUCH-MAINGE-09-04-2010-COCV.pdf
DOI : 10.1051/cocv/2010024

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
A multifunction Γ is called a Kakutani multifunction if there exist two nonempty convex sets X and Y , each in a Hausdorff topological vector space, such that Γ : X → Y is upper semi-continuous with nonempty compact convex values. We prove the following extension of the Kakutani fixed point theorem : Let Γ : X → X be a multi-function from a simplex X into itself ; if Γ can be factorized by an arbitrary finite number of Kakutani multifunctions, then Γ has a fixed point. The proof relies on a simplicial approximation technique and the Brouwer fixed point theorem. Extensions to infinite-dimensional spaces and applications to game theory are given.
Journal of Mathematical Analysis and applications

hal-00699225
https://hal.archives-ouvertes.fr/hal-00699225

Éditeur(s) : HAL CCSD Springer Verlag
Résumé : International audience
Given a pair of Banach spaces X and Y such that one is the dual of the other, we study the relationships between generic Fréchet differentiability of convex continuous functions on Y (Asplund property), generic existence of linear perturbations for lower semicontinuous functions on X to have a strong minimum (Stegall variational principle), and dentability of bounded subsets of X (Radon-Nikodým Property).
ISSN: 0927-6947

hal-00694604
https://hal.archives-ouvertes.fr/hal-00694604

Éditeur(s) : HAL CCSD American Mathematical Society
Résumé : International audience
A topological description of various generalized function algebras over corresponding basic locally convex algebras is given. The framework consists of algebras of sequences with appropriate ultra( pseudo)metrics defined by sequences of exponential weights. Such an algebra with embedded Dirac's delta distribution induces discrete topology on the basic space. This result is in analogy to Schwartz' impossibility result concerning multiplication of distributions.
ISSN: 0002-9939

hal-00758571
https://hal.univ-antilles.fr/hal-00758571
https://hal.univ-antilles.fr/hal-00758571/document
https://hal.univ-antilles.fr/hal-00758571/file/0206039v1.pdf

Éditeur(s) : HAL CCSD International Publishing Service/IPS
Résumé : International audience
We give a description of various algebras of generalized functions based on the introduction of pseudo-ultranorms on spaces of sequences in given locally convex function algebras. We study sheaf properties of these algebras, needed for microlocal analysis, and also consider regularity theory, functoriality and different concepts of association and weak equality in a unified setting. Using this approach, we also give new results on embeddings of ultradistribution and hyperfunction spaces into corresponding algebras of generalized functions.
ISSN: 0012-3862

hal-00758044
https://hal.univ-antilles.fr/hal-00758044
https://hal.univ-antilles.fr/hal-00758044/document
https://hal.univ-antilles.fr/hal-00758044/file/DHPV-26oct-8nov06.pdf
DOI : 10.4064/dm447-0-1

Éditeur(s) : HAL CCSD
Résumé : International audience
We start from a basic version of the Hahn-Banach theorem, of which we provide a proof based on Tychonoff's theorem on the product of compact intervals. Then, in the first section, we establish conditions ensuring the existence of affine functions lying between a convex function and a concave one in the setting of vector spaces -- this directly leads to the theorems of Hahn-Banach, Mazur-Orlicz and Fenchel. In the second section, we caracterize those topological vector spaces for which certain convex functions are continuous -- this is connected to the uniform boundedness theorem of Banach-Steinhaus and to the closed graph and open mapping theorems of Banach. Combining both types of results readily yields topological versions of the theorems of the first section.
Nonconvex Optim. Appl.
Erice, Italy

Droits : info:eu-repo/semantics/OpenAccess
hal-00699215
https://hal.archives-ouvertes.fr/hal-00699215
https://hal.archives-ouvertes.fr/hal-00699215/document
https://hal.archives-ouvertes.fr/hal-00699215/file/M_NonconvexOptimAppl98.pdf

Éditeur(s) : HAL CCSD
Résumé : International audience
It is an admitted fact that mainstream boosting algorithms like AdaBoost do not perform well to estimate class conditional probabilities. In this paper, we analyze, in the light of this problem, a recent algorithm, unn, which leverages nearest neighbors while minimizing a convex loss. Our contribution is threefold. First, we show that there exists a subclass of surrogate losses, elsewhere called balanced, whose minimization brings simple and statistically efficient estimators for Bayes posteriors. Second, we show explicit convergence rates towards these estimators for unn, for any such surrogate loss, under a Weak Learning Assumption which parallels that of classical boosting results. Third and last, we provide experiments and comparisons on synthetic and real datasets, including the challenging SUN computer vision database. Results clearly display that boosting nearest neighbors may provide highly accurate estimators, sometimes more than a hundred times more accurate than those of other contenders like support vector machines.
European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD)
Bristol, United Kingdom

hal-00702771
https://hal.archives-ouvertes.fr/hal-00702771
https://hal.archives-ouvertes.fr/hal-00702771/document
https://hal.archives-ouvertes.fr/hal-00702771/file/ecml12-dnbnb-sub.pdf

Éditeur(s) : HAL CCSD Taylor & Francis
Résumé : International audience
We introduce the notions of conditioning and well-posedness for equilibrium problems. Using these concepts, we obtain finite and strong convergence results for the proximal method that improve, develop, and unify several theorems in optimization and nonlinear analysis.
ISSN: 0163-0563

hal-00779994
https://hal.univ-antilles.fr/hal-00779994
DOI : 10.1080/01630560701766684