Publications

2016

• A new comment on the computation of non conservative products using Roe-type path conservative schemes
  
  • Chalons Christophe
  • Coquel Frédéric
  , 2016. We are interested in the numerical approximation of the discontinuous solutions of non conservative hyperbolic systems. We more precisely consider a non conservative formulation of the usual gas dynamics equations and show how to slightly modify the so-called Roe-type path-conservative schemes to properly capture the underlying shock discontinuities. Numerical evidences are proposed. The present note follows a first comment on the computation of non conservative products in [1].
• The effect of competition and horizontal trait inheritance on invasion, fixation and polymorphism
  
  • Billiard Sylvain
  • Collet Pierre
  • Ferrière Régis
  • Méléard Sylvie
  • Tran Viet Chi
  Journal of Theoretical Biology, Elsevier, 2016, 411, pp.48-58. Horizontal transfer (HT) of heritable information or 'traits' (carried by genetic elements, plasmids, endosymbionts, or culture) is widespread among living organisms. Yet current ecological and evolutionary theory addressing HT is scant. We present a modeling framework for the dynamics of two populations that compete for resources and horizontally exchange (transfer) an otherwise vertically inherited trait. Competition infuences individual demographics, thereby affecting population size, which feeds back on the dynamics of transfer. This feedback is captured in a stochastic individual-based model, from which we derive a general model for the contact rate, with frequency-dependent (FD) and density-dependent (DD) rates as special cases. Taking a large-population limit on the stochastic individual-level model yields a deterministic Lotka-Volterra competition system with additional terms accounting for HT. The stability analysis of this system shows that HT can revert the direction of selection: HT can drive invasion of a deleterious trait, or prevent invasion of an advantageous trait. Due to HT, invasion does not necessarily imply fixation. Two trait values may coexist in a stable polymorphism even if their invasion fitnesses have opposite signs, or both are negative. Adressing the question of how the stochasticity of individual processes influences population fluctuations, we identify conditions on competition and mode of transfer (FD versus DD) under which the stochasticity of transfer events overwhelms demographic stochasticity. Assuming that one trait is initially rare, we derive invasion and fixation probabilities and time. In the case of costly plasmids, which are transfered unilaterally, invasion is always possible if the transfer rate is large enough; under DD and for intermediate values of the transfer rate, maintenance of the plasmid in a polymorphic population is possible. In conclusion, HT interacts with ecology (competition) in non-trivial ways. Our model provides a basis to study the influence of HT on evolutionary adaptation. (10.1016/j.jtbi.2016.10.003)
  DOI : 10.1016/j.jtbi.2016.10.003
• Lower large deviations for supercritical branching processes in random environment
  
  • Bansaye Vincent
  • Boeinghoff Christian
  , 2016. Branching Processes in Random Environment (BPREs) $(Z_n:n\geq0)$ are the generalization of Galton-Watson processes where in each generation the reproduction law is picked randomly in an i.i.d. manner. In the supercritical regime, the process survives with a positive probability and grows exponentially on the non-extinction event. We focus on rare events when the process takes positive values but lower than expected. More precisely, we are interested in the lower large deviations of $Z$, which means the asymptotic behavior of the probability $\{1 \leq Z_n \leq \exp(n\theta)\}$ as $n\rightarrow \infty$. We provide an expression of the rate of decrease of this probability, under some moment assumptions, which yields the rate function. This result generalizes the lower large deviation theorem of Bansaye and Berestycki (2009) by considering processes where $\P(Z_1=0 \vert Z_0=1)>0$ and also much weaker moment assumptions.
• Non-asymptotic convergence analysis for the Unadjusted Langevin Algorithm
  
  • Durmus Alain
  • Moulines Éric
  , 2016. In this paper, we study a method to sample from a target distribution $\pi$ over $\mathbb{R}^d$ having a positive density with respect to the Lebesgue measure, known up to a normalisation factor. This method is based on the Euler discretization of the overdamped Langevin stochastic differential equation associated with $\pi$. For both constant and decreasing step sizes in the Euler discretization, we obtain non-asymptotic bounds for the convergence to the target distribution $\pi$ in total variation distance. A particular attention is paid to the dependency on the dimension $d$, to demonstrate the applicability of this method in the high dimensional setting. These bounds improve and extend the results of (Dalalyan 2014).
• Mechanism design and allocation algorithms for network markets with piece-wise linear costs and externalities
  
  • Heymann Benjamin
  • Jofré Alejandro
  , 2016. Motivated by market power in electricity market we introduce a mechanism design in [1] for simplified markets of two agents with linear production cost functions. In standard procurement auctions, the market power resulting from the quadratic transmission losses allow the producers to bid above their true value (i.e. production cost). The mechanism proposed in the previous paper reduces the producers margin to the society benefit. We extend those results to a more general market made of a finite number of agents with piecewise linear cost functions, which make the problem more difficult, but at the same time more realistic. We show that the methodology works for a large class of externalities. We also provide two algorithms to solve the principal allocation problem.
• Dimension reduction for the micromagnetic energy functional on curved thin films
  
  • Di Fratta Giovanni
  , 2016. Micromagnetic con gurations of the vortex and onion type have beenwidely studied in the context of planar structures. Recently a signi cant interest to micromagnetic curved thin lms has appeared. In particular, thin spherical shells are currently of great interest due to their capability to support skyrmion solutions which can be stabilized by curvature e ects only, in contrast to the planar case where the intrinsic Dzyaloshinsky-Moriya interaction is required.The aimof the paper is to performa $Γ$-development analysis of the micromagnetic energy functional, when the shell is generated, like in the case of a sphere, by a bounded and convex smooth surface.
• Nonmixing layers
  
  • Gaillard Pierre
  • Giovangigli Vincent
  • Matuszewski Lionel
  Physical Review Fluids, American Physical Society, 2016, 1 (8), pp.page 084001-1 - 084001-12. We investigate the impact of nonideal diffusion on the structure of supercritical cryogenic binary mixing layers. This situation is typical of liquid fuel injection in high-pressure rocket engines. Nonideal diffusion has a dramatic impact in the neighborhood of chemical thermodynamic stability limits where the components become quasi-immiscible and ultimately form a nonmixing layer. Numerical simulations are performed for mixing layers of H2 and N2 at a pressure of 100 atm and temperature around 120–150 K near chemical thermodynamic stability limits. (10.1103/PhysRevFluids.1.084001)
  DOI : 10.1103/PhysRevFluids.1.084001
• A probabilistic max-plus numerical method for solving stochastic control problems
  
  • Akian Marianne
  • Fodjo Eric
  , 2016. We consider fully nonlinear Hamilton-Jacobi-Bellman equations associated to diffusion control problems involving a finite set-valued (or switching) control and possibly a continuum-valued control. We construct a lower complexity probabilistic numerical algorithm by combining the idempotent expansion properties obtained by McEneaney, Kaise and Han (2011) for solving such problems with a numerical probabilistic method such as the one proposed by Fahim, Touzi and Warin (2011) for solving some fully nonlinear parabolic partial differential equations. Numerical tests on a small example of pricing and hedging an option are presented.
• Quantitative a posteriori error estimators in Finite Element-based shape optimization
  
  • Giacomini Matteo
  , 2016. Gradient-based shape optimization strategies rely on the computation of the so-called shape gradient. In many applications, the objective functional depends both on the shape of the domain and on the solution of a PDE which can only be solved approximately (e.g. via the Finite Element Method). Hence, the direction computed using the discretized shape gradient may not be a genuine descent direction for the objective functional. This Ph.D. thesis is devoted to the construction of a certification procedure to validate the descent direction in gradient-based shape optimization methods using a posteriori estimators of the error due to the Finite Element approximation of the shape gradient.By means of a goal-oriented procedure, we derive a fully computable certified upper bound of the aforementioned error. The resulting Certified Descent Algorithm (CDA) for shape optimization is able to identify a genuine descent direction at each iteration and features a reliable stopping criterion basedon the norm of the shape gradient.Two main applications are tackled in the thesis. First, we consider the scalar inverse identification problem of Electrical Impedance Tomography and we investigate several a posteriori estimators. A first procedure is inspired by the complementary energy principle and involves the solution of additionalglobal problems. In order to reduce the computational cost of the certification step, an estimator which depends solely on local quantities is derived via an equilibrated fluxes approach. The estimators are validated for a two-dimensional case and some numerical simulations are presented to test the discussed methods. A second application focuses on the vectorial problem of optimal design of elastic structures. Within this framework, we derive the volumetric expression of the shape gradient of the compliance using both H 1 -based and dual mixed variational formulations of the linear elasticity equation. Some preliminary numerical tests are performed to minimize the compliance under a volume constraint in 2D using the Boundary Variation Algorithm and an a posteriori estimator of the error in the shape gradient is obtained via the complementary energy principle.
• High-dimensional Bayesian inference via the Unadjusted Langevin Algorithm
  
  • Durmus Alain
  • Moulines Éric
  , 2016. We consider in this paper the problem of sampling a high-dimensional probability distribution $\pi$ having a density wrt the Lebesgue measure on $\mathbb{R}^d$, known up to a normalisation factor $x \mapsto \mathrm{e}^{−U (x)} / \int_{\mathbb{R}^d} \mathrm{e}^{−U (y)}\mathrm{d}y$. Such problem naturally occurs for example in Bayesian inference and machine learning. Under the assumption that $U$ is continuously differentiable, $\nabla U$ is globally Lipschitz and $U$ is strongly convex, we obtain non-asymptotic bounds for the convergence to stationarity in Wasserstein distance of order $2$ and total variation distance of the sampling method based on the Euler discretization of the Langevin stochastic differential equation, for both constant and decreasing step sizes. The dependence on the dimension of the state space of the obtained bounds is studied to demonstrate the applicability of this method. The convergence of an appropriately weighted empirical measure is also investigated and bounds for the mean square error and exponential deviation inequality are reported for functions which are either Lipchitz continuous or measurable and bounded. An illustration to a Bayesian inference for binary regression is presented.
• Reconstruction methods for inverse problems for Helmholtz-type equations
  
  • Agaltsov Alexey
  , 2016. This work is devoted to study of some inverse problems for the gauge-covariant Helmholtz equation, whose particular cases include the Schrödinger equation for a charged elementary particle in a magnetic field and the time-harmonic wave equation describing sound waves in a moving fluid. These problems are mainly motivated by applications in different tomographies, including acoustic tomography, tomography using elementary particles and electrical impedance tomography. In particular, we study inverse problems motivated by applications in acoustic tomography of moving fluid. We present formulas and equations which allow to reduce the acoustic tomography problem to an appropriate inverse scattering problem. Next, we develop a functional-analytic algorithm for solving this inverse scattering problem. However, in general, the solution to the latter problem is unique only up to an appropriate gauge transformation. In this connection, we give formulas and equations which allow to get rid of this gauge non-uniqueness and recover the fluid parameters, by measuring acoustic fields at several frequencies. We also present examples of fluids which are not distinguishable in this acoustic tomography setting. Next, we consider the inverse scattering problem without phase information. This problem is motivated by applications in tomography using elementary particles, where only the absolute value of the scattering amplitude can be measured relatively easily. We give estimates in the configuration space for the phaseless Born-type reconstructions, which are needed for the further development of precise inverse scattering algorithms. Finally, we consider the problem of determination of a Riemann surface in the complex projective plane from its boundary. This problem arises as a part of the inverse Dirichlet-to-Neumann problem for the Laplace equation on an unknown 2-dimensional surface, and is motivated by applications in electrical impedance tomography.
• Stochastic Gradient Richardson-Romberg Markov Chain Monte Carlo
  
  • Durmus Alain
  • Şimşekli Umut
  • Moulines Éric
  • Badeau Roland
  • Richard Gael
  , 2016. Stochastic Gradient Markov Chain Monte Carlo (SG-MCMC) algorithms have become increasingly popular for Bayesian inference in large-scale applications. Even though these methods have proved useful in several scenarios, their performance is often limited by their bias. In this study, we propose a novel sampling algorithm that aims to reduce the bias of SG-MCMC while keeping the variance at a reasonable level. Our approach is based on a numerical sequence acceleration method, namely the Richardson-Romberg extrapolation, which simply boils down to running almost the same SG-MCMC algorithm twice in parallel with different step sizes. We illustrate our framework on the popular Stochastic Gradient Langevin Dynamics (SGLD) algorithm and propose a novel SG-MCMC algorithm referred to as Stochastic Gradient Richardson-Romberg Langevin Dynamics (SGRRLD). We provide formal theoretical analysis and show that SGRRLD is asymptotically consistent, satisfies a central limit theorem, and its non-asymptotic bias and the mean squared-error can be bounded. Our results show that SGRRLD attains higher rates of convergence than SGLD in both finite-time and asymptotically, and it achieves the theoretical accuracy of the methods that are based on higher-order integrators. We support our findings using both synthetic and real data experiments.
• On the convergence of monotone schemes for path-dependent PDE *
  
  • Ren Zhenjie
  • Tan Xiaolu
  Stochastic Processes and their Applications, Elsevier, 2016. We propose a reformulation of the convergence theorem of monotone numerical schemes introduced by Zhang and Zhuo [32] for viscosity solutions of path-dependent PDEs (PPDE), which extends the seminal work of Barles and Souganidis [1] on the viscosity solution of PDE. We prove the convergence theorem under conditions similar to those of the classical theorem in [1]. These conditions are satisfied, to the best of our knowledge, by all classical monotone numerical schemes in the context of stochastic control theory. In particular, the paper provides a unified approach to prove the convergence of numerical schemes for non-Markovian stochastic control problems, second order BSDEs, stochastic differential games etc.
• Homogenized and analytical models for the diffusion MRI signal
  
  • Schiavi Simona
  , 2016. Diffusion magnetic resonance imaging (dMRI) is an imaging modality that probes the diffusion characteristics of a sample via the application of magnetic field gradient pulses. More specifically, it encodes water displacement due to diffusion and is then a powerful tool to obtain information on the tissue microstructure. The signal measured by the MRI scanner is a mean-value measurement in a physical volume, called a voxel, whose size, due to technical reasons, is much larger than the scale of the microscopic variations of the cellular structure. It follows that the microscopic components of the tissues are not visible at the spatial resolution of dMRI. Rather, their geometric features are aggregated into the macroscopic signal coming from the voxels. An important quantity measured in dMRI in each voxel is the Apparent Diffusion Coefficient (ADC) and it is well-established from imaging experiments that, in the brain, in-vivo, the ADC is dependent on the diffusion time. There is a large variety (phenomenological, probabilistic, geometrical, PDE based model, etc.) of macroscopic models for ADC in the literature, ranging from simple to complicated. Indeed, each of these models is valid under a certain set of assumptions. The goal of this thesis is to derive simple (but sufficiently sound for applications) models starting from fine PDE modelling of diffusion at microscopic scale using homogenization techniques.In a previous work, the homogenized FPK model was derived starting from the Bloch-Torrey PDE equation under the assumption that membrane's permeability is small and diffusion time is large. We first analyse this model and establish a convergence result to the well known K{"a}rger model as the magnetic pulse duration goes to 0. In that sense, our analysis shows that the FPK model is a generalisation of the K{"a}rger one for the case of arbitrary duration of the magnetic pulses. We also give a mathematically justified new definition of the diffusion time for the K{"a}rger model (the one that provides the highest rate of convergence).The ADC for the FPK model is time-independent which is not compatible with some experimental observations. Our goal next is to correct this model for small so called b-values so that the resulting homogenised ADC is sensitive to both the pulses duration and the diffusion time. To achieve this goal, we employed a similar homogenization technique as for FPK, but we include a suitable time and gradient intensity scalings for the range of considered b-values. Numerical simulations show that the derived asymptotic new model provides a very accurate approximation of the dMRI signal at low b-values. We also obtain some analytical approximations (using short time expansion of surface potentials for the heat equation and eigenvalue decompositions) of the asymptotic model that yield explicit formulas of the time dependency of ADC. Our results are in concordance with classical ones in the literature and we improved some of them by accounting for the pulses duration.Finally we explored the inverse problem of determining qualitative information on the cells volume fractions from measured dMRI signals. While finding sphere distributions seems feasible from measurement of the whole dMRI signal, we show that ADC alone would not be sufficient to obtain this information.
• An Explicit Martingale Version of the One-dimensional Brenier's Theorem with Full Marginals Constraint
  
  • Henry-Labordère Pierre
  • Tan Xiaolu
  • Touzi Nizar
  Stochastic Processes and their Applications, Elsevier, 2016. We provide an extension of the martingale version of the Fréchet-Hoeffding coupling to the infinitely-many marginals constraints setting. In the two-marginal context, this extension was obtained by Beiglböck & Juillet [7], and further developed by Henry-Labordère & Touzi [40], see also [6]. Our main result applies to a special class of reward functions and requires some restrictions on the marginal distributions. We show that the optimal martingale transference plan is induced by a pure downward jump local Lévy model. In particular, this provides a new martingale peacock process (PCOC " Processus Croissant pour l'Ordre Convexe, " see Hirsch, Profeta, Roynette & Yor [43]), and a new remarkable example of discontinuous fake Brownian motions. Further, as in [40], we also provide a duality result together with the corresponding dual optimizer in explicit form. As an application to financial mathematics, our results give the model-independent optimal lower and upper bounds for variance swaps.
• Spectral theory near thresholds for weak interactions with massive particles
  
  • Barbaroux Jean-Marie
  • Faupin Jérémy
  • Guillot Jean-Claude
  Journal of Spectral Theory, European Mathematical Society, 2016, 6 (3), pp.505–555. We consider a Hamiltonian describing the weak decay of the massive vector boson Z0 into electrons and positrons. We show that the spectrum of the Hamiltonian is composed of a unique isolated ground state and a semi-axis of essential spectrum. Using a suitable extension of Mourre's theory, we prove that the essential spectrum below the boson mass is purely absolutely continuous. (10.4171/JST/131)
  DOI : 10.4171/JST/131
• Optimal Skorokhod embedding under finitely-many marginal constraints *
  
  • Guo Gaoyue
  • Tan Xiaolu
  • Touzi Nizar
  SIAM Journal on Control and Optimization, Society for Industrial and Applied Mathematics, 2016. The Skorokhod embedding problem aims to represent a given probability measure on the real line as the distribution of Brownian motion stopped at a chosen stopping time. In this paper, we consider an extension of the optimal Skorokhod embedding problem in Beiglböck , Cox & Huesmann [1] to the case of finitely-many marginal constraints 1. Using the classical convex duality approach together with the optimal stopping theory, we obtain the duality results which are formulated by means of probability measures on an enlarged space. We also relate these results to the problem of martingale optimal transport under multiple marginal constraints.
• Mathematical modeling and numerical simulation of a bioreactor landfill using Feel++
  
  • Dollé Guillaume
  • Duran O
  • Feyeux Nelson
  • Frénod Emmanuel
  • Giacomini Matteo
  • Prud'Homme Christophe
  ESAIM: Proceedings and Surveys, EDP Sciences, 2016, 55, pp.83-110. In this paper, we propose a mathematical model to describe the functioning of a bioreactor landfill, that is a waste management facility in which biodegradable waste is used to generate methane. The simulation of a bioreactor landfill is a very complex multiphysics problem in which bacteria catalyze a chemical reaction that starting from organic carbon leads to the production of methane, carbon dioxide and water. The resulting model features a heat equation coupled with a non-linear reaction equation describing the chemical phenomena under analysis and several advection and advection-diffusion equations modeling multiphase flows inside a porous environment representing the biodegradable waste. A framework for the approximation of the model is implemented using Feel++, a C++ open-source library to solve Partial Differential Equations. Some heuristic considerations on the quantitative values of the parameters in the model are discussed and preliminary numerical simulations are presented. (10.1051/proc/201655083)
  DOI : 10.1051/proc/201655083
• Stability of non-autonomous difference equations with applications to transport and wave propagation on networks
  
  • Chitour Yacine
  • Mazanti Guilherme
  • Sigalotti Mario
  Networks and Heterogeneous Media, American Institute of Mathematical Sciences, 2016, 11, pp.563-601. In this paper, we address the stability of transport systems and wave propagation on networks with time-varying parameters. We do so by reformulating these systems as non-autonomous difference equations and by providing a suitable representation of their solutions in terms of their initial conditions and some time-dependent matrix coefficients. This enables us to characterize the asymptotic behavior of solutions in terms of such coefficients. In the case of difference equations with arbitrary switching, we obtain a delay-independent generalization of the well-known criterion for autonomous systems due to Hale and Silkowski. As a consequence, we show that exponential stability of transport systems and wave propagation on networks is robust with respect to variations of the lengths of the edges of the network preserving their rational dependence structure. This leads to our main result: the wave equation on a network with arbitrarily switching damping at external vertices is exponentially stable if and only if the network is a tree and the damping is bounded away from zero at all external vertices but at most one. (10.3934/nhm.2016010)
  DOI : 10.3934/nhm.2016010
• On the monotonicity principle of optimal Skorokhod embedding problem *
  
  • Guo Gaoyue
  • Tan Xiaolu
  • Touzi Nizar
  SIAM Journal on Control and Optimization, Society for Industrial and Applied Mathematics, 2016. This is a continuation of our accompanying paper [18]. We provide an alternative proof of the monotonicity principle for the optimal Skorokhod embedding problem established in Beiglböck , Cox and Huesmann [2]. Our proof is based on the adaptation of the Monge-Kantorovich duality in our context, a delicate application of the optional cross-section theorem, and a clever conditioning argument introduced in [2].
• Convergence of an infinite dimensional stochastic process to a spatially structured trait substitution sequence
  
  • Leman Hélène
  Stochastics and Partial Differential Equations: Analysis and Computations, Springer US, 2016, 4 (4), pp.791 - 826. We consider an individual-based spatially structured population for Darwinian evolution in an asexual population. The individuals move randomly on a bounded continuous space according to a reflected brownian motion. The dynamics involves also a birth rate, a density-dependent logistic death rate and a probability of mutation at each birth event. We study the convergence of the microscopic process when the population size grows to $+\infty$ and the mutation probability decreases to $0$. We prove a convergence towards a jump process that jumps in the infinite dimensional space of the stable spatial distributions. The proof requires specific studies of the microscopic model. First, we examine the large deviation principle around the deterministic large population limit of the microscopic process. Then, we find a lower bound on the exit time of a neighborhood of a stationary spatial distribution. Finally, we study the extinction time of the branching diffusion processes that approximate small size populations. (10.1007/s40072-016-0077-y)
  DOI : 10.1007/s40072-016-0077-y
• An iterative inversion of weighted Radon transforms along hyperplanes
  
  • Goncharov F O
  , 2016. We propose iterative inversion algorithms for weighted Radon transforms $R_W$ along hyperplanes in $R^3$. More precisely, expanding the weight $W = W (x, \theta), x \in R^3 , \theta \in S^2$ , into the series of spherical harmonics in $\theta$ and assuming that the zero order term $w_{0,0}(x)$ is not zero at any $x \in R^3$ , we reduce the inversion of $R_W$ to solving a linear integral equation. In addition, under the assumption that the even part of $W$ in $\theta$ (i.e., $1/2(W (x, \theta) + W (x, −\theta))$) is close to $w_{0,0}$, the aforementioned linear integral equation can be solved by the method of successive approximations. Approximate inversions of $R_W$ are also given. Our results can be considered as an extension to 3D of two-dimensional results of Kunyansky (1992), Novikov (2014), Guillement, Novikov (2014). In our studies we are motivated, in particular, by problems of emission tomographies in 3D. In addition, we generalize our results to the case of dimension $n > 3$.
• Précision de modèle et efficacité algorithmique : exemples du traitement de l'occultation en stéréovision binoculaire et de l'accélération de deux algorithmes en optimisation convexe
  
  • Tan Pauline
  , 2016. Le présent manuscrit est composé de deux parties relativement indépendantes.La première partie est consacrée au problème de la stéréovision binoculaire, et plus particulièrement au traitement de l'occultation. En partant d'une analyse de ce phénomène, nous en déduisons un modèle de régularité qui inclut une contrainte convexe de visibilité. La fonctionnelle d'énergie qui en résulte est minimisée par relaxation convexe. Les zones occultées sont alors détectées grâce à la pente horizontale de la carte de disparité avant d'être densifiées.Une autre méthode gérant l'occultation est la méthode des graph cuts proposée par Kolmogorov et Zabih. L'efficacité de cette méthode justifie son adaptation à deux problèmes auxiliaires rencontrés en stéréovision, qui sont la densification de cartes éparses et le raffinement subpixellique de cartes pixelliques.La seconde partie de ce manuscrit traite de manière plus générale de deux algorithmes d'optimisation convexe, pour lequels deux variantes accélérées sont proposées. Le premier est la méthode des directions alternées (ADMM). On montre qu'un léger relâchement de contraintes dans les paramètres de cette méthode permet d'obtenir un taux de convergence théorique plus intéressant.Le second est un algorithme de descentes proximales alternées, qui permet de paralléliser la résolution approchée du problème Rudin-Osher-Fatemi (ROF) de débruitage pur dans le cas des images couleurs. Une accélération de type FISTA est également proposée.
• Rare events simulation by shaking transformations : Non-intrusive resampler for dynamic programming
  
  • Liu Gang
  , 2016. This thesis contains two parts: rare events simulation and non-intrusive stratified resampler for dynamic programming. The first part consists of quantifying statistics related to events which are unlikely to happen but which have serious consequences. We propose Markovian transformation on path spaces and combine them with the theories of interacting particle system and of Markov chain ergodicity to propose methods which apply very generally and have good performance. The second part consists of resolving dynamic programming problem numerically in a context where we only have historical observations of small size and we do not know the values of model parameters. We propose and analyze a new scheme with stratification and resampling techniques.
• Solving Hamilton-Jacobi-Bellman equations by combining a max-plus linear approximation and a probabilistic numerical method
  
  • Akian Marianne
  , 2016.