Search

Heterogeneity of the tumor microenvironment (TME) is one of the major causes of treatment resistance in breast cancer. Among TME components, nervous system role in clinical outcome has been underestimated. Identifying neuronal signatures associated with treatment response will help to characterize neuronal influence on tumor progression and identify new treatment targets. The search for hormonotherapy-predictive biomarkers was implemented by supervised machine learning (ML) analysis on merged transcriptomics datasets from public databases. ML-derived genes were investigated by pathway enrichment analysis, and potential gene signatures were curated by removing the variables that were not strictly nervous system specific. The predictive and prognostic abilities of the generated signatures were examined by Cox models, in the initial cohort and seven external cohorts. Generated signature performances were compared with 14 other published signatures, in both the initial and external cohorts. Underlying biological mechanisms were explored using deconvolution tools (CIBERSORTx and xCell). Our pipeline generated two nervous system-related signatures of 24 genes and 97 genes (NervSign24 and NervSign97). These signatures were prognostic and hormonotherapy-predictive, but not chemotherapy-predictive. When comparing their predictive performance with 14 published risk signatures in six hormonotherapy-treated cohorts, NervSign97 and NervSign24 were the two best performers. Pathway enrichment score and deconvolution analysis identified brain neural progenitor presence and perineural invasion as nervous system-related mechanisms positively associated with NervSign97 and poor clinical prognosis in hormonotherapy-treated patients. Transcriptomic profiling has identified two nervous system–related signatures that were validated in clinical samples as hormonotherapy-predictive signatures, meriting further exploration of neuronal component involvement in tumor progression. Significance: The development of personalized and precision medicine is the future of cancer therapy. With only two gene expression signatures approved by FDA for breast cancer, we are in need of new ones that can reliably stratify patients for optimal treatment. This study provides two hormonotherapy-predictive and prognostic signatures that are related to nervous system in TME. It highlights tumor neuronal components as potential new targets for breast cancer therapy.

Heterogeneity of the tumor microenvironment (TME) is one of the major causes of treatment resistance in breast cancer. Among TME components, nervous system role in clinical outcome has been underestimated. Identifying neuronal signatures associated with treatment response will help to characterize neuronal influence on tumor progression and identify new treatment targets. The search for hormonotherapy-predictive biomarkers was implemented by supervised machine learning (ML) analysis on merged transcriptomics datasets from public databases. ML-derived genes were investigated by pathway enrichment analysis, and potential gene signatures were curated by removing the variables that were not strictly nervous system specific. The predictive and prognostic abilities of the generated signatures were examined by Cox models, in the initial cohort and seven external cohorts. Generated signature performances were compared with 14 other published signatures, in both the initial and external cohorts. Underlying biological mechanisms were explored using deconvolution tools (CIBERSORTx and xCell). Our pipeline generated two nervous system-related signatures of 24 genes and 97 genes (NervSign24 and NervSign97). These signatures were prognostic and hormonotherapy-predictive, but not chemotherapy-predictive. When comparing their predictive performance with 14 published risk signatures in six hormonotherapy-treated cohorts, NervSign97 and NervSign24 were the two best performers. Pathway enrichment score and deconvolution analysis identified brain neural progenitor presence and perineural invasion as nervous system-related mechanisms positively associated with NervSign97 and poor clinical prognosis in hormonotherapy-treated patients. Transcriptomic profiling has identified two nervous system–related signatures that were validated in clinical samples as hormonotherapy-predictive signatures, meriting further exploration of neuronal component involvement in tumor progression.Significance:The development of personalized and precision medicine is the future of cancer therapy. With only two gene expression signatures approved by FDA for breast cancer, we are in need of new ones that can reliably stratify patients for optimal treatment. This study provides two hormonotherapy-predictive and prognostic signatures that are related to nervous system in TME. It highlights tumor neuronal components as potential new targets for breast cancer therapy.

Supplementary Table S1: R package referencesSupplementary Table S2: Merged Affymetrix cohort patient characteristicsSupplementary Table S3: Validation cohort patient characteristicsSupplementary Table S4: Algorithm parameter optimizationSupplementary Figure S1: Patient selection and algorithm pipeline configurationSupplementary Figure S2: Generation of NervSign24Supplementary Figure S3: ML-generated gene lists for the first signature and second signature generationsSupplementary Figure S4: Supplementary information related to Figure 2Supplementary Figure S5: Supplementary information related to Figure 3Supplementary Figure S6: Univariate Cox analyses in cohorts of patients with early or late recurrenceSupplementary Figure S7: Supplementary information related to Figure 4Supplementary Figure S8: Multivariable and univariate Cox analyses in merged Affymetrix, METABRIC and Saal cohortsSupplementary Figure S9: Correlation between nervous system mechanisms and NervSign97Supplementary Figure S10: Kaplan-Meier analysis of previously published biomarkers

International audience; We build and study a data-driven procedure for the estimation of the stationary density f of an additive fractional SDE. To this end, we also prove some new concentrations bounds for discrete observations of such dynamics in stationary regime.

We give bounds for the total variation distance between the solutions to two stochastic differential equations starting at the same point and with close coefficients, which applies in particular to the distance between an exact solution and its Euler-Maruyama scheme in small time. We show that for small $t$, the total variation distance is of order $t^{r/(2r+1)}$ if the noise coefficient $\sigma$ of the SDE is elliptic and $\mathcal{C}^{2r}_b$, $r\in \mathbb{N}$ and if the drift is $C^1$ with bounded derivatives, using multi-step Richardson-Romberg extrapolation. We do not require the drift to be bounded. Then we prove with a counterexample that we cannot achieve a bound better than $t^{1/2}$ in general., Comment: 20 pages

International audience; In this paper we consider the drift estimation problem for a general differential equation driven by an additive multidimensional fractional Brownian motion, under ergodic assumptions on the drift coefficient. Our estimation procedure is based on the identification of the invariant measure, and we provide consistency results as well as some information about the convergence rate. We also give some examples of coefficients for which the identifiability assumption for the invariant measure is satisfied.

A probabilistic reconstruction of genealogies in a polyploid population (from 2x to 4x) is investigated, by considering genetic data analyzed as the probability of allele presence in a given genotype. Based on the likelihood of all possible crossbreeding patterns, our model enables us to infer and to quantify the whole potential genealogies in the population. We explain in particular how to deal with the uncertain allelic multiplicity that may occur with polyploids. Then we build an \textit{ad hoc} penalized likelihood to compare genealogies and to decide whether a particular individual brings sufficient information to be included in the taken genealogy. This decision criterion enables us in a next part to suggest a greedy algorithm in order to explore missing links and to rebuild some connections in the genealogies, retrospectively. As a by-product, we also give a way to infer the individuals that may have been favored by breeders over the years. In the last part we highlight the results given by our model and our algorithm, firstly on a simulated population and then on a real population of rose bushes. Most of the methodology relies on the maximum likelihood principle and on graph theory., Comment: 26 pages, 14 figures, 3 tables

We investigate the problem of the rate of convergence to equilibrium for ergodic stochastic differential equations driven by fractional Brownian motion with Hurst parameter $H\in(1/3,1)$ and multiplicative noise component $\sigma$. When $\sigma$ is constant and for every $H\in(0,1)$, it was proved in [Ann. Probab. 33 (2005) 703–758] that, under some mean-reverting assumptions, such a process converges to its equilibrium at a rate of order $t^{-\alpha}$ where $\alpha\in(0,1)$ (depending on $H$). In [Ann. Inst. Henri Poincaré Probab. Stat. 53 (2017) 503–538], this result has been extended to the multiplicative case when $H>1/2$. In this paper, we obtain these types of results in the rough setting $H\in(1/3,1/2)$. Once again, we retrieve the rate orders of the additive setting. Our methods also extend the multiplicative results of [Ann. Inst. Henri Poincaré Probab. Stat. 53 (2017) 503–538] by deleting the gradient assumption on the noise coefficient $\sigma$. The main theorems include some existence and uniqueness results for the invariant distribution.

International audience; In the continuity of a recent paper ([6]), dealing with finite Markov chains, this paper proposes and analyzes a recursive algorithm for the approximation of the quasi-stationary distribution of a general Markov chain living on a compact metric space killed in finite time. The idea is to run the process until extinction and then to bring it back to life at a position randomly chosen according to the (possibly weighted) empirical occupation measure of its past positions. General conditions are given ensuring the convergence of this measure to the quasi-stationary distribution of the chain. We then apply this method to the numerical approximation of the quasi-stationary distribution of a diffusion process killed on the boundary of a compact set and to the estimation of the spectral gap of irreducible Markov processes. Finally, the sharpness of the assumptions is illustrated through the study of the algorithm in a non-irreducible setting.

The goal of this paper is to present a series of recent contributions on some various problems of numerical probability. Beginning with the Richardson-Romberg Multilevel Monte-Carlo method which, among other fields of applications, is a very efficient method for the approximation of diffusion processes, we focus on some adaptive multilevel splitting algorithms for rare event simulation. Then, the third part is devoted to the simulation of McKean-Vlasov forward and decoupled forward-backward stochastic differential equations by some cubature algorithms. Finally, we tackle the problem of the weak error estimation in total variation norm for a general Markov semi-group.

International audience; We investigate a weighted Multilevel Richardson-Romberg extrapolation for the ergodic approximation of invariant distributions of diffusions adapted from the one introduced in~[Lemaire-Pag\`es, 2013] for regular Monte Carlo simulation. In a first result, we prove under weak confluence assumptions on the diffusion, that for any integer $R\ge2$, the procedure allows us to attain a rate $n^{\frac{R}{2R+1}}$ whereas the original algorithm convergence is at a weak rate $n^{1/3}$. Furthermore, this is achieved without any explosion of the asymptotic variance. In a second part, under stronger confluence assumptions and with the help of some second order expansions of the asymptotic error, we go deeper in the study by optimizing the choice of the parameters involved by the method. In particular, for a given $\varepsilon>0$, we exhibit some semi-explicit parameters for which the number of iterations of the Euler scheme required to attain a Mean-Squared Error lower than $\varepsilon^2$ is about $\varepsilon^{-2}\log(\varepsilon^{-1})$. Finally, we numerically this Multilevel Langevin estimator on several examples including the simple one-dimensional Ornstein-Uhlenbeck process but also on a high dimensional diffusion motivated by a statistical problem. These examples confirm the theoretical efficiency of the method.

We investigate the problem of the rate of convergence to equilibrium for ergodic stochastic differential equations driven by fractional Brownian motion with Hurst parameter $H\textgreater{}1/2$ and multiplicative noise component $\sigma$. When $\sigma$ is constant and for every $H\in(0,1)$, it was proved in \cite{hairer} that, under some mean-reverting assumptions, such a process converges to its equilibrium at a rate of order $t^{-\alpha}$ where $\alpha\in(0,1)$ (depending on $H$). The aim of this paper is to extend such types of results to some multiplicative noise setting. More precisely, we show that we can recover such convergence rates when $H\textgreater{}1/2$ and the inverse of the diffusion coefficient $\sigma$ is a Jacobian matrix. The main novelty of this work is a type of extension of Foster-Lyapunov like techniques to this non-Markovian setting, which allows us to put in place an asymptotic coupling scheme such as in \cite{hairer} without resorting to deterministic contracting properties.

This paper deals with a natural stochastic optimization procedure derived from the so-called Heavy-ball method differential equation, which was introduced by Polyak in the 1960s with his seminal contribution [Pol64]. The Heavy-ball method is a second-order dynamics that was investigated to minimize convex functions f . The family of second-order methods recently received a large amount of attention, until the famous contribution of Nesterov [Nes83], leading to the explosion of large-scale optimization problems. This work provides an in-depth description of the stochastic heavy-ball method, which is an adaptation of the deterministic one when only unbiased evalutions of the gradient are available and used throughout the iterations of the algorithm. We first describe some almost sure convergence results in the case of general non-convex coercive functions f . We then examine the situation of convex and strongly convex potentials and derive some non-asymptotic results about the stochastic heavy-ball method. We end our study with limit theorems on several rescaled algorithms., 39 pages, 3 pages

We consider a sequence ( ξ n ) n ≥ 1 of i.i.d. random values residing in the domain of attraction of an extreme value distribution. For such a sequence, there exist ( a n ) and ( b n ) , with a n > 0 and b n ∈ R for every n ≥ 1 , such that the sequence ( X n ) defined by X n = ( max ( ξ 1 , … , ξ n ) − b n ) / a n converges in distribution to a non-degenerated distribution. In this paper, we show that ( X n ) can be viewed as an Euler scheme with a decreasing step of an ergodic Markov process solution to a SDE with jumps and we derive a functional limit theorem for the sequence ( X n ) from some methods used in the long time numerical approximation of ergodic SDEs.

International audience; We study the rate of convergence of some recursive procedures based on some "exact" or "approximate" Euler schemes which converge to the invariant measure of an ergodic SDE driven by a Lévy process. The main interest of this work is to compare the rates induced by exact and approximate Euler schemes. In our main result, we show that replacing the small jumps by a Brownian component in the approximate case preserves the rate induced by the exact Euler scheme for a large class of Lévy processes.

With a view to numerical applications we address the following question: given an ergodic Brownian diffusion with a unique invariant distribution, what are the invariant distributions of the duplicated system consisting of two trajectories? We mainly focus on the interesting case where the two trajectories are driven by the same Brownian path. Under this assumption, we first show that uniqueness of the invariant distribution (weak confluence) of the duplicated system is essentially always true in the one-dimensional case. In the multidimensional case, we begin by exhibiting explicit counter-examples. Then, we provide a series of weak confluence criterions (of integral type) and also of a.s. pathwise confluence, depending on the drift and diffusion coefficients through a non-infinitesimal Lyapunov exponent. As examples, we apply our criterions to some non-trivially confluent settings such as classes of gradient systems with non-convex potentials or diffusions where the confluence is generated by the diffusive component. We finally establish that the weak confluence property is connected with an optimal transport problem. ¶ As a main application, we apply our results to the optimization of the Richardson–Romberg extrapolation for the numerical approximation of the invariant measure of the initial ergodic Brownian diffusion.

This note provides several recent progresses in the study of long time behavior of Markov processes. The examples presented below are related to other scientific fields as PDE's, physics or biology. The involved mathematical tools as propagation of chaos, coupling, functional inequalities, provide a good picture of the classical methods that furnish quantitative rates of convergence to equilibrium., arXiv admin note: text overlap with arXiv:1405.2573

In a previous paper, we studied the ergodic properties of an Euler scheme of a stochastic differential equation with a Gaussian additive noise in order to approximate the stationary regime of such equation. We now consider the case of multiplicative noise when the Gaussian process is a fractional Brownian Motion with Hurst parameter H>1/2 and obtain some (functional) convergences properties of some empirical measures of the Euler scheme to the stationary solutions of such SDEs., To appear in Stochastic Processes and their Applications

International audience; In this paper, we consider a class of diffusion processes based on a memory gradient descent, i.e. whose drift term is built as the average all along the past of the trajectory of the gradient of a coercive function U . Under some classical assumptions on U , this type of diffusion is ergodic and admits a unique invariant distribution. In view to optimization applications, we want to understand the behaviour of the invariant distribution when the diffusion coefficient goes to 0. In the non-memory case, the invariant distribution is explicit and the so-called Laplace method shows that a Large Deviation Principle (LDP) holds with an explicit rate function, that leads to a concentration of the invariant distribution around the global minima of U . Here, except in the linear case, we have no closed formula for the invariant distribution but we show that a LDP can still be obtained. Then, in the one- dimensional case, we get some bounds for the rate function that lead to the concentration around the global minimum under some assumptions on the second derivative of U .

41p.; International audience; In some recent papers, some procedures based on some weighted empirical measures related to decreasing-step Euler schemes have been investigated to approximate the stationary regime of a diffusion (possibly with jumps) for a class of functionals of the process. This method is efficient but needs the computation of the function at each step. To reduce the complexity of the procedure (especially for functionals), we propose in this paper to study a new scheme, called mixed-step scheme where we only keep some regularly time-spaced values of the Euler scheme. Our main result is that, when the coefficients of the diffusion are smooth enough, this alternative does not change the order of the rate of convergence of the procedure. We also investigate a Richardson-Romberg method to speed up the convergence and show that the variance of the original algorithm can be preserved under a uniqueness assumption for the invariant distribution of the ''duplicated'' diffusion, condition which is extensively discussed in the paper. Finally, we end by giving some sufficient ''asymptotic confluence'' conditions for the existence of a smooth solution to a discrete version of the associated Poisson equation, condition which is required to ensure the rate of convergence results.

We extend to Lipschitz continuous functionals either of the true paths or of the Euler scheme with decreasing step of a wide class of Brownian ergodic diffusions, the Central Limit Theorems formally established for their marginal empirical measure of these processes (which is classical for the diffusions and more recent as concerns their discretization schemes). We illustrate our results by simulations in connection with barrier option pricing., Comment: 33 pages

This paper is concerned with the estimation of the volatility process in a stochastic volatility model of the following form: $dX_t=a_tdt+\sigma_tdW_t$, where $X$ denotes the log-price and $\sigma$ is a c\`adl\`ag semi-martingale. In the spirit of a series of recent works on the estimation of the cumulated volatility, we here focus on the instantaneous volatility for which we study estimators built as finite differences of the \textit{power variations} of the log-price. We provide central limit theorems with an optimal rate depending on the local behavior of $\sigma$. In particular, these theorems yield some confidence intervals for $\sigma_t$., Comment: Submitted to Statistical Inference for Stochastic Processes. 28 pages, 4 figures

In this paper, we are interested in a diffusion process based on a gradient descent. The process is non Markov and has a memory term which is built as a weighted average of the drift term all along the past of the trajectory. For this type of diffusion, we study the long time behaviour of the process in terms of the memory. We exhibit some conditions for the long-time stability of the dynamical system and then provide, when stable, some convergence properties of the occupation measures and of the marginal distribution, to the associated steady regimes. When the memory is too long, we show that in general, the dynamical system has a tendency to explode, and in the particular Gaussian case, we explicitly obtain the rate of divergence.

We study sequences of empirical measures of Euler schemes associated to some non-Markovian SDEs: SDEs driven by Gaussian processes with stationary increments. We obtain the functional convergence of this sequence to a stationary solution to the SDE. Then, we end the paper by some specific properties of this stationary solution. We show that, in contrast to Markovian SDEs, its initial random value and the driving Gaussian process are always dependent. However, under an integral representation assumption, we also obtain that the past of the solution is independent of the future of the underlying innovation process of the Gaussian driving process.

We build a sequence of empirical measures on the space $\mathbb{D}(\mathbb{R}_{+},\mathbb{R}^{d})$ of $ℝ^d$-valued cadlag functions on $ℝ_+$ in order to approximate the law of a stationary $ℝ^d$-valued Markov and Feller process $(X_t)$. We obtain some general results on the convergence of this sequence. We then apply them to Brownian diffusions and solutions to Lévy-driven SDE’s under some Lyapunov-type stability assumptions. As a numerical application of this work, we show that this procedure provides an efficient means of option pricing in stochastic volatility models.

We study some recursive procedures based on exact or approximate Euler schemes with decreasing step to compute the invariant measure of Lévy driven SDEs. We prove the convergence of these procedures toward the invariant measure under weak conditions on the moment of the Lévy process and on the mean-reverting of the dynamical system. We also show that an a.s. CLT for stable processes can be derived from our main results. Finally, we illustrate our results by several simulations.

In this thesis, we focus on three problems related to the ergodicity of stochastic dynamics with memory (in a discrete-time or continuous-time setting) and especially of Stochastic Differential Equations (SDE) driven by fractional Brownian motion. In the first chapter, we study the long-time behavior of a general class of discrete-time stochastic dynamics driven by an ergodic and stationary Gaussian noise. Following the seminal paper written by Hairer (2005) on the ergodicity of fractional SDE (see also Fontbona-Panloup (2017) and Deya-Panloup-Tindel (2019)), we first build a Markovian structure above the dynamics, we show existence and uniqueness of the invariant distribution and then we exhibit some upper-bounds on the rate of convergence to equilibrium in terms of the asymptotic behavior of the covariance function of the Gaussian noise (or more precisely, of the asymptotic behavior of the coefficients appearing in its moving average representation). The second chapter establishes long-time concentration inequalities both for functionals of the whole solution on an interval [0,T] of an additive fractional SDE and for functionals of discrete-time observations of this process. Then, we apply this general result to specific functionals related to discrete and continuous-time occupation measures of the process. The last chapter, which uses the results developed in Chapter 2, is a joint work with Panloup and Tindel which focuses on the parametric estimation of the (non-linear) drift term in an additive fractional SDE. We use a minimum contrast estimation based on the identification of the invariant distribution (for which we build an approximation from discrete-time observations of the SDE). We provide consistency results as well as non-asymptotic estimates of the corresponding quadratic error. Some of our results are illustrating through numerical discussions. We also give some examples for which the identifiability condition related to our estimation procedure (intrinsically linked to the invariant distribution) is fulfilled.; Dans cette thèse, nous nous intéressons à trois problèmes en lien avec l'ergodicité de dynamiques aléa-toires à mémoire (discrètes ou continues) et tout particulièrement des Équations Différentielles Stochas-tiques (EDS) dirigées par un mouvement brownien fractionnaire. Le premier chapitre porte sur l'étude du comportement en temps long pour une classe générale de dynamiques aléatoires discrètes dirigées par un processus gaussien stationnaire ergodique. En s'inspirant des travaux de Hairer (2005), Fontbona-Panloup (2017), Deya-Panloup-Tindel (2019) sur l'ergodicité des EDS fractionnaires, nous construisons une structure markovienne au-dessus de la dynamique considérée, nous démontrons l'existence et l'unicité d'une mesure invariante puis nous donnons une borne sur la vitesse de convergence de la loi du processus vers cette mesure. La vitesse obtenue dépend du comportement asymp-totique de la fonction de covariance du processus gaussien qui dirige la dynamique (ou plus précisément de celui des coefficients intervenant dans sa représentation en moyenne mobile). Le deuxième chapitre expose des résultats sur la concentration en temps long à la fois pour des fonctionnelles de la solution d'une EDS fractionnaire additive sur un intervalle [0,T] et pour des fonctionnelles d'observations discrètes de ce processus. Ce résultat général est ensuite appliqué à des fonctionnelles spécifiques liées aux mesures d'occupations (discrètes ou continues) de la solution de l'EDS. Le dernier chapitre, dont les résultats utilisent ceux du chapitre 2, est un travail effectué en collaboration avec Panloup et Tindel qui porte sur l'estimation paramétrique du drift (non linéaire) pour une EDS fractionnaire additive. Nous utilisons une estimation par minimum de contraste basée sur l'identification de la mesure invariante (dont une approximation est construite à partir d'observations discrètes de l'EDS). Nous démontrons la consistance des estimateurs considérés et obtenons des bornes non asymptotiques sur l'erreur quadratique. Nos résultats sont illustrés par des simulations numériques. Enfin, nous montrons sur une classe d'exemples que l'hypothèse d'identifiabilité relative à ce problème d'estimation (intrinsèquement liée à la mesure invariante) est satisfaite.

In this thesis, we focus on three problems related to the ergodicity of stochastic dynamics with memory (in a discrete-time or continuous-time setting) and especially of Stochastic Differential Equations (SDE) driven by fractional Brownian motion. In the first chapter, we study the long-time behavior of a general class of discrete-time stochastic dynamics driven by an ergodic and stationary Gaussian noise. Following the seminal paper written by Hairer (2005) on the ergodicity of fractional SDE (see also Fontbona-Panloup (2017) and Deya-Panloup-Tindel (2019)), we first build a Markovian structure above the dynamics, we show existence and uniqueness of the invariant distribution and then we exhibit some upper-bounds on the rate of convergence to equilibrium in terms of the asymptotic behavior of the covariance function of the Gaussian noise (or more precisely, of the asymptotic behavior of the coefficients appearing in its moving average representation). The second chapter establishes long-time concentration inequalities both for functionals of the whole solution on an interval [0,T] of an additive fractional SDE and for functionals of discrete-time observations of this process. Then, we apply this general result to specific functionals related to discrete and continuous-time occupation measures of the process. The last chapter, which uses the results developed in Chapter 2, is a joint work with Panloup and Tindel which focuses on the parametric estimation of the (non-linear) drift term in an additive fractional SDE. We use a minimum contrast estimation based on the identification of the invariant distribution (for which we build an approximation from discrete-time observations of the SDE). We provide consistency results as well as non-asymptotic estimates of the corresponding quadratic error. Some of our results are illustrating through numerical discussions. We also give some examples for which the identifiability condition related to our estimation procedure (intrinsically linked to the invariant distribution) is fulfilled.; Dans cette thèse, nous nous intéressons à trois problèmes en lien avec l'ergodicité de dynamiques aléa-toires à mémoire (discrètes ou continues) et tout particulièrement des Équations Différentielles Stochas-tiques (EDS) dirigées par un mouvement brownien fractionnaire. Le premier chapitre porte sur l'étude du comportement en temps long pour une classe générale de dynamiques aléatoires discrètes dirigées par un processus gaussien stationnaire ergodique. En s'inspirant des travaux de Hairer (2005), Fontbona-Panloup (2017), Deya-Panloup-Tindel (2019) sur l'ergodicité des EDS fractionnaires, nous construisons une structure markovienne au-dessus de la dynamique considérée, nous démontrons l'existence et l'unicité d'une mesure invariante puis nous donnons une borne sur la vitesse de convergence de la loi du processus vers cette mesure. La vitesse obtenue dépend du comportement asymp-totique de la fonction de covariance du processus gaussien qui dirige la dynamique (ou plus précisément de celui des coefficients intervenant dans sa représentation en moyenne mobile). Le deuxième chapitre expose des résultats sur la concentration en temps long à la fois pour des fonctionnelles de la solution d'une EDS fractionnaire additive sur un intervalle [0,T] et pour des fonctionnelles d'observations discrètes de ce processus. Ce résultat général est ensuite appliqué à des fonctionnelles spécifiques liées aux mesures d'occupations (discrètes ou continues) de la solution de l'EDS. Le dernier chapitre, dont les résultats utilisent ceux du chapitre 2, est un travail effectué en collaboration avec Panloup et Tindel qui porte sur l'estimation paramétrique du drift (non linéaire) pour une EDS fractionnaire additive. Nous utilisons une estimation par minimum de contraste basée sur l'identification de la mesure invariante (dont une approximation est construite à partir d'observations discrètes de l'EDS). Nous démontrons la consistance des estimateurs considérés et obtenons des bornes non asymptotiques sur l'erreur quadratique. Nos résultats sont illustrés par des simulations numériques. Enfin, nous montrons sur une classe d'exemples que l'hypothèse d'identifiabilité relative à ce problème d'estimation (intrinsèquement liée à la mesure invariante) est satisfaite.

In this thesis, we are studying severa! stochastic algorithms with different purposes and this is why we will start this manuscript by giving historicals results to define the framework of our work. Then, we will study a bandit algorithm due to the work of Narendra and Shapiro whose objectif was to determine among a choice of severa! sources which one is the most profitable without spending too much times on the wrong orres. Our goal is to understand the weakness of this algorithm in order to propose an optimal procedure for a quantity measuring the performance of a bandit algorithm, the regret. In our results, we will propose an algorithm called NS over-penalized which allows to obtain a minimax regret bound. A second work will be to understand the convergence in law of this process. The particularity of the algorith is that it converges in law toward a non-diffusive process which makes the study more intricate than the standard case. We will use coupling techniques to study this process and propose rates of convergence. The second work of this thesis falls in the scope of optimization of a function using a stochastic algorithm. We will study a stochastic version of the so-called heavy bali method with friction. The particularity of the algorithm is that its dynamics is based on the ali past of the trajectory. The procedure relies on a memory term which dictates the behavior of the procedure by the form it takes. In our framework, two types of memory will investigated : polynomial and exponential. We will start with general convergence results in the non-convex case. In the case of strongly convex functions, we will provide upper-bounds for the rate of convergence. Finally, a convergence in law result is given in the case of exponential memory. The third part is about the McKean-Vlasov equations which were first introduced by Anatoly Vlasov and first studied by Henry McKean in order to mode! the distribution function of plasma. Our objective is to propose a stochastic algorithm to approach the invariant distribution of the McKean Vlasov equation. Methods in the case of diffusion processes (and sorne more general pro cesses) are known but the particularity of McKean Vlasov process is that it is strongly non-linear. Thus, we will have to develop an alternative approach. We will introduce the notion of asymptotic pseudotrajectory in odrer to get an efficient procedure.; Dans cette thèse, nous étudions des thématiques autour des algorithmes stochastiques et c'est pour cette raison que nous débuterons ce manuscrit par des éléments généraux sur ces algorithmes en donnant des résultats historiques pour poser les bases de nos travaux. Ensuite, nous étudierons un algorithme de bandit issu des travaux de N arendra et Shapiro dont l'objectif est de déterminer parmi un choix de plusieurs sources laquelle profite le plus à l'utilisateur en évitant toutefois de passer trop de temps à tester celles qui sont moins per­formantes. Notre but est dans un premier temps de comprendre les faiblesses structurelles de cet algorithme pour ensuite proposer une procédure optimale pour une quantité qui mesure les performances d'un algorithme de bandit, le regret. Dans nos résultats, nous proposerons un algorithme appelé NS sur-pénalisé qui permet d'obtenir une borne de regret optimale au sens minimax au travers d'une étude fine de l'algorithme stochastique sous-jacent à cette procédure. Un second travail sera de donner des vitesses de convergence pour le processus apparaissant dans l'étude de la convergence en loi de l'algorithme NS sur-pénalisé. La par­ticularité de l'algorithme est qu'il ne converge pas en loi vers une diffusion comme la plupart des algorithmes stochastiques mais vers un processus à sauts non-diffusif ce qui rend l'étude de la convergence à l'équilibre plus technique. Nous emploierons une technique de couplage afin d'étudier cette convergence. Le second travail de cette thèse s'inscrit dans le cadre de l'optimisation d'une fonc­tion au moyen d'un algorithme stochastique. Nous étudierons une version stochastique de l'algorithme déterministe de boule pesante avec amortissement. La particularité de cet al­gorithme est d'être articulé autour d'une dynamique qui utilise une moyennisation sur tout le passé de sa trajectoire. La procédure fait appelle à une fonction dite de mémoire qui, selon les formes qu'elle prend, offre des comportements intéressants. Dans notre étude, nous verrons que deux types de mémoire sont pertinents : les mémoires exponentielles et poly­nomiales. Nous établirons pour commencer des résultats de convergence dans le cas général où la fonction à minimiser est non-convexe. Dans le cas de fonctions fortement convexes, nous obtenons des vitesses de convergence optimales en un sens que nous définirons. En­fin, l'étude se termine par un résultat de convergence en loi du processus après une bonne renormalisation. La troisième partie s'articule autour des algorithmes de McKean-Vlasov qui furent intro­duit par Anatoly Vlasov et étudié, pour la première fois, par Henry McKean dans l'optique de la modélisation de la loi de distribution du plasma. Notre objectif est de proposer un al­gorithme stochastique capable d'approcher la mesure invariante du processus. Les méthodes pour approcher une mesure invariante sont connues dans le cas des diffusions et de certains autre processus mais ici la particularité du processus de McKean-Vlasov est de ne pas être une diffusion linéaire. En effet, le processus a de la mémoire comme les processus de boule pesante. De ce fait, il nous faudra développer une méthode alternative pour contourner ce problème. Nous aurons besoin d'introduire la notion de pseudo-trajectoires afin de proposer une procédure efficace.