Exposé à Metz. Titre et résumé à venir.

Exposé à Metz. Titre et résumé à venir.

We study the optimal transport problem on the real line with the cost given by the distance, a setting in which solutions (called optimal transport plans) are typically non-unique. The first part of the talk presents a decomposition theorem: every optimal transport plan admits a unique decomposition into components, each acting on a specific region where the mass moves forward, moves backward, or remains stationary. Building on this structure, the second part investigates the behaviour of an entropically regularized version of the problem as the regularization parameter tends to zero. A natural candidate for the limit is constructed from our decomposition together with a Strassen-type theorem for a strengthened stochastic order. When the source and target distributions are sufficiently singular, the entropic minimizers converge to this plan. In general, all limit points satisfy a structural property known as weak multiplicativity.

Single-cell data yield profound insight into the complex nature of molecular feature distributions. However, they also pose statistical analysis challenges. A key challenge is the intricate geometry of these distributions, which requires non-linear analysis methods. We propose a kernel-based framework for comparing conditions in single-cell experiments that allows non-linear comparisons of different cell populations. In this talk, I will explain how embedding the data in an infinite-dimensional reproducing kernel Hilbert space (RKHS) facilitates non-linear operations on the data via linear operations in the feature space. I will present a linear model in the RKHS and introduce a truncated kernel Hotelling-Lawley statistic with an associated kernel trick. This statistic has been shown to have an asymptotic chi-squared distribution, which allows to quantify the significance of the test results. The functionality and flexibility of the proposed approach will be demonstrated on scRNA-Seq data obtained in the context of cerebral arteries profiling. The goal of this analysis is to gain insight into the appearance of intracranial aneurysms.

Plus d’informations ici.

Programme disponible ici.

Le but de cet exposé sera d’étudier le comportement limite d’arbres de Bienaymé-Galton-Watson conditionnés à avoir un grand nombre de sommets, dont un nombre fixé de feuilles ou de nœuds internes. Dans le premier cas, nous obtiendrons un résultat universel quelle que soit la loi de reproduction. En revanche, le conditionnement par le nombre de sommets et de noeuds internes donnera lieu à une diversité de comportements asymptotiques selon les propriétés de la loi de reproduction, allant de phénomènes de condensation à des structures d’arbres plus allongées.

Stochastic partial differential equations (SPDEs) are a major subject of current research in probability and analysis, with rich methodologies for studying existence and regularity. At the same time, SPDEs are increasingly used as statistical models for spatially and temporally structured data, where inference requires learning unknown parameters or functions from observations. In this talk, we consider Bayesian inference for the reaction function in a stochastic reaction-diffusion equation, based on a single solution trajectory observed continuously in space over a fixed time interval. We place a Gaussian process prior on the reaction function and derive posterior contraction rates in a novel asymptotic regime in which the spatial domain grows while the observation horizon remains fixed. In this setting, the SPDE solution becomes spatially ergodic and converges to a stationary process, which allows us to prove concentration inequalities for spatial averages of the solution. The proofs combine tools from Malliavin calculus – most notably the Clark–Ocone formula – with sharp bounds on the marginal densities of the SPDE.

The aim of this presentation is to show different results obtained in the field of semi-Markov processes focusing on risk/reliability indicators and related statistical inference aspects. The first part concerns discrete-time semi-Markov processes which are defined in a finite state space. Theoretical results are obtained in terms of evaluation and statistical estimation of risk indicators such as the mean time to failure. The asymptotic behavior of the empirical estimators is studied in the case of one single (large) trajectory. The results are illustrated on both real and simulated wind data. The second part of the presentation concerns partially observed semi-Markov chains such as the hidden Markov renewal and the hidden semi-Markov chains. Statistical estimation results are presented for reliability indicators such as the failure occurrence rate. In the third part, we present bootstrap estimators of risk indicators when the state space is discrete along with posterior concentration rates of the kernel density when the state space is continuous.

(Exposé en français avec des diapositives en anglais.)

Les structures de régularités permettent de construire des théories de solutions afin de résoudre des équations aux dérivées partielles stochastiques (EDPs).
Après une introduction aux structures de régularité, nous verrons qu’il est possible de les interpréter dans un cadre catégorique via la théorie des espèces combinatoires.
Nous verrons ensuite deux applications de ce point de vue, tout d’abord un théorème de Y. Bruned et V. Dotsenko, puis un préprint récent de Y. Bruned et P. Laubie.

Nous discuterons de la limite de faible intensité d’une mosaïque de Poisson-Voronoï, alias « ideal Poisson-Voronoi tessellation » (IPVT). Dans l’espace hyperbolique de dimension d, une simple description poissonienne de la cellule qui contient l’origine (cellule zéro) permet d’étudier des propriétés fines des tuiles de l’IPVT. Cette description poissonienne de l’IPVT reste simple dans d’autres cas, tels que le produit cartésien de plans hyperboliques.

Exposé basé sur un travail en collaboration avec Nicolas Curien, Nathanaël Enriquez, Russell Lyons et Meltem Ünel (à paraître sur Ann. Probab.), et sur 2412.00822.

Avec des réalisations physiques de la cellule zéro de l’IPVT de l’espace hyperbolique tridimensionnel dans le modèle de la boule de Poincaré (« bijou »).

The study of combinatoric properties of tilings on lattice models has a long history of interactions with both computability (e.g. the undecidability of the domino problem) and statistical physics (e.g. the Peierls argument), but the joining of those two interfaces is relatively recent. Notably, the question “chaotic temperature dependence” originates from the spin-glass literature, and has been active for the last two decades.

In this context, chaoticity can be summarised as the fact that no converging behaviour can occur in a given model as its temperature goes to 0. First formally established for an infinite spin alphabet, this property was later refined using a finite alphabet with long-range 1D interactions, and then finite-range interactions in higher dimensions.

In this talk, I will notably focus on how the simulation of Turing machines within tilings has played a key role in this evolution, up to and including a realisation result on the zero-temperature limit accumulation sets of chaotic models.

(L’exposé sera en français.)

Cet exposé porte sur les algorithmes stochastiques pour le traitement séquentiel des données. Après un rappel des techniques classiques d’approximation stochastique, je présenterai des contributions récentes concernant le développement et l’analyse d’algorithmes avancés tels que Full AdaGrad et les méthodes de Newton stochastiques. Les applications incluent la régression en ligne, l’estimation de la médiane géométrique et des cadres plus généraux. Nous discuterons des garanties théoriques, des propriétés de convergence et des expériences numériques.

Wei Lu est maître de conférences à la FST à Nancy dans notre équipe depuis le 1er septembre.

Les systèmes de particules sont des outils souvent utilisés pour modéliser des populations interagissant entre elles et/ou avec un environnement extérieur. Je présenterai un modèle appelé N-BMP (N-processus de Markov branchant) dans lequel les particules ont des trajectoires indépendantes sur la droite réelle, et se séparent en deux (on parle de branchement) à des instants aléatoires successifs indépendants et distribués selon une loi exponentielle. Pour garder une population de taille constante au cours du temps, la particule la plus basse est tuée à chaque instant de branchement. Je poserai la question de la limite hydrodynamique de ce processus, c’est-à-dire le comportement du processus obtenu quand le nombre N de particules tend vers l’infini. Le cas où les particules ont des trajectoires browniennes a notamment été étudié depuis 2017, et mis en relation avec un problème à frontière libre associé à l’équation de la chaleur. Je présenterai les résultats que nous avons obtenus, qui donnent un cadre plus général dans lequel le N-BMP a une limite hydrodynamique, en faisant intervenir une frontière analogue à celle qui apparaît dans le cas brownien. Travail en collaboration avec Jean Bérard.

Statistical methodology for spatial point patterns has traditionally focused on Euclidean data and planar surfaces. However, with recent advances in 3D biological imaging technologies targeting protein molecules on a cell’s plasma membrane, spatial point patterns are now being observed on complex shapes and manifolds whose geometry must be respected for principled inference. Consequently, there is now a demand for tools that can analyse these data for important scientific studies in cellular and micro-biology. Motivated by studying the spatial distribution of LPS proteins on the surface of E-Coli, we develop the fundamental functional summary statistics for the analysis of point patterns to general convex bounded shapes and demonstrate how they can be used to test for complete spatial randomness. We then develop their multi-type extensions, together with a test for independence of the component marginal processes. To support these methods, we introduce a plug-in estimator for the intensity of a spatial point process on a manifold. We conclude with a discussion on how these methods can readily be extended to a class of non-convex shapes. This talk will aim to provide an accessible overview of the references below.

References:

Ward, E.A.K. Cohen, N. M. Adams. Testing for complete spatial randomness on 3-dimensional bounded convex shapes. Spatial Statistics, Vol. 41, 2021.

Ward, H. S. Battey and E. A. K. Cohen. Nonparametric estimation of the intensity function of a spatial point process on a Riemannian manifold. Biometrika, Vol. 110, 2023.

Kumar, P. Inns, S. Ward, V. Lagage, J. Wang, R. Kaminska, S. Uphoff, E. A. K. Cohen, G. Mamou and C. Kleanthous. Immobile lipopolysaccharides and outer membrane proteins differentially segregate in growing Escherichia coli. Proceedings of the National Academy of Sciences, 122 (10), 2025

Ward, E. A. K. Cohen and N. M. Adams. Functional summary statistics and testing for independence in marked point patterns on the surface of three-dimensional convex shapes. Spatial Statistics, Vol. 67, 2025

We study a system of partial differential equations (PDEs) with interconnected obstacles and Neumann-type boundary conditions on a smooth bounded domain D. This system is the Hamilton-Jacobi-Bellman system of equations associated with multidimensional switching problem in finite horizon when the state process is constrained to live in the domain D. We prove the existence of a unique continuous viscosity solution. The existence of a viscosity solution is obtained using a probabilistic approach which connects the system of PDEs to a system of backward stochastic differential equations, where randomness is constrained to stay in the domain D. The second main result consists in verifying the maximum principle, which ensures the comparison between any viscosity sub-solution and super-solution of the PDEs system. This guarantees the uniqueness and continuity of the solution.

In this talk, we consider the problem of testing for the sphericity of a collection of random vectors. It is well known that in the classical elliptical model, testing for rotational symmetry of the underlying distribution is equivalent to testing that a scatter parameter is a multiple of the identity matrix. We consider the more general model of random vectors with elliptical directions introduced by R.H. Randles and present a few scenarios where testing for sphericity is still equivalent to testing that the scatter parameter is a multiple of the identity. These new scenarios include, for instance, non-classical settings where some dependence of a rather general form studied here for the first time may be present between observations. We study, under these new assumptions, the behavior of the classical spatial sign test and show that under certain mild assumptions, the test is asymptotically valid and has the same local asymptotic power as in the classical elliptical scenario. We then show that, contrary to some commonly held belief, the spatial sign test enjoys some local asymptotic optimality properties when it comes to testing for sphericity when the underlying distribution is strongly heavy-tailed.

(L’exposé sera en français, avec des slides en anglais.)