Ilya Molchanov’s research while affiliated with Institute of Mathematical Statistics and other places

The standard closed convex hull of a set is defined as the intersection of all images, under the action of a group of rigid motions, of a half-space containing the given set. In this paper we propose a generalisation of this classical notion, that we call a (K, ℍ)-hull, and which is obtained from the above construction by replacing a half-space with some other closed convex subset K of the Euclidean space, and a group of rigid motions by a subset ℍ of the group of invertible affine transformations. The main focus is on the analysis of (K, ℍ)-convex hulls of random samples from K.

Since its introduction by J. Karamata, regular variation has evolved from a purely mathematical concept into a cornerstone of theoretical probability and data analysis. It is extensively studied and applied in different areas. Its significance lies in characterising large deviations, determining the limits of partial sums, and predicting the long-term behaviour of extreme values in stochastic processes. Motivated by various applications, the framework of regular variation has expanded over time to incorporate random observations in more general spaces, including Banach spaces and Polish spaces. In this monograph, we identify three fundamental components of regular variation: scaling, boundedness, and the topology of the underlying space. We explore the role of each component in detail and extend a number of previously obtained results to general topological spaces. Our more abstract approach unifies various concepts appearing in the literature, streamlines existing proofs and paves the way for novel contributions, such as: a generalised theory of (hidden) regular variation for random measures and sets; an innovative treatment of regularly varying random functions and elements scaled by independent random quantities and numerous other advancements. Throughout the text, key results and definitions are illustrated by instructive examples, including extensions of several established models from the literature. By bridging abstraction with practicality, this work aims to deepen both theoretical understanding and methodological applicability of regular variation.

We obtain a complete characterization of planar monotone $\sigma$-continuous valuations taking integer values, without assuming invariance under any group of transformations. We further investigate the consequences of dropping monotonicity or $\sigma$-continuity and give a full classification of line valuations. We also introduce a construction of the product for valuations of this type.

We study vantage-point trees constructed using an independent sample from the uniform distribution on a fixed convex body K in $(\mathbb {R}^d,\Vert \cdot \Vert )$, where $\Vert \cdot \Vert$ is an arbitrary norm on $\mathbb {R}^d$. We prove that a sequence of sets, associated with the left boundary of a vantage-point tree, forms a recurrent Harris chain on the space of convex bodies in $(\mathbb {R}^d,\Vert \cdot \Vert )$. The limiting object is a ball polyhedron, that is, an a.s. finite intersection of closed balls in $(\mathbb {R}^d,\Vert \cdot \Vert )$ of possibly different radii. As a consequence, we derive a limit theorem for the length of the leftmost path of a vantage-point tree.

We consider Gaussian approximation in a variant of the classical Johnson–Mehl birth–growth model with random growth speed. Seeds appear randomly in $\mathbb{R}^d$ at random times and start growing instantaneously in all directions with a random speed. The locations, birth times, and growth speeds of the seeds are given by a Poisson process. Under suitable conditions on the random growth speed, the time distribution, and a weight function $h\;:\;\mathbb{R}^d \times [0,\infty) \to [0,\infty)$, we prove a Gaussian convergence of the sum of the weights at the exposed points, which are those seeds in the model that are not covered at the time of their birth. Such models have previously been considered, albeit with fixed growth speed. Moreover, using recent results on stabilization regions, we provide non-asymptotic bounds on the distance between the normalized sum of weights and a standard Gaussian random variable in the Wasserstein and Kolmogorov metrics.

We consider the normal approximation of Kabanov–Skorohod integrals on a general Poisson space. Our bounds are for the Wasserstein and the Kolmogorov distance and involve only difference operators of the integrand of the Kabanov–Skorohod integral. The proofs rely on the Malliavin–Stein method and, in particular, on multiple applications of integration by parts formulae. As examples, we study some linear statistics of point processes that can be constructed by Poisson embeddings and functionals related to Pareto optimal points of a Poisson process.

We study empirical variants of the halfspace (Tukey) depth of a probability measure $\mu$, which are obtained by replacing $\mu$ with the corresponding weighted empirical measure. We prove analogues of the Marcinkiewicz--Zygmund strong law of large numbers and of the law of the iterated logarithm in terms of set inclusions and for the Hausdorff distance between the theoretical and empirical variants of depth trimmed regions. In the special case of $\mu$ being the uniform distribution on a convex body K, the depth trimmed regions are convex floating bodies of K, and we obtain strong limit theorems for their empirical estimators.

We introduce a hull operator on Poisson point processes, the easiest example being the convex hull of the support of a point process in Euclidean space. Assuming that the intensity measure of the process is known on the set generated by the hull operator, we discuss estimation of the expected linear statistics built on the Poisson process. In special cases, our general scheme yields an estimator of the volume of a convex body or an estimator of an integral of a H\"older function. We show that the estimation error is given by the Kabanov--Skorohod integral with respect to the underlying Poisson process. A crucial ingredient of our approach is a spatial Markov property of the underlying Poisson process with respect to the hull. We derive the rate of normal convergence for the estimation error, and illustrate it on an application to estimators of integrals of a H\"older function. We also discuss estimation of higher order symmetric statistics.