?
Statistical inference for Bures-Wasserstein barycenters
Annals of Applied Probability. 2021. Vol. 31. No. 3. P. 1264-1298.
n this work we introduce the concept of Bures-Wasserstein barycenter $Q_*$, that is essentially a Fr\'echet mean of some distribution $P$ supported on a subspace of positive semi-definite Hermitian operators $\mathbb{H}_{+}(d)$.
We allow a barycenter to be constrained to some affine subspace of $\mathbb{H}_{+}(d)$ and provide conditions ensuring its existence and uniqueness.
We also investigate convergence and concentration properties of an empirical counterpart of $Q_*$ in both Frobenius norm and Bures-Wasserstein distance, and explain, how obtained results are connected to optimal transportation theory and can be applied to statistical inference in quantum mechanics.
Publication based on the results of:
Carlier G., Eichinger K., Kroshnin A., SIAM Journal on Mathematical Analysis 2021 Vol. 53 No. 5 P. 5880-5914
In this paper, we investigate properties of entropy-penalized Wasserstein barycenters introduced in [J. Bigot, E. Cazelles, and N. Papadakis, SIAM J. Math. Anal., 51 (2019), pp. 2261--2285] as a regularization of Wasserstein barycenters [M. Agueh and G. Carlier, SIAM J. Math. Anal., 43 (2011), pp. 904--924]. After characterizing these barycenters in terms of a system of Monge--Ampère ...
Added: October 27, 2021
Kroshnin A., Journal of Convex Analysis 2018 Vol. 25 No. 4 P. 1371-1395
We consider the space P(X) of probability measures on arbitrary Radon space X endowed with a transportation cost J(μ, ν) generated by a nonnegative continuous cost function. For a probability distribution on P(X) we formulate a notion of average with respect to this transportation cost, called here the Fréchet barycenter, prove a version of the law ...
Added: November 23, 2018
Durmus A., Moulines E., Naumov A. et al., / Cornell University. Series arXiv "math". 2023.
In this paper, we establish novel deviation bounds for additive functionals of geometrically ergodic Markov chains similar to Rosenthal and Bernstein-type inequalities for sums of independent random variables. We pay special attention to the dependence of our bounds on the mixing time of the corresponding chain. Our proof technique is, as far as we know, ...
Added: June 18, 2023
Kolesnikov A., Milman E., Annals of Probability 2018 Vol. 46 No. 6 P. 3578-3615
What is the optimal way to cut a convex bounded domain K in Euclidean space (Rn,|⋅|) into two halves of equal volume, so that the interface between the two halves has least surface area? A conjecture of Kannan, Lov\'asz and Simonovits asserts that, if one does not mind gaining a universal numerical factor (independent of n) in the surface area, ...
Added: March 26, 2018
Shchegolev A., Управление большими системами: сборник трудов 2023 № 102 С. 5-14
The class of nonlinear Markov processes is characterized by the dependence of the current state of the process on its current distribution in addition to the dependence on the previous state. Due to this feature, these processes are characterized by complex limit behavior and ergodic properties, for which the usual criteria for Markov processes are ...
Added: June 12, 2023
Nikita Puchkin, Vladimir Ulyanov, Annales de l'institut Henri Poincare (B) Probability and Statistics 2023 Vol. 59 No. 3 P. 1508-1529
We show that external randomization may enforce the convergence of test statistics to their limiting distributions in particular cases. This results in a sharper inference. Our approach is based on a central limit theorem for weighted sums. We apply our method to a family of rank-based test statistics and a family of phi-divergence test statistics ...
Added: September 3, 2023
Goetze F., Naumov A.A., Tikhomirov A., Bernoulli: a journal of mathematical statistics and probability 2017 Vol. 23 No. 4B P. 3067-3113
In this paper we consider the product of two independent random matrices X^(1) and X^(2). Assume that X_{jk}^{(q)},1\le j,k \le n,q=1,2,, are i.i.d. random variables with \EX_{jk}^{q}=0, VarX_{jk}^{(q)}=1/ Denote by s_1(W),…,s_n(W) the singular values of W:=n^{-1}X^(1)X^(2). We prove the central limit theorem for linear statistics of the squared singular values s_1^2(W),…,s_n^2(W) showing that the limiting variance depends on \kappa_4:=\E(X_{11}^{(1)})^4−3. ...
Added: April 28, 2018
Bufetov A. I., Mkrtchyan S., Scherbina M. et al., / Cornell University. Series math "arxiv.org". 2013. No. 1301.0342.
We show that beta ensembles in Random Matrix Theory with generic real analytic potential have the asymptotic equipartition property. In addition, we prove a Central Limit Theorem for the density of the eigenvalues of these ensembles. ...
Added: February 21, 2013
Kolesnikov A., Tikhonov S. Y., / Cornell University. Series math "arxiv.org". 2012. No. 1203.3457.
Let $\mu = e^{-V} \ dx$ be a probability measure and $T = \nabla \Phi$ be the optimal transportation mapping pushing forward $\mu$ onto a log-concave compactly supported measure $\nu = e^{-W} \ dx$. In this paper, we introduce a new approach to the regularity problem for the corresponding Monge--Amp{\`e}re equation $e^{-V} = \det D^2 ...
Added: March 28, 2013
Annales de l'institut Henri Poincare (B) Probability and Statistics 2017 Vol. 53 No. 4 P. 1719-1746
Nous relions les inégalités de transport-entropie à l'étude des points critiques de la fonction. Cette approche conduit en particulier à une nouvelle preuve du résultat de Otto et Villani 2000 montrant que le droit logarithmique de Sobolev implique l'inégalité des transports chez Talagrand. ...
Added: June 7, 2018
Delon J., Salomon J., Sobolevski A., SIAM Journal of Discrete Mathematics 2012 Vol. 26 No. 2 P. 801-827
In this paper, we introduce a class of local indicators that enable us to compute efficiently optimal transport plans associated with arbitrary weighted distributions of N demands and M supplies in R in the case where the cost function is concave. Indeed, whereas this problem can be solved linearly when the cost is a convex ...
Added: May 30, 2012
Kolesnikov A., Werner E., Advances in Mathematics 2022 Vol. 396 Article 108110
Motivated by the geodesic barycenter problem from optimal transportation theory, we prove a natural generalization of the Blaschke–Santaló inequality and the affine isoperimetric inequalities for many sets and many functions. We derive from it an entropy bound for the total Kantorovich cost appearing in the barycenter problem. We also establish a “pointwise Prékopa–Leindler inequality” and show a monotonicity property of the multimarginal Blaschke–Santaó functional. ...
Added: December 4, 2021
Tolmachov K., / Cornell University. Series arXiv:1202.1806 "ArXiv.org". 2012.
In this paper we compute the precise asymptotics of the variance of linear statistic of descents on a growing interval for Plancherel Young diagrams (following Vershik and Kerov, diagrams are considered rotated by π/4). We also give an example of a local configuration with linearly growing variance in a fixed regime and prove the central limit ...
Added: December 8, 2013
Kelbert M., Karpikov I., Science and Business: Ways of Development 2018 Vol. 79 No. 1 P. 56-68
This article gives a brief summary on the main theoretical and practical results for the Scale functions. The article is organized in the following way: the first part describes the main theoretical concepts of Lévy processes, gives the formal definition and analytical properties of the Scale function. The second part describes the most significant practical ...
Added: April 5, 2018
Panov V., / Cornell University. Series arXiv "math". 2017. No. 1703.10463.
In this paper, we study the fluctuations of sums of random variables with distribution defined as a mixture of light-tail and truncated heavy-tail distributions. We focus on the case when both the mixing coefficient and the truncation level depend on the number of summands. The aim of this research is to characterize the limiting distributions ...
Added: March 31, 2017
Zhivotovskiy N., Electronic Journal of Probability 2020 Vol. 25 P. 1-30
This paper is devoted to uniform versions of the Hanson-Wright inequality for a random vector X∈RnX∈Rn with independent subgaussian components. The core technique of the paper is based on the entropy method combined with truncations of both gradients of functions of interest and of the components of XX itself. Our results recover, in particular, the ...
Added: August 2, 2020
Nikitin Y. Y., Petrov V. V., Zaitsev A. Y. et al., Vestnik of the St. Petersburg University: Mathematics 2018 Vol. 51 No. 2 P. 201-232
This is the first in a series of reviews devoted to the scientific achievements of the Leningrad–St. Petersburg school of probability and statistics in the period from 1947 to 2017. It is devoted to limit theorems for sums of independent random variables—a traditional subject for St. Petersburg. It refers to the classical limit theorems: the ...
Added: October 1, 2019
Spokoiny V., Theory of Probability and Its Applications 2014 Vol. 58 No. 2 P. 314-323
In this paper we derive a sharp concentration inequality for the supremum of a smooth random field over a finite dimensional set. It is shown that this supremum can be bounded with high probability by the value of the field at some deterministic point plus an intrinsic dimension of the optimization problem. As an application ...
Added: October 22, 2014
Le Gouic T., Paris Q., Rigollet P. et al., Journal of the European Mathematical Society 2022 P. 1-17
This work establishes fast rates of convergence for empirical barycenters over a large class of geodesic spaces with curvature bounds in the sense of Alexandrov. More specifically, we show that parametric rates of convergence are achievable under natural conditions that characterize the bi-extendibility of geodesics emanating from a barycenter. These results largely advance the state-of-the-art ...
Added: November 14, 2019
Gribkova N., Probability and Mathematical Statistics 2017 Vol. 37 No. 1 P. 101-118
In this paper, we propose a new approach to the investigation of asymptotic properties of trimmed L-statistics and we apply it to the Cramér type large deviation problem. Our results can be compared with those in Callaert et al. (1982) – the first and, as far as we know, the single article where some results ...
Added: February 28, 2020
Bressaud X., Bufetov A.I., Hubert P., Proceedings of the London Mathematical Society 2014 Vol. 109 No. 2 P. 483-522
Deviation of ergodic sums is studied for substitution dynamical systems with a matrix that admits eigenvalues of modulus 1. The functions γ we consider are the corresponding eigenfunctions. In Theorem 1.1, we prove that the limit inferior of the ergodic sums (n,γ(x_0)+⋯+γ(x_{n−1})) n∈N is bounded for every point x in the phase space. In Theorem ...
Added: October 23, 2014
Kroshnin A., Tupitsa Nazarii, Dvinskikh D. et al., , in : Proceedings of Machine Learning Research. Vol. 97: International Conference on Machine Learning, 9-15 June 2019, Long Beach, California, USA.: PMLR, 2019. P. 3530-3540.
We study the complexity of approximating the Wasserstein barycenter of m discrete measures, or histograms of size n, by contrasting two alternative approaches that use entropic regularization. The first approach is based on the Iterative Bregman Projections (IBP) algorithm for which our novel analysis gives a complexity bound proportional to $m n^2 / \epsilon^2$ to approximate the original non-regularized barycenter. ...
Added: June 11, 2019
Bobkov S., Ulyanov V. V., Theory of Probability and Its Applications 2022 Vol. 66 No. 4 P. 537-549
We give a short overview of the results related to the refined forms of the central limit theorem, with a focus on independent integer-valued random variables (r.v.'s). In the independent and non-identically distributed (non-i.i.d.) case, an approximation is then developed for the distribution of the sum by means of the Chebyshev--Edgeworth correction containing the moments ...
Added: February 22, 2022
Sirotin V., Arkhipova M., Dubrova T. A. et al., Bielsko-Biala : University of Bielsko-Biala Press, 2016
The main attributes of modern enterprises should be the flexibility and the ability of forecasting the future. Constant adaptation to the changing environment and the rapidity of undertaking certain actions which are conditioned by specific situations determine the rules for the future position of market competition. Effective and efficient adjustment of the company in line ...
Added: November 2, 2016