Large deviation principles induced by the Stiefel manifold, and random multidimensional projections

DOI10.1214/23-EJP1023MaRDI QIDQ6177516zbMATHOpenAlexFDO

Authors Steven Soojin Kim, Kavita Ramanan

Publication date 17 January 2024

Published in Electronic Journal of Probability (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2105.04685

zbMATH Keywords

Stiefel manifoldlarge deviationsrandom projectionsrate functionquenchedvariational formulaannealedasymptotic convex geometry\(\ell_p^n\) balls

Mathematics Subject Classification ID

Large deviations (60F10) Random matrices (probabilistic aspects) (60B20) Asymptotic theory of convex bodies (52A23)

Abstract: Given an $n$-dimensional random vector $X^{(n)}$ , for $k<n$, consider its $k$-dimensional projection $mathbf{a}_{n,k}{X}^{(n)}$, where $mathbf{a}_{n,k}$ is an $nimesk$-dimensional matrix belonging to the Stiefel manifold $mathbb{V}_{n,k}$ of orthonormal $k$-frames in $mathbb{R}^n$. For a class of sequences $X^{(n)}$ that includes the uniform distributions on scaled $el{l}_p^n$ balls, $pin(1,infty]$, and product measures with sufficiently light tails, it is shown that the sequence of projected vectors $mathbf{a}_{n,k}^{i}ntercal{X}^{(n)}$ satisfies a large deviation principle whenever the empirical measures of the rows of $sqrtnmathbf{a}_{n,k}$ converge, as $nightarrowinfty$, to a probability measure on $mathbb{R}^k$. In particular, when $mathbf{A}_{n,k}$ is a random matrix drawn from the Haar measure on $mathbb{V}_{n,k}$, this is shown to imply a large deviation principle for the sequence of random projections $mathbf{A}_{n,k}^{i}ntercal{X}^{(n)}$ in the quenched sense (that is, conditioned on almost sure realizations of $mathbf{A}_{n,k}$). Moreover, a variational formula is obtained for the rate function of the large deviation principle for the annealed projections $mathbf{A}_{n,k}^{i}ntercal{X}^{(n)}$, which is expressed in terms of a family of quenched rate functions and a modified entropy term. A key step in this analysis is a large deviation principle for the sequence of empirical measures of rows of $sqrtnmathbf{A}_{n,k}$, which may be of independent interest. The study of multi-dimensional random projections of high-dimensional measures is of interest in asymptotic functional analysis, convex geometry and statistics. Prior results on quenched large deviations for random projections of $el{l}_p^n$ balls have been essentially restricted to the one-dimensional setting.

Recommendations

• Large deviations for random matrices in the orthogonal group and Stiefel manifold with applications to random projections of product distributions
• Large deviations for uniform projections of $p$-radial distributions on $\ell_p^n$-balls
• Large deviations for random projections of \(\ell^{p}\) balls
• An asymptotic thin shell condition and large deviations for random multidimensional projections

Cites work

• A Fourier view on the \(R\)-transform and related asymptotics of spherical integrals
• A central limit theorem for convex sets
• A comparison principle for functions of a uniformly random subspace
• A conditional limit theorem for high-dimensional \(\ell^p\)-spheres
• A new look at random projections of the cube and general product measures
• Aging of spherical spin glasses
• An asymptotic thin shell condition and large deviations for random multidimensional projections
• Approximate independence of distributions on spheres and their stability properties
• Asymptotic power of sphericity tests for high-dimensional data
• Asymptotics of graphical projection pursuit
• Convex Analysis
• Cramér's theorem is atypical
• Extensions of Lipschitz mappings into a Hilbert space
• Generalized Dirichlet distributions on the ball and moments
• High-dimensional limit theorems for random vectors in \(\ell_p^n\)-balls
• Large deviations for high-dimensional random projections of \(\ell_p^n\)-balls
• Large deviations for random projections of \(\ell^{p}\) balls
• Large deviations for transient random walks in random environment on a Galton-Watson tree
• On general minimax theorems
• On the Volume of the Intersection of Two L n p Balls
• Projections of probability distributions: a measure-theoretic Dvoretzky theorem
• Quenched, annealed and functional large deviations for one-dimensional random walk in random environment
• Sanov's theorem in the Wasserstein distance: a necessary and sufficient condition
• Sharp asymptotics for \(q\)-norms of random vectors in high-dimensional \(\ell_p^n\)-balls
• Signal detection in high dimension: the multispiked case

Cited in

(1)

• Large deviations for random matrices in the orthogonal group and Stiefel manifold with applications to random projections of product distributions

This page was built for publication: Large deviation principles induced by the Stiefel manifold, and random multidimensional projections

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6177516)