Abstract:It is well known that fluctuations of marginals of high-dimensional random vectors that satisfy a certain concentration estimate called the thin shell condition are approximately Gaussian. In this article we identify a general condition on a sequence of high-dimensional random vectors under which one can identify the exponential decay rate of large deviation probabilities of the corresponding sequence of marginals. More precisely, consider the projection of an n-dimensional random vector onto a random k n -dimensional basis, k n ≤ n , drawn uniformly from the Haar measure on the Stiefel manifold of orthonormal k n -frames in R n , in three different asymptotic regimes as n → ∞ : “constant” ( k n = k ), “sublinear” ( k n → ∞ but k n / n → 0 ) and “linear” ( k n / n → λ with 0 < λ ≤ 1 ). When the sequence of random vectors satisfies a certain “asymptotic thin shell condition”, we establish large deviation principles for the corresponding sequence of random projections in the constant regime, and for the sequence of empirical measures of the coordinates of the random projections in the sublinear and linear regimes. We also establish large deviation principles for scaled ℓ q norms of the random projections in all three regimes. Moreover, we show that the asymptotic thin shell condition holds for various sequences of random vectors of interest, including the uniform measure on suitably scaled ℓ p n balls, for p ∈ [ 1 , ∞ ) , and generalized Orlicz balls defined via a superquadratic function, as well as a class of Gibbs measures with superquadratic interaction potential. Along the way, we obtain logarithmic asymptotics of volumes of high-dimensional Orlicz balls, which may be of independent interest. We also show that the decay rate of large deviation probabilities of Euclidean norms of multi-dimensional projections of ℓ p n balls, when p ∈ [ 1 , 2 ) , exhibits an unexpected phase transition in the sublinear regime, thus disproving an earlier conjecture due to Alonso-Gutiérrez et al. Random projections of high-dimensional random vectors are of interest in a range of fields including asymptotic convex geometry and high-dimensional statistics.

Off-diagonal elements of projection matrices and dimension asymptotics

Deviation Inequalities on the Spectral Norm of Random Matrices with Dependent Entries

On asymptotics of eigenvectors of large sample covariance matrix

An asymptotic thin shell condition and large deviations for random multidimensional projections

Invariance principle of random projection for the norm

Asymptotics for high-dimensional covariance matrices and quadratic forms with applications to the trace functional and shrinkage

Inference for Projection Parameters in Linear Regression: beyond $d = o(n^{1/2})$

Random weighted projections, random quadratic forms and random eigenvectors

EIGENVALUE DISTRIBUTIONS OF VARIANCE COMPONENTS ESTIMATORS IN HIGH-DIMENSIONAL RANDOM EFFECTS MODELS

Random projections beyond zero overlap

Asymptotic Theory of Least Distances Estimate in Multivariate Linear Models

High-dimensional analysis of double descent for linear regression with random projections

A Note on the Convergence Rate of the Spectral Distributions of Large Sample Covariance Matrices

Accurate Analysis of Sparse Random Projections

Asymptotics of estimators for structured covariance matrices

On the mean projection theorem for determinantal point processes

Reviving pseudo-inverses: Asymptotic properties of large dimensional Moore-Penrose and Ridge-type inverses with applications

Non-asymptotic theory of random matrices: extreme singular values

On Extreme Value Asymptotics of Projected Sample Covariances in High Dimensions with Applications in Finance and Convolutional Networks

The High-Dimensional Asymptotics of Principal Component Regression

Quenched large deviation principles for random projections of $\ell_p^n$ balls