A Maxwell principle for generalized Orlicz balls

DOI10.1214/22-AIHP1298MaRDI QIDQ6147644zbMATH OpenOpenAlexFDO

Authors Samuel Johnston, Joscha Prochno

Publication date 16 January 2024

Published in Annales de l'Institut Henri Poincaré. Probabilités et Statistiques (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2012.11568

zbMATH Keywords

large deviation principleGibbs measuresGibbs conditioning principlegeneralized Orlicz ballsMaxwell principlelow-dimensional projectionsquantitative Cramér theorem

Mathematics Subject Classification ID

Large deviations (60F10) Central limit and other weak theorems (60F05) Convex sets in (n) dimensions (including convex hypersurfaces) (52A20)

Abstract: In [A dozen de {F}inetti-style results in search of a theory, Ann. Inst. H. Poincar'{e} Probab. Statist. 23(2)(1987), 397--423], Diaconis and Freedman studied low-dimensional projections of random vectors from the Euclidean unit sphere and the simplex in high dimensions, noting that the individual coordinates of these random vectors look like Gaussian and exponential random variables respectively. In subsequent works, Rachev and R"uschendorf and Naor and Romik unified these results by establishing a connection between $el{l}_p^N$ balls and a $p$-generalized Gaussian distribution. In this paper, we study similar questions in a significantly generalized and unifying setting, looking at low-dimensional projections of random vectors uniformly distributed on sets of the form [B_{phi,t}^N := Big{(s_1,ldots,s_N)inmathbb{R}^N : sum_{ i =1}^Nphi(s_i)leq t NBig},] where $phi:mathbbRo[0,infty]$ is a potential (including the case of Orlicz functions). Our method is different from both Rachev-R"uschendorf and Naor-Romik, based on a large deviation perspective in the form of quantitative versions of Cram'er's theorem and the Gibbs conditioning principle, providing a natural framework beyond the $p$-generalized Gaussian distribution while simultaneously unraveling the role this distribution plays in relation to the geometry of $el{l}_p^N$ balls. We find that there is a critical parameter $t_{mathrmcrit}$ at which there is a phase transition in the behaviour of the projections: for $t>t_{mathrmcrit}$ the coordinates of random points sampled from $B_{phi,t}^N$ behave like uniform random variables, but for $tleq{t}_{mathrmcrit}$ the Gibbs conditioning principle comes into play, and here there is a parameter $$ (the inverse temperature) such that the coordinates are approximately distributed according to a density proportional to $$.

Recommendations

• Volume properties of high-dimensional Orlicz balls
• Sanov-type large deviations and conditional limit theorems for high-dimensional Orlicz balls
• The maximum entropy principle and volumetric properties of Orlicz balls
• High-dimensional limit theorems for random vectors in \(\ell_p^n\)-balls. II
• Thin-shell concentration for random vectors in Orlicz balls via moderate deviations and Gibbs measures

Cites work

• scientific article; zbMATH DE number 3349081 (Why is no real title available?)
• A Note on the Zero-one Law
• A conditional limit theorem for high-dimensional \(\ell^p\)-spheres
• A dozen de Finetti-style results in search of a theory
• An asymptotic thin shell condition and large deviations for random multidimensional projections
• Approximate independence of distributions on spheres and their stability properties
• Conditional limit theorems for exponential families and finite versions of de Finetti's theorem
• High-dimensional limit theorems for random vectors in \(\ell_p^n\)-balls
• Large deviations techniques and applications.
• On Deviations of the Sample Mean
• On the Volume of the Intersection of Two L n p Balls
• On the isotropic constant of random polytopes with vertices on an _p-sphere
• Projecting the surface measure of the sphere of \({\ell}_p^n\)
• Saddlepoint Approximations in Statistics
• The isotropic constant of random polytopes with vertices on convex surfaces
• The maximum entropy principle and volumetric properties of Orlicz balls
• Thin-shell concentration for random vectors in Orlicz balls via moderate deviations and Gibbs measures
• Yet another note on the arithmetic-geometric mean inequality

Cited in

(3)

• Sanov-type large deviations and conditional limit theorems for high-dimensional Orlicz balls
• Limit theorems for mixed-norm sequence spaces with applications to volume distribution
• A probabilistic approach to Lorentz balls \(\ell_{q, 1}^n\)

This page was built for publication: A Maxwell principle for generalized Orlicz balls

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6147644)