Full Large Deviation Research Articles

An exact formula for the variance of linear statistics in the one-dimensional jellium model

We consider the jellium model of N particles on a line confined in an external harmonic potential and with a pairwise one-dimensional Coulomb repulsion of strength α > 0. Using a Coulomb gas method, we study the statistics of where f(x), in principle, is an arbitrary smooth function. While the mean of s is easy to compute, the variance is nontrivial due to the long-range Coulomb interactions. In this paper we demonstrate that the fluctuations around this mean are Gaussian with a variance for large N. In this paper, we provide an exact compact formula for the constant . In addition, we also calculate the full large deviation function characterizing the tails of the full distribution for several different examples of f(x). Our analytical predictions are confirmed by numerical simulations.

Large deviation principles for renewal–reward processes

We establish a sharp large deviation principle for renewal–reward processes, supposing that each renewal involves a broad-sense reward taking values in a real separable Banach space. In fact, we demonstrate a weak large deviation principle without assuming any exponential moment condition on the law of waiting times and rewards by resorting to a sharp version of Cramér’s theory. We also exhibit sufficient conditions for exponential tightness of renewal–reward processes, which leads to a full large deviation principle.

Matrix Liberation Process II: Relation to Orbital Free Entropy

AbstractWe investigate the concept of orbital free entropy from the viewpoint of the matrix liberation process. We will show that many basic questions around the definition of orbital free entropy are reduced to the question of full large deviation principle for the matrix liberation process. We will also obtain a large deviation upper bound for a certain family of random matrices that is essential to define the orbital free entropy. The resulting rate function is made up into a new approach to free mutual information.

From the master equation to mean field game limit theory: Large deviations and concentration of measure

We study a sequence of symmetric $n$-player stochastic differential games driven by both idiosyncratic and common sources of noise, in which players interact with each other through their empirical distribution. The unique Nash equilibrium empirical measure of the $n$-player game is known to converge, as $n$ goes to infinity, to the unique equilibrium of an associated mean field game. Under suitable regularity conditions, in the absence of common noise, we complement this law of large numbers result with nonasymptotic concentration bounds for the Wasserstein distance between the $n$-player Nash equilibrium empirical measure and the mean field equilibrium. We also show that the sequence of Nash equilibrium empirical measures satisfies a weak large deviation principle, which can be strengthened to a full large deviation principle only in the absence of common noise. For both sets of results, we first use the master equation, an infinite-dimensional partial differential equation that characterizes the value function of the mean field game, to construct an associated McKean–Vlasov interacting $n$-particle system that is exponentially close to the Nash equilibrium dynamics of the $n$-player game for large $n$, by refining estimates obtained in our companion paper. Then we establish a weak large deviation principle for McKean–Vlasov systems in the presence of common noise. In the absence of common noise, we upgrade this to a full large deviation principle and obtain new concentration estimates for McKean–Vlasov systems. Finally, in two specific examples that do not satisfy the assumptions of our main theorems, we show how to adapt our methodology to establish large deviations and concentration results.

Sample path large deviations for Lévy processes and random walks with regularly varying increments

Let $X$ be a Lévy process with regularly varying Lévy measure $\nu$. We obtain sample-path large deviations for scaled processes $\bar{X}_{n}(t)\triangleq X(nt)/n$ and obtain a similar result for random walks with regularly varying increments. Our results yield detailed asymptotic estimates in scenarios where multiple big jumps in the increment are required to make a rare event happen; we illustrate this through detailed conditional limit theorems. In addition, we investigate connections with the classical large deviations framework. In that setting, we show that a weak large deviation principle (with logarithmic speed) holds, but a full large deviation principle does not hold.

Large deviation principle in one-dimensional dynamics

We study the dynamics of smooth interval maps with non-flat critical points. For every such a map that is topologically exact, we establish the full (level-2) Large Deviation Principle for empirical means. In particular, the Large Deviation Principle holds for every non-renormalizable quadratic map. This includes the maps without physical measure found by Hofbauer and Keller, and challenges the widely-shared view of the Large Deviation Principle as a refinement of laws of large numbers.

On the large deviation principle for maximum likelihood estimator of [formula omitted]-Brownian bridge

We consider the large deviation principle for maximum likelihood estimator of α-Brownian bridge. The full large deviation with explicit rate function is obtained in the case of non-Gaussian limit distribution by local large deviation principle and exponential tightness.

Large deviations for some logarithmic means in the case of random variables with thin tails

In this paper we generalize Theorem 3.3 in Giuliano and Macci (2011). More precisely we prove the full large deviation principle without assuming a particular condition in that theorem and, moreover, we give some results for the case of random variables with thin tails (and not super-exponential tails). As an application we deduce some consequences for the logarithmic means of some random variables expressed in terms of a C-process.

Energy Statistics in Open Harmonic Networks

We relate the large time asymptotics of the energy statistics in open harmonic networks to the variance-gamma distribution and prove a full Large Deviation Principle. We consider both Hamiltonian and stochastic dynamics, the later case including electronic RC networks. We compare our theoretical predictions with the experimental data obtained by Ciliberto et al. [Phys. Rev. Lett. 110, 180601 (2013)].

Large deviations on Markov towers

We give a sufficient condition to hold a full large deviation principle for Markov tower maps induced from return time functions. As an application of this result we show the large deviation principle of level 2 for some class of smooth dynamical systems with nonuniform hyperbolicity.

Large Deviations in Quantum Spin Chains

We show the full large deviation principle for KMS-states and C*-finitely correlated states on a quantum spin chain. We cover general local observables. Our main tool is Ruelle’s transfer operator method.

Large deviations and Chernoff bound for certain correlated states on a spin chain

In this paper we extend the results of Lenci and Rey-Bellet [J. Stat. Phys. 119, 715 (2005)] on the large deviation upper bound of the distribution measures of local Hamiltonians with respect to a Gibbs state in the setting of translation-invariant finite-range interactions. We show that a certain factorization property of the reference state is sufficient for a large deviation upper bound to hold and that this factorization property is satisfied by Gibbs states of the above kind as well as finitely correlated states. As an application of the methods, the Chernoff bound for correlated states with factorization property is studied. In the specific case of the distributions of the ergodic averages of a one-site observable with respect to an ergodic finitely correlated state, the spectral theory of positive maps is applied to prove the full large deviation principle.

Occupation time processes of super-Brownian motion with cut-off branching

In this article we investigate a class of superprocess with cut-off branching, studying the long-time behavior of the occupation time process. Persistence of the process holds in all dimensions. Central-limit-type theorems are obtained, and the scales are dimension dependent. The Gaussian limit holds only when d ≤ 4. In dimension one, a full large deviation principle is established and the rate function is identified explicitly. Our result shows that the super-Brownian motion with cut-off branching in dimension one has many features that are similar to super-Brownian motion in dimension three.

Occupation time processes of super-Brownian motion with cut-off branching

In this article we investigate a class of superprocess with cut-off branching, studying the long-time behavior of the occupation time process. Persistence of the process holds in all dimensions. Central-limit-type theorems are obtained, and the scales are dimension dependent. The Gaussian limit holds only when d ≤ 4. In dimension one, a full large deviation principle is established and the rate function is identified explicitly. Our result shows that the super-Brownian motion with cut-off branching in dimension one has many features that are similar to super-Brownian motion in dimension three.

Self-normalized Large Deviations for Markov Chains

We prove a self-normalized large deviation principle for sums of Banach space valued functions of a Markov chain. Self-normalization applies to situations for which a full large deviation principle is not available. We follow the lead of Dembo and Shao [DemSha98b] who state partial large deviations principles for independent and identically distributed random sequences.

Principe de grandes déviations autonormalisées pour des chaı̂nes de Markov

We prove a self-normalized large deviation principle for sums of Banach space valued functions of a Markov chain. Self-normalization applies to situations for which a domination hypothesis would be necessary in order to obtain a full large deviation principle. We follow the lead of Dembo and Shao [2] who state partial large deviations principles for independent and identically distributed random sequences.

Space–time large deviation lower bounds for spin particle systems

Abstract In this Letter we study large deviation lower bounds for space–time empirical processes of spin particle systems with possibly vanishing spin flip rates. We present an approach for obtaining such lower bounds for systems starting from almost all configurations w.r.t. an translation invariant and extremal invariant measure of the systems, so that in such cases we have full large deviation principle.

Ordered overlaps in disordered mean-field models

Let M(N) be a sequence of integers with M→∞ as N→∞ and M=o(N). For bounded i.i.d. r.v. ξik and bounded i.i.d. r.v. σi, we study the large deviation of the family of (ordered) scalar products Xk=N−1∑i=1Nσiξik,k≤M, under the distribution conditioned on the ξik's. To get a full large deviation principle, it is necessary to specify also the total norm(∑k≤M(Xk)2)1/2, which turns to be associated with some extra Gaussian distribution. Our results apply to disordered, mean-field systems, including generalized Hopfield models in the regime of a sublinear number of patterns. We build also a class of examples where this norm is the crucial order parameter.

Large deviations for empirical processes of interacting particle systems

In this paper we study process-level large deviations for spin-flip particle systems and more general Markov processes. We prove that a modification of a well known Donsker-Varadhan entropy function can be used to govern the large deviation lower bounds for any such process starting from a class of initial distributions related to an extremal invariant measure of the process. In some specific cases, we obtain the full large deviation principle.

Large deviations in the geometry of convex lattice polygons

We provide a full large deviation principle (LDP) for the uniform measure on certain ensembles of convex lattice polygons. This LDP provides for the analysis of concentration of the measure on convex closed curves. In particular, convergence to a limiting shape results in some particular cases, including convergence to a circle when the ensemble is defined as those centered convex polygons, with vertices on a scaled two dimensional lattice, and with length bounded by a constant. The Gauss-Minkowskii transform of convex curves plays a crucial role in our approach.