Ergodic Averages Research Articles

Pseudo-marginal Metropolis–Hastings sampling using averages of unbiased estimators

SummaryWe consider a pseudo-marginal Metropolis–Hastings kernel ${\mathbb{P}}_m$ that is constructed using an average of $m$ exchangeable random variables, and an analogous kernel ${\mathbb{P}}_s$ that averages $s<m$ of these same random variables. Using an embedding technique to facilitate comparisons, we provide a lower bound for the asymptotic variance of any ergodic average associated with ${\mathbb{P}}_m$ in terms of the asymptotic variance of the corresponding ergodic average associated with ${\mathbb{P}}_s$. We show that the bound is tight and disprove a conjecture that when the random variables to be averaged are independent, the asymptotic variance under ${\mathbb{P}}_m$ is never less than $s/m$ times the variance under ${\mathbb{P}}_s$. The conjecture does, however, hold for continuous-time Markov chains. These results imply that if the computational cost of the algorithm is proportional to $m$, it is often better to set $m=1$. We provide intuition as to why these findings differ so markedly from recent results for pseudo-marginal kernels employing particle filter approximations. Our results are exemplified through two simulation studies; in the first the computational cost is effectively proportional to $m$ and in the second there is a considerable start-up cost at each iteration.

On the homogeneous ergodic bilinear averages with Möbius and Liouville weights

It is shown that the homogeneous ergodic bilinear averages with Möbius or Liouville weight converge almost surely to zero; that is, if T is a map acting on a probability space (X,A,ν), and a,b∈Z, then for any f,g∈L2(X), for almost all x∈X, 1N∑n=1Nν(n)f(Tanx)g(Tbnx)→N→+∞0, where ν is the Liouville function or the Möbius function. We further obtain that the convergence almost everywhere holds for the short interval with the help of Zhan’s estimation. Moreover, we establish that if T is weakly mixing and its restriction to its Pinsker algebra has singular spectrum, then for any integer k≥1, for any fj∈L∞(X), j=1,…,k, for almost all x∈X, we have 1 N ∑ n = 1 N ν ( n ) ∏ j = 1 k f j ( T n j x ) → N → + ∞ 0 .

A performance analysis of ensemble averaging for high fidelity turbulence simulations at the strong scaling limit

We analyze the potential performance benefits of estimating expected quantities in large eddy simulations of turbulent flows using true ensembles rather than ergodic time averaging. Multiple realizations of the same flow are simulated in parallel, using slightly perturbed initial conditions to create unique instantaneous evolutions of the flow field. Each realization is then used to calculate statistical quantities. Provided each instance is sufficiently de-correlated, this approach potentially allows considerable reduction in the time to solution beyond the strong scaling limit for a given accuracy. This paper focuses on the theory and implementation of the methodology in Nek5000, a massively parallel open-source spectral element code.

Multiple ergodic theorems for arithmetic sets

We establish results with an arithmetic flavor that generalize the polynomial multidimensional Szemerédi theorem and related multiple recurrence and convergence results in ergodic theory. For instance, we show that in all these statements we can restrict the implicit parameter n n to those integers that have an even number of distinct prime factors or satisfy any other congruence condition. In order to obtain these refinements we study the limiting behavior of some closely related multiple ergodic averages with weights given by appropriately chosen multiplicative functions. These averages are then analyzed using a recent structural result for bounded multiplicative functions proved by the authors.

Maximal functions and ergodic averages related to Waring’s problem

We study the arithmetic analogue of maximal functions on diagonal hypersurfaces. This paper is a natural step following the papers of [13], [14] and [16]. We combine more precise knowledge of oscillatory integrals and exponential sums to generalize the asymptotic formula in Waring’s problem to an approximation formula for the Fourier transform of the solution set of lattice points on hypersurfaces arising in Waring’s problem and apply this result to arithmetic maximal functions and ergodic averages. In sufficiently large dimensions, the approximation formula, l2-maximal theorems and ergodic theorems were previously known. Our contribution is in reducing the dimensional constraint in the approximation formula using recent bounds of Wooley, and improving the range of lp spaces in the maximal and ergodic theorems. We also conjecture the expected range of spaces.

Dynamical multifractal zeta-functions, multifractal pressure and fine multifractal spectra

We introduce multifractal pressure and dynamical multifractal zeta-functions, providing precise information on a very general class of multifractal spectra, including, for example, the fine multifractal spectra of self-conformal measures and the fine multifractal spectra of ergodic Birkhoff averages of continuous functions.

Strictly Ergodic Models Under Face and Parallelepiped Group Actions

The Jewett–Krieger theorem states that each ergodic system has a strictly ergodic topological model. In this article, we show that for an ergodic system one may require more properties on its strictly ergodic model. For example, the orbit closure of points in diagonal under face transforms may be also strictly ergodic. As an application, we show the pointwise convergence of ergodic averages along cubes, which was firstly proved by Assani (J Anal Math 110:241–269, 2010).

On ergodic averages for parabolic product flows

We consider a direct product of a suspension flow over a substitution dynamical system and an arbitrary ergodic flow and give quantitative estimates for the speed of convergence for ergodic integrals of such systems. Our argument relies on new uniform estimates of the spectral measure for suspension flows over substitution dynamical systems. The paper answers a question by Jon Chaika.

Estimating the speed-up of adaptively restrained Langevin dynamics

We consider Adaptively Restrained Langevin dynamics, in which the kinetic energy function vanishes for small velocities. Properly parameterized, this dynamics makes it possible to reduce the computational complexity of updating inter-particle forces, and to accelerate the computation of ergodic averages of molecular simulations. In this paper, we analyze the influence of the method parameters on the total achievable speed-up. In particular, we estimate both the algorithmic speed-up, resulting from incremental force updates, and the influence of the change of the dynamics on the asymptotic variance. This allows us to propose a practical strategy for the parameterization of the method. We validate these theoretical results by representative numerical experiments.

On the pointwise entangled ergodic theorem

We present some twisted compactness conditions for almost everywhere convergence of one-parameter entangled ergodic averages of Dunford–Schwartz operators T0,…,Ta on a Borel probability space of the form1N∑n=1NTanAa−1Ta−1nAa−1⋅…⋅A0T0nf for f∈Lp(X,μ), p≥1. We also discuss examples and present a continuous version of the result.

Variation-norm and fluctuation estimates for ergodic bilinear averages

For any dynamical system, we show that higher variation-norms for the sequence of ergodic bilinear averages of two functions satisfy a large range of bilinear Lp estimates. It follows that, with probability one, the number of fluctuations along this sequence may grow at most polynomially with respect to (the growth of) the underlying scale. These results strengthen previous works of Lacey and Bourgain where almost surely convergence of the sequence was proved (which is equivalent to the qualitative statement that the number of fluctuations is finite at each scale). Via transference, the proof reduces to establishing new bilinear Lp bounds for variation-norms of truncated bilinear operators on R, and the main ingredient of the proof of these bounds is a variation-norm extension of maximal Bessel inequalities of Lacey and Demeter--Tao--Thiele.

Spectral Measure at Zero for Self-Similar Tilings

The goal of this paper is to study the action of the group of translations over self-similar tilings in the euclidian space $\mathbb{R}^d$. It investigates the behaviour near zero for spectral measures for such dynamical systems. Namely the paper gives a Holder asymptotic expansion near zero for these spectral measures. It is a generalization to higher dimension of a result by Bufetov and Solomyak who studied self similar-suspension flows for substitutions. The study of such asymptotics mostly involves the understanding of the deviations of some ergodic averages.

Topological pressure for the completely irregular set of birkhoff averages

It is well-known that for certain dynamical systems (satisfying specification or its variants), the set of irregular points w.r.t. a continuous function $\phi$ (i.e. points with divergent Birkhoff ergodic averages observed by $\phi$ ) either is empty or carries full topological entropy (or pressure, see [6,17,36,37] etc. for example). In this paper we study the set of irregular points w.r.t. a collection $D$ of finite or infinite continuous functions (that is, points with divergent Birkhoff ergodic averages simultaneously observed by all $\phi∈D$ ) and obtain some generalized results. As consequences, these results are suitable for systems such as mixing shifts of finite type, uniformly hyperbolic diffeomorphisms, repellers and $β-$ shifts.

A good universal weight for nonconventional ergodic averages in norm

A good universal weight for nonconventional ergodic averages in norm

Multifractal spectra and multifractal zeta-functions

We introduce multifractal zetafunctions providing precise information of a very general class of multifractal spectra, including, for example, the multifractal spectra of self-conformal measures and the multifractal spectra of ergodic Birkhoff averages of continuous functions. More precisely, we prove that these and more general multifractal spectra equal the abscissae of convergence of the associated zeta-functions.

Non-conventional ergodic averages for several commuting actions of an amenable group

Let (X, μ) be a probability space, G a countable amenable group, and (Fn)n a left Folner sequence in G. This paper analyzes the non-conventional ergodic averages $$\frac{1}{{\left| {{F_n}} \right|}}\sum\limits_{g \in {F_n}} {\mathop \Pi \limits_{i = 1}^d } \left( {{f_i} \circ T_1^g \cdots T_i^g} \right)$$ associated to a commuting tuple of μ-preserving actions \({T_1}, \ldots {T_d}:G \curvearrowright X\) and f1,..., fd ∈ L∞(μ). We prove that these averages always converge in \({\left\| \cdot \right\|_2}\), and that they witness a multiple recurrence phenomenon when f1 =... = fd = 1A for a non-negligible set A ⊆ X. This proves a conjecture of Bergelson, McCutcheon and Zhang. The proof relies on an adaptation from earlier works of the machinery of sated extensions.

Norm convergence of multiple ergodic averages on amenable groups

We apply Walsh's method for proving norm convergence of multiple ergodic averages to arbitrary amenable groups. We obtain convergence in the uniform Ces\`aro sense for their polynomial actions and for ``triangular'' averages associated to commuting homomorphic actions. The latter generalizes a result due to Bergelson, McCutcheon, and Zhang in the case of two actions.

Random differences in Szemerédi’s theorem and related results

We introduce a new, elementary method for studying random differences in arithmetic progressions and convergence phenomena along random sequences of integers. We apply our method to obtain significant improvements on two results. The first improvement is the following. Let l be a positive integer and {u 1 ≥ u 2 ≥ · · · } be a decreasing sequence of probabilities satisfying u n · n 1/(l+1)→∞. Let R = R ω be the random sequence obtained by selecting the natural number n with probability u n . Then every set A of natural numbers with positive upper density contains an arithmetic progression a, a+r, a+2r,..., a+lr of length l + 1 with difference r ∈ R ω . The best previous result (by M. Christ and us) was the condition u n ·n 2−l+1 → ∞ with a logarithmic rate. The new bound is better when l ≥ 4. Our second improvement concerns almost everywhere convergence of double ergodic averages. We construct a (random) sequence {r 1 < r 2 < · · · } of positive integers such that r n /n 2−ϵ → ∞ for all ϵ > 0 and, for any measure preserving transformation T of a probability space, the averages $$\frac{1}{N}\sum\limits_{n < N} {{T^n}{F_1}\left( x \right)} {T^{{r_n}}}{F_2}\left( x \right)$$ converge for almost every x. Our best previous result was the growth rate r n /n (1+1/14)−ϵ → ∞ of the sequence {r n }.

Criteria of Divergence Almost Everywhere in Ergodic Theory

In this expository paper, we survey nowadays classical tools or criteria used in problems of convergence everywhere to build counterexamples: the Stein continuity principle, Bourgain’s entropy criteria, and Kakutani–Rokhlin lemma, the most classical device for these questions in ergodic theory. First, we state a L1-version of the continuity principle and give an example of its usefulness by applying it to a famous problem on divergence almost everywhere of Fourier series. Next we particularly focus on entropy criteria in Lp, 2 ≤ p ≤ ∞, and provide detailed proofs. We also study the link between the associated maximal operators and the canonical Gaussian process on L2. We further study the corresponding criterion in Lp, 1 < p < 2, using properties of pstable processes. Finally, we consider Kakutani–Rokhlin’s lemma, one of the most frequently used tools in ergodic theory, by stating and proving a criterion for a.e. divergence of weighted ergodic averages. Bibliography: 38 titles.

Vector valued q-variation for differential operators and semigroups I

In this paper, we establish $\mathcal B$-valued variational inequalities for differential operators, ergodic averages and symmetric diffusion semigroups under the condition that Banach space $\mathcal B$ has martingale cotype property. These results generalize, on the one hand Pisier and Xu's result on the variational inequalities for $\mathcal B$-valued martingales, on the other hand many classical variational inequalities in harmonic analysis and ergodic theory. Moreover, we show that Rademacher cotype $q$ is necessary for the $\mathcal B$-valued $q$-variational inequalities. As applications of the variational inequalities, we deduce the jump estimates and obtain quantitative information on the rate of convergence. It turns out the rate of convergence depends on the geometric property of the Banach space under consideration, which considerably improve Cowling and Leinert's result where it is shown that the convergence always holds for all Banach spaces.