arXiv++ Combinatorics

Two versions of the Ajtai-Szemerédi Theorem are considered in the Cartesian square of a finite non-Abelian group $G$. In case $G$ is sufficiently quasirandom, we obtain strong forms of both versions: if $E \subseteq G\times G$ is fairly dense, then $E$ contains a large number of the desired patterns for most individual choices of `common difference'. For one of the versions, we also show that this set of good common differences is syndetic.

In a recent paper (arXiv:1211.6372), Bergelson and Tao proved that if $G$ is a $D$-quasi-random group, and $x$,$g$ are drawn uniformly and independently from $G$, then the quadruple $(g,x,gx,xg)$ is roughly equidistributed in the subset of $G^4$ defined by the constraint that the last two coordinates lie in the same conjugacy class. Their proof gives only a qualitative version of this result. The present notes gives a rather more elementary proof which improves this to an explicit polynomial bound in $D^{-1}$.

Let $(X,μ)$ be a probability space, $G$ a countable amenable group and $(F_n)_n$ a left Følner sequence in $G$. This paper analyzes the non-conventional ergodic averages \[\frac{1}{|F_n|}\sum_{g \in F_n}\prod_{i=1}^d (f_i\circ T_1^g\cdots T_i^g)\] associated to a commuting tuple of $μ$-preserving actions $T_1$, ..., $T_d:G\curvearrowright X$ and $f_1$, ..., $f_d \in L^\infty(μ)$. We prove that these averages always converge in $\|\cdot\|_2$, and that they witness a multiple recurrence phenomenon when $f_1 = \ldots = f_d = 1_A$ for a non-negligible set $A\subseteq X$. This proves a conjecture of Bergelson, McCutcheon and Zhang. The proof relies on an adaptation from earlier works of the machinery of sated extensions.

This paper continues an earlier work on the structure of solutions to two classes of functional equation. Let $Z$ be a compact Abelian group and $U_1$, \ldots, $U_k \leq Z$ be closed subgroups. Given $f:Z\to\mathbb{T}$ and $w \in Z$, one defines the differenced function \[d_wf(z) := f(z+w) - f(z).\] In this notation, we shall study solutions to the system of difference equations \[d_{u_1}\cdots d_{u_k}f \equiv 0 \quad \forall (u_1,\ldots,u_k) \in \prod_{i\leq k}U_i,\] and to the zero-sum problem \[f_1 + \cdots + f_k = 0\] for functions $f_i:Z\to \mathbb{T}$ that are $U_i$-invariant for each $i$. Part I of this work showed that the $Z$-modules of solutions to these problems can be described using a general theory of `almost modest $¶$-modules'. Much of the global structure of these solution $Z$-modules could then be extracted from results about the closure of this general class under certain natural operations, such as forming cohomology groups. The main result of the present paper is that solutions to either problem can always be decomposed into summands which either solve a simpler system of equations, or have some special `step polynomial' structure. This will be proved by augmenting the definition of `almost modest $\mathcal{P}$-modules' further, to isolate a subclass in which elements can be represented by the desired `step polynomials'. We will then find that this subclass is closed under the same operations.

Consider a compact Abelian group $Z$ and closed subgroups $U_1$, \ldots, $U_k \leq Z$. Let $\mathbb{T} := \mathbb{R}/\mathbb{Z}$. This paper examines two kinds of functional equation for measurable functions $Z\to \mathbb{T}$. First, given $f:Z\to \mathbb{T}$ and $w \in Z$, the resulting differenced function is \[d_wf(z) := f(z-w) - f(z).\] In this notation, we study solutions to the system of difference equations \[d_{u_1}\cdots d_{u_k}f \equiv 0 \quad \forall u_1 \in U_1,\ u_2 \in U_2,\ \ldots\ u_k \in U_k.\] Second, we study tuples of measurable functions $f_i:Z\to \mathbb{T}$ such that $f_i$ is invariant under translation by $U_i$ and also \[f_1 + \cdots + f_k = 0.\] For these equations, the solutions form a subgroup of $\mathcal{F}(Z)$ or $\mathcal{F}(Z)^k$, where $\mathcal{F}(Z)$ is the group of measurable functions $Z\to \mathbb{T}$ modulo Haar-a.e. equality. The subgroup of solutions is closed under convergence in probability and is globally invariant under rotations of $Z$, so it is a complete metrizable $Z$-module. We will give a recursive description of the structure of this $Z$-module relative to the solution-modules of lower-order equations of the same kind. These results are obtained as applications of an abstract theory of a special class of $Z$-modules. Most of our work will go into showing that this class of modules is closed under various natural operations. Knowing that, the above descriptions follow as easy consequences. Partial difference equations of the above kind can be seen as an extremal version of the inverse problem for the higher-dimensional, directional analogs of Gowers' uniformity norms. Our methods also give some information about the `stability' version of this inverse problem, which concerns functions whose Gowers norm is sufficiently close to being maximal.

We exhibit proofs of two ergodic-theoretic results in the study of multiple recurrence using an analog of the density-increment argument of Roth and Gowers: Furstenberg's Multiple Recurrence Theorem (which implies Szemerédi's Theorem), and a two-dimensional special case of Furstenberg and Katznelson's multidimensional version of this theorem. The second of these requires also an analog of some recent finitary work by Shkredov. Many proofs of these multiple recurrence theorems are now known, but our main goal is to shed some further light on the heuristic correspondence principle that has grown up between the ergodic-theoretic and combinatorial aspects of multiple recurrence and Szemerédi's Theorem. Focusing on the density-increment strategy highlights several close points of connection between these settings.

In 1975 Szemerédi proved the long-standing conjecture of Erdős and Turán that any subset of $\bbZ$ having positive upper Banach density contains arbitrarily long arithmetic progressions. Szemerédi's proof was entirely combinatorial, but two years later Furstenberg gave a quite different proof of Szemerédi's Theorem by first showing its equivalence to an ergodic-theoretic assertion of multiple recurrence, and then bringing new machinery in ergodic theory to bear on proving that. His ergodic-theoretic approach subsequently yielded several other results in extremal combinatorics, as well as revealing a range of new phenomena according to which the structures of probability-preserving systems can be described and classified. In this work I survey some recent advances in understanding these ergodic-theoretic structures. It contains proofs of the norm convergence of the `nonconventional' ergodic averages that underly Furstenberg's approach to variants of Szemerédi's Theorem, and of two of the recurrence theorems of Furstenberg and Katznelson: the Multidimensional Multiple Recurrence Theorem, which implies a multidimensional generalization of Szemerédi's Theorem; and a density version of the Hales-Jewett Theorem of Ramsey Theory.

The culmination of the papers (arXiv:0905.0518, arXiv:0910.0909) was a proof of the norm convergence in $L^2(μ)$ of the quadratic nonconventional ergodic averages \frac{1}{N}\sum_{n=1}^N(f_1\circ T_1^{n^2})(f_2\circ T_1^{n^2}T_2^n)\quad\quad f_1,f_2\in L^\infty(μ) associated to an arbitrary probability-preserving \bbZ^2-system (X,μ,T_1,T_2). This is a special case of the Bergelson-Leibman conjecture on the norm convergence of polynomial nonconventional ergodic averages. That proof relied on some new machinery for extending probability-preserving $\bbZ^d$-systems to obtain simplified asymptotic behaviour for various nonconventional averages such as the above. The engine of this machinery is formed by some detailed structure theorems for the `characteristic factors' that are available for some such averages after ascending to a suitably-extended system. However, these new structure theorems underwent two distinct phases of development, separated by the discovery of some new technical results in Moore's cohomology theory for locally compact groups (arXiv:1004.4937). That discovery enabled a significant improvement to the main structure theorem (Theorem 1.1 in (arXiv:0905.0518)), which in turn afforded a much shortened proof of convergence. However, since the proof of convergence using the original structure theorem required some quite different ideas that are now absent from these other papers, I have recorded it here in case it has some independent interest.

In this paper we combine the general tools developed in (arXiv:0905.0518) with several ideas taken from earlier work on one-dimensional nonconventional ergodic averages by Furstenberg and Weiss, Host and Kra and Ziegler to study the averages $\frac{1}{N}\sum_{n=1}^N(f_1\circ T^{n\bf{p}_1})(f_2\circ T^{n\bf{p}_2})(f_3\circ T^{n\bf{p}_3})$ for $f_1,f_2,f_3 \in L^\infty(μ)$ associated to a triple of directions $\bf{p}_1,\bf{p}_2,\bf{p}_3 \in \mathbb{Z}^2$ that lie in general position along with $0 \in \mathbb{Z}^2$. We will show how to construct a `pleasant' extension of an initially-given $\mathbb{Z}^2$-system for which these averages admit characteristic factors with a very concrete description, involving one-dimensional isotropy factors and two-step pro-nilsystems. We also use this analysis to construct pleasant extensions and then prove norm convergence for the polynomial nonconventional ergodic averages $\frac{1}{N}\sum_{n=1}^N(f_1\circ T_1^{n^2})(f_2\circ T_1^{n^2}T_2^n)$ associated to two commuting transformations $T_1$, $T_2$.

We study a generalized notion of a homogeneous skew-product extension of a probability-preserving system in which the homogeneous space fibres are allowed to vary over the ergodic decomposition of the base. The construction of such extensions rests on a simple notion of `direct integral' for a `measurable family' of homogeneous spaces, which has a number of precedents in older literature. The main contribution of the present paper is the systematic development of a formalism for handling such extensions, including non-ergodic versions of the results of Mackey describing ergodic components of such extensions, of the Furstenberg-Zimmer Structure Theory and of results of Mentzen describing the structure of automorphisms of such extensions when they are relatively ergodic. We then offer applications to two structural results for actions of several commuting transformations: firstly to describing the possible joint distributions of three isotropy factors corresponding to three commuting transformations; and secondly to describing the characteristic factors for a system of double nonconventional ergodic averages. Although both applications are modest in themselves, we hope that they point towards a broader usefulness of this formalism in ergodic theory.

In two recent papers we introduced some new techniques for constructing an extension of a probability-preserving system $T:\mathbb{Z}^d\curvearrowright (X,μ)$ that enjoys certain desirable properties in connexion with the asymptotic behaviour of some related nonconventional ergodic averages. The present paper is the first of two that will explore various refinements and extensions of these ideas. This first part is dedicated to some much more general machinery for the construction of extensions that can be used to recover various earlier results. It also contains two relatively simple new applications of this machinery to the study of certain families of nonconventional averages, one in discrete and one in continuous time (convergence being a new result for the latter). In the forthcoming second part (arXiv:0910.0907) we will introduce the problem of describing the characteristic factors and the limit of the linear nonconventional averages $\frac{1}{N}\sum_{n=1}^N \prod_{i=1}^kf_i\circ T^{n\bf{p}_i}$ when the directions $\bf{p}_1$, $\bf{p}_2$, \ldots, $\bf{p}_k \in \mathbb{Z}^d$ are not assumed to be linearly independent, and provide a fairly detailed solution in the case when k = 3, d = 2 and any pair of directions is linearly independent. This will then be used to prove the convergence in $L^2(μ)$ of the quadratic nonconventional averages $\frac{1}{N}\sum_{n=1}^N (f_1\circ T_1^{n^2})(f_2\circ T_1^{n^2}T_2^n)$.

We offer a new proof of Furstenberg and Katznelson's density version of the Hales-Jewett Theorem: For any $δ> 0$ there is some $N_0 \geq 1$ such that whenever $A \subseteq [k]^N$ with $N \geq N_0$ and $|A|\geq δk^N$, $A$ contains a \textbf{combinatorial line}: that is, for some $I \subseteq [N]$ nonempty and $w_0 \in [k]^{[N]\setminus I}$ we have A \supseteq \{w: w|_{[N]\setminus I} = w_0, w|_I = \rm{const.}\}. Following Furstenberg and Katznelson, we first show that this result is equivalent to a `multiple recurrence' assertion for a class of probability measures enjoying a certain kind of stationarity. However, we then give a quite different proof of this latter assertion through a reduction to an infinitary removal lemma in the spirit of Tao's work on infinite random hypergraphs (and also its recent re-interpretation in a new proof of the multidimensional Szemeredi Theorem by the present author). This reduction is based on a structural analysis of these stationary laws closely analogous to the classical representation theorems for various partial exchangeable stochastic processes in the sense of Hoover, Aldous and Kallenberg. However, the underlying combinatorial arguments used to prove this theorem are rather different from those required to work with exchangeable arrays, and involve crucially an observation that arose during ongoing work by a collaborative team of authors to give a purely finitary proof of the above theorem.

In his 1985 survey of notions of exchangeability, Aldous introduced a form of exchangeability corresponding to the symmetries of the infinite discrete cube, and asked whether these exchangeable probability measures enjoy a representation theorem similar to those for exchangeable sequences, arrays and set-indexed families. In this note we to prove that, whereas the known representation theorems for different classes of partially exchangeable probability measure imply that the compact convex set of such measures is a Bauer simplex (that is, its subset of extreme points is closed), in the case of cube-exchangeability it is a copy of the Poulsen simplex (in which the extreme points are dense). This follows from the arguments used by Glasner and Weiss' for their characterization of property (T) in terms of the geometry of the simplex of invariant measures for associated generalized Bernoulli actions. The emergence of this Poulsen simplex suggests that, if a representation theorem for these processes is available at all, it must take a very different form from the case of set-indexed exchangeable families.

Recent works of Alon-Shapira and Rödl-Schacht have demonstrated that every hereditary property of undirected graphs or hypergraphs is testable with one-sided error; informally, this means that if a graph or hypergraph satisfies that property "locally" with sufficiently high probability, then it can be perturbed (or "repaired") into a graph or hypergraph which satisfies that property "globally". In this paper we make some refinements to these results, some of which may be surprising. In the positive direction, we strengthen the results to cover hereditary properties of multiple directed polychromatic graphs and hypergraphs. In the case of undirected graphs, we extend the result to continuous graphs on probability spaces, and show that the repair algorithm is "local" in the sense that it only depends on a bounded amount of data; in particular, the graph can be repaired in a time linear in the number of edges. We also show that local repairability also holds for monotone or partite hypergraph properties (this latter result is also implicitly in work of Ishigami). In the negative direction, we show that local repairability breaks down for directed graphs, or for undirected 3-uniform hypergraphs. The reason for this contrast in behavior stems from (the limitations of) Ramsey theory.

De Finetti's classical result of [18] identifying the law of an exchangeable family of random variables as a mixture of i.i.d. laws was extended to structure theorems for more complex notions of exchangeability by Aldous [1,2,3], Hoover [41,42], Kallenberg [44] and Kingman [47]. On the other hand, such exchangeable laws were first related to questions from combinatorics in an independent analysis by Fremlin and Talagrand [29], and again more recently in Tao [62], where they appear as a natural proxy for the `leading order statistics' of colourings of large graphs or hypergraphs. Moreover, this relation appears implicitly in the study of various more bespoke formalisms for handling `limit objects' of sequences of dense graphs or hypergraphs in a number of recent works, including Lovász and Szegedy [52], Borgs, Chayes, Lovász, Sós, Szegedy and Vesztergombi [17], Elek and Szegedy [24] and Razborov [54,55]. However, the connection between these works and the earlier probabilistic structural results seems to have gone largely unappreciated. In this survey we recall the basic results of the theory of exchangeable laws, and then explain the probabilistic versions of various interesting questions from graph and hypergraph theory that their connection motivates (particularly extremal questions on the testability of properties for graphs and hypergraphs). We also locate the notions of exchangeability of interest to us in the context of other classes of probability measures subject to various symmetries, in particular contrasting the methods employed to analyze exchangeable laws with related structural results in ergodic theory, particular the Furstenberg-Zimmer structure theorem for probability-preserving $\mathbb {Z}$-systems, which underpins Furstenberg's ergodic-theoretic proof of Szemerédi's Theorem. The forthcoming paper [10]--hereditarytest will make a much more elaborate appeal to the link between exchangeable laws and dense (directed) hypergraphs to establish various results in property testing.