Boundaries of random walks on graphs and groups with infinitely many ends

Consider an irreducible random walk {Z _n} on a locally finite graphG with infinitely many ends, and assume that its transition probabilities are invariant under a closed group Γ of automorphisms ofG which acts transitively on the vertex set. We study the limiting behaviour of {Z _n} on the spaceΩ of ends ofG. With the exception of a degenerate case,Ω always constitutes a boundary of Γ in the sense of Furstenberg, and {Z _n} converges a.s. to a random end. In this case, the Dirichlet problem for harmonic functions is solvable with respect toΩ. The degenerate case may arise when Γ is amenable; it then fixes a unique end, and it may happen that {Z _n} converges to this end. If {Z _n} is symmetric and has finite range, this may be excluded. A decomposition theorem forΩ, which may also be of some purely graph-theoretical interest, is derived and applied to show thatΩ can be identified with the Poisson boundary, if the random walk has finite range. Under this assumption, the ends with finite diameter constitute a dense subset in the minimal Martin boundary. These results are then applied to random walks on discrete groups with infinitely many ends.     

Hypergroup deformations and Markov chains

Persi Diaconis and Phil Hanlon in their interesting paper⁽⁴⁾ give the rates of convergence of some Metropolis Markov chains on the cubeZ ^d (2). Markov chains on finite groups that are actually random walks are easier to analyze because the machinery of harmonic analysis is available. Unfortunately, Metropolis Markov chains are, in general, not random walks on group structure. In attempting to understand Diaconis and Hanlon's work, the authors were led to the idea of a hypergroup deformation of a finite groupG, i.e., a continuous family of hypergroups whose underlying space isG and whose structure is naturally related to that ofG. Such a deformation is provided forZ ^d (2), and it is shown that the Metropolis Markov chains studied by Diaconis and Hanlon can be viewed as random walks on the deformation. A direct application of the Diaconis-Shahshahani Upper Bound Lemma, which applies to random walks on hypergroups, is used to obtain the rate of convergence of the Metropolis chains starting at any point. When the Markov chains start at 0, a result in Diaconis and Hanlon⁽⁴⁾ is obtained with exactly the same rate of convergence. These results are extended toZ ^d (3).Research supported in part by the Office of Research and Sponsored Programs, University of Oregon.     

Deviation inequality for monotonic Boolean functions with application to the number of k‐cycles in a random graph

Using Talagrand's concentration inequality on the discrete cube {0, 1}^m we show that given a real‐valued function Z(x) on {0, 1}^m that satisfies certain monotonicity conditions one can control the deviations of Z(x) above its median by a local Lipschitz norm of Z at the point x. As one application, we obtain a deviation inequality for the number of k‐cycles in a random graph. © 2003 Wiley Periodicals, Inc. Random Struct. Alg., 2004     

A characterization of a class of [Z] groups via Korovkin theory

We characterize all the central topological groupsG for which the centreZ(L ¹(G)) of the group algebra admits a finite universal Korovkin set. It is proved thatZ(L ¹(G)) has a finite universal Korovkin set iffĜ is a finite dimensional, separable metric space. This is equivalent to the fact thatG is separable, metrizable andG/K has finite torsion free rank, whereK is a compact open normal subgroup of certain direct summand ofG.     

Lower degree bounds for modular invariants and a question of I. Hughes

We prove two statements. The first one is a conjecture of Ian Hughes which states that iff ₁, ..., f_n are primary invariants of a finite linear groupG, then the least common multiple of the degrees of thef _i is a multiple of the exponent ofG.The second statement is about vector invariants: IfG is a permutation group andK a field of positive characteristicp such thatp divides |G|, then the invariant ringK[V ^m]^G ofm copies of the permutation moduleV overK requires a generator of degreem(p–1). This improves a bound given by Richman [6], and implies that there exists no degree bound for the invariants ofG that is independent of the representation.     

Markov random fields and gibbs random fields

Spitzer has shown that every Markov random field (MRF) is a Gibbs random field (GRF) and vice versa when (i) both are translation invariant, (ii) the MRF is of first order, and (iii) the GRF is defined by a binary, nearest neighbor potential. In both cases, the field (iv) is defined onZ ^v, and (v) at anyxεZ^v, takes on one of two states. The current paper shows that a MRF is a GRF and vice versa even when (i)−(v) are relaxed, i.e., even if one relaxes translation invariance, replaces first order bykth order, allows for many states and replaces finite domains of Z^v by arbitrary finite sets. This is achieved at the expense of using a many body rather than a pair potential, which turns out to be natural even in the classical (nearest neighbor) case when Z^v is replaced by a triangular lattice. The contents of this paper were presented in August, 1971, at a seminar of the Battelle Rencontre in Statistical Mechanics and also at a pair of seminars in December, 1971, at the Weizmann Institute of Science. Partially supported by NSF GP 7469 and a Weizmann Institute senior fellowship while on sabbatical leave from Indiana University.     

Stochastic Dynamics of Compact Spins: Ergodicity and Irreducibility

Stochastic dynamics associated with Gibbs measures on M ^{Z d} , where M is a compact Riemannian manifold and Z ^d is an integer lattice, is considered. Equivalence of its L ²-ergodicity and the extremality of the corresponding Gibbs measure is proved.     

On the full automorphism group of a graph

While it is easy to characterize the graphs on which a given transitive permutation groupG acts, it is very difficult to characterize the graphsX with Aut (X)=G. We prove here that for the certain transitive permutation groups a simple necessary condition is also sufficient. As a corollary we find that, whenG is ap-group with no homomorphism ontoZ _p wrZ _p , almost all Cayley graphs ofG have automorphism group isomorphic toG.     

On infinite tournaments with regular automorphism groups

Atournament regular representation (TRR) of an abstract groupG is a tournamentT whose automorphism group is isomorphic toG and is a regular permutation group on the vertices ofT. L. Babai and W. Imrich have shown that every finite group of odd order exceptZ ₃ ×Z ₃ admits a TRR. In the present paper we give several sufficient conditions for an infinite groupG with no element of order 2 to admit a TRR. Among these are the following: (1)G is a cyclic extension byZ of a finitely generated group; (2)G is a cyclic extension byZ _2n+1 of any group admitting a TRR; (3)G is a finitely generated abelian group; (4)G is a countably generated abelian group whose torsion subgroup is finite.     

Finite nearest particle systems

Summary Finite nearest particle systems are certain continuous time Markov chains on the collection of finite subsets ofZ ¹. In this paper, we give a sufficient condition for such a system to survive, in the sense that the probability of absorption at 0 is less than one. This theorem generalizes earlier results for the one-dimensional contact process.Research supported in part by NSF Grant MCS83-00836     

Random Measures and Applications

Abstract

A mapping Z(·) from a δ-ring ?₀(?) into the vector space of random variables L ^p (P) is a vector-valued measure if it is σ-additive in the metric of its range. It is a vector measure if the range is a Banach space and a random measure if also its values are independent on disjoint sets. An important reason for this study is to construct integrals relative to such Zs, which typically do not have finite variation. For this, it is essential to find a controlling (σ-finite) measure for Z that is not available if 0 <p < 1, and here the random measure is taken to be p-stable and utilize properties of infinitely divisible distributions. In the case of p = 2, Z(·) induces a bimeasure, and if p > 2 is an integer it induces a polymeasure, either of which need not be (signed) measures on product spaces. Important applications lead to all these possibilities. In all those cases, a detailed analysis of vector-valued set functions is presented, with special focus for the cases of 0 <p < 1 and p = 2 where probability and Bochner's L ^{2, 2} boundedness plays a key role. Specialization if Z is stationary, harmonizable, and/or isotropic are discussed using the group structure of ? ⁿ , n ≥ 1, extending it for an lca group G. If Z is Banach valued or a quasi-martingale measure, methods of obtaining integrals are outlined in the last section, and open problems motivated by applications are pointed out at various places.     

Decomposition of ranges of vector measures

The following conditions on a zonoidZ, i.e., a range of a non-atomic vector measure, are equivalent: (i) the extreme set containing 0 in its relative interior is a parallelepiped; (ii) the zonoidZ determines them-range of any non-atomic vector measure with rangeZ, where them-range of a vector measure μ is the set ofm-tuples (μ(S ₁), …, μ(S _m), whereS ₁, …S _m are disjoint measurable sets and (iii) there is avector measure space (X, Σ, μ) such that any finite factorization ofZ, Z =ΣZ _i , in the class of zonoids could be achieved by decomposing (X, Σ). In the case of ranges of non-atomic probability measures (i) is automatically satisfied, so (ii) and (iii) hold. Partially supported by NSF grant MCS-79-06634     

Properties of a representation of a basis for the null space

Given a rectangular matrixA(x) that depends on the independent variablesx, many constrained optimization methods involve computations withZ(x), a matrix whose columns form a basis for the null space ofA ^T(x). WhenA is evaluated at a given point, it is well known that a suitableZ (satisfyingA ^T Z = 0) can be obtained from standard matrix factorizations. However, Coleman and Sorensen have recently shown that standard orthogonal factorization methods may produce orthogonal bases that do not vary continuously withx; they also suggest several techniques for adapting these schemes so as to ensure continuity ofZ in the neighborhood of a given point.This paper is an extension of an earlier note that defines the procedure for computingZ. Here, we first describe howZ can be obtained byupdating an explicit QR factorization with Householder transformations. The properties of this representation ofZ with respect to perturbations inA are discussed, including explicit bounds on the change inZ. We then introduceregularized Householder transformations, and show that their use implies continuity of the full matrixQ. The convergence ofZ andQ under appropriate assumptions is then proved. Finally, we indicate why the chosen form ofZ is convenient in certain methods for nonlinearly constrained optimization.The research of the Stanford authors was supported by the U.S. Department of Energy Contract DE-AM03-76SF00326, PA No. DE-AT03-76ER72018; the National Science Foundation Grants MCS-7926009 and ECS-8312142; the Office of Naval Research Contract N00014-75-C-0267; and the U.S. Army Research Office Contract DAAG29-84-K-0156.The research of G.W. Stewart was supported by the Air Force Office of Scientific Research Contract AFOSR-82-0078.     

On stationary and cycle-stationary sequences

Consider random sequences in two-sided time split into cycles by the visits to a recurrent set of states A. The two Palm dualities between stationary sequences Z and sequences with stationary cycles Z ^° are constructed using the change-of-measure and change-of-origin method. The first duality has the standard interpretation that Z ^° behaves like Z conditioned on Z ₀ ∈ A. The second duality has the less known but no less important interpretation that Z ^° behaves like Z seen from a typical visit to A. Received: 6 September 2002     

Automorphism groups of finite distributive lattices with a given sublattice of fixed points

For a latticeL, aL is a fixed point ofL if and only iff(a)=a for every automorphismf ofL. Let Aut (L) andS (L) denote the group of automorphisms ofL and the sublattice of fixed points ofL, respectively. It is shown that ifG is a finite group other thanZ ₂ ² ,Z ₂ ³ ,Z ₂ ⁴ ,Z ₃ ² or the quaternion group of order 8 andL is a finite automorphism free distributive lattice that is not near Boolean then there is a finite distributive latticeL such that Aut (L)G andS(L)L.The support of the National Research Council of Canada is gratefully acknowledged.     

Combinatorial criteria for uniqueness of Gibbs measures

We generalize previously known conditions for uniqueness of the Gibbs measure in statistical physics models by presenting conditions of any finite size for models on any underlying graph. We give two dual conditions, one requiring that the total influence on a site is small, and the other that the total influence of a site is small. Our proofs are combinatorial in nature and use tools from the analysis of discrete Markov chains, in particular the path coupling method. The implications of our conditions for the mixing time of natural Markov chains associated with the models are discussed as well. We also present some examples of models for which the conditions hold. © 2005 Wiley Periodicals, Inc. Random Struct. Alg., 2005     

Markov random fields,Markov cocycles and the 3-colored chessboard

The well-known Hammersley–Clifford Theorem states (under certain conditions) that any Markov random field is a Gibbs state for a nearest neighbor interaction. In this paper we study Markov random fields for which the proof of the Hammersley–Clifford Theorem does not apply. Following Petersen and Schmidt we utilize the formalism of cocycles for the homoclinic equivalence relation and introduce “Markov cocycles”, reparametrizations of Markov specifications. The main part of this paper exploits this to deduce the conclusion of the Hammersley–Clifford Theorem for a family of Markov random fields which are outside the theorem’s purview where the underlying graph is Z_d. This family includes all Markov random fields whose support is the d-dimensional “3-colored chessboard”. On the other extreme, we construct a family of shift-invariant Markov random fields which are not given by any finite range shift-invariant interaction.     

A property of strictly singular one-to-one operators

We prove that ifT is a strictly singular one-to-one operator defined on an infinite dimensional Banach spaceX, then for every infinite dimensional subspaceY ofX there exists an infinite dimensional subspaceZ ofX such thatZ∩Y is infinite dimensional,Z contains orbits ofT of every finite length and the restriction ofT toZ is a compact operator. The research was partially supported by NSF.     

Sur Les Groupes D’Homotopie Des Espaces Dont La Cohomologie Modulo 2 Est Nilpotente

The main object of this note is to prove the following generalisation of a theorem of Serre. A simply connected space of finite type whose mod. 2 cohomology is nilpotent (and non-trivial) has infinitely many homotopy groups which are not of odd torsion. Incidentally we show that for every fibrationF( _→ ^ί )E ( _→ ^p )B, satisfying certain mild conditions, the following holds. If a classx in the mod. 2 cohomology ofE belongs to the kernel ofi*, then some power ofx belongs to the ideal generated by the image underp* of the mod. 2 reduced cohomology ofB.      

Construction of Gibbs measures for 1-dimensional continuum fields

We study 1-dimensional continuum fields of Ginzburg-Landau type under the presence of an external and a long-range pair interaction potentials. The corresponding Gibbs states are formulated as Gibbs measures relative to Brownian motion [17]. In this context we prove the existence of Gibbs measures for a wide class of potentials including a singular external potential as hard-wall ones, as well as a non-convex interaction. Our basic methods are: (i) to derive moment estimates via integration by parts; and (ii) in its finite-volume construction, to represent the hard-wall Gibbs measure on C(ℝ;ℝ⁺) in terms of a certain rotationally invariant Gibbs measure on C(ℝ;ℝ³).