Unions and Intersections of Families of Lp-Spaces

Let L^Ψ and E^Ψ be the ORLICZ space and the space of finite elements respectively, on a measure space (Ω, Σ, μ), and let T ? (0, ∞). It is proved that if inf {p: p ? T} ? T or sup {p: p ? T} ? T and μ is an infinite atomless measure, then there is no ORLICZ function Ψ such that: \documentclass{article}\pagestyle{empty}\begin{document}$ L^\varphi = Lin\mathop { \cup L^p }\limits_{p\varepsilon T} $\end{document} or \documentclass{article}\pagestyle{empty}\begin{document}$ E^\varphi = Lin\mathop { \cup L^p }\limits_{p\varepsilon T} $\end{document} and moreover, there is no ORLICZ function Ψ such that: \documentclass{article}\pagestyle{empty}\begin{document}$ L^\varphi = Lin\mathop { \cap L^p }\limits_{p\varepsilon T} $\end{document} or \documentclass{article}\pagestyle{empty}\begin{document}$ E^\varphi = Lin\mathop { \cap L^p }\limits_{p\varepsilon T} $\end{document}.     

The PENROSE Transform for Bundles Non-Trivial on the General Line

Let L₀ be a fixed projective line in CP ³ and let M ? C ⁴ be the complexified MINKOWSKI space interpreted as the manifold of all projective lines L ? CP ³ with L ∩ L ₀ ?? Ø. Let D ? M , D ′ ? CP ³/ L ₀ be open sets such that \documentclass{article}\pagestyle{empty}\begin{document}$ D' = \mathop \cup \limits_{L \in D} $\end{document}. Under certain topological conditions on D, R. S. WARD'S PENROSE transform sets up an 1–1 correspondence between holomorphic vector bundles over D ′ trivial over each L ? D and holomorphic connections with anti-self-dual curvature over D (anti-self-dual YANG-MILLIS fields). In the present paper WARD'S construction is generalized to holomorphic vector bundles E over D′ satisfying the condition that \documentclass{article}\pagestyle{empty}\begin{document}$ E|_L \cong E|_{\tilde L} $\end{document} for all \documentclass{article}\pagestyle{empty}\begin{document}$ L,\tilde L \in D $\end{document}.     

Local and Global LIPSCHITZian Mappings on Ordered Metric Spaces

We consider the set ?? of nonhomogeneous Markov fields on T = N or T = Z with finite state spaces E_n, n ? T , with fixed local characteristics. For T = N we show that ?? has at most \documentclass{article}\pagestyle{empty}\begin{document}$ \mathop N\nolimits_\infty = \mathop {\lim \inf}\limits_{n \to \infty} \left| {\mathop E\nolimits_n} \right| $\end{document} phases. If T = Z , ?? has at most N_-∞ · N_∞; phases, where \documentclass{article}\pagestyle{empty}\begin{document}$ \mathop N\nolimits_{-\infty} = \mathop {\lim \inf}\limits_{n \to -\infty} \left| {\mathop E\nolimits_n} \right| $\end{document}. We give examples, that for T = N for any number k, 1 ≦ k ≦ N_∞, there are local characteristics with k phases, whereas for T = Z every number l · k, 1 ≦ l ≦ N_-∞, 1 ≦ k ≦ N_∞ occurs. We describe the inner structure of ??, the behaviour at infinity and the connection between the one-sided and the two-sided tail-fields. Simple examples of Markov fields which are no Markov processes are given.     

What is the smallest number of dicycles in a dicycle decomposition of an eulerian digraph?

A digraph D with n vertices is said to be decomposable into a set S of dicycles if every arc of D is contained in exactly one member of S. Counterexamples are given to the following conjectures which are generalizations of three well-known conjectures by G. Hajós, P. Erd?s, and P.J. Kelly: (1) [B. Jackson] Every eulerian-oriented graph is decomposable into at most \documentclass{article}\pagestyle{empty}\begin{document}$ \frac{n}{2} $\end{document} dicycles. (2) [W. Bienia & H. Meyniel] Every eulerian digraph is decomposable into at most n dicycles. Certain observations lead us to make three other conjectures: (a) Every eulerian-oriented graph is decomposable into at most \documentclass{article}\pagestyle{empty}\begin{document}$ \frac{{2n}}{3} $\end{document} dicycles. (b) Every symmetric digraph with n > 1 is decomposable into at most 2n – 3 dicycles. (c) Every eulerian digraph with n > 1 is decomposable into at most \documentclass{article}\pagestyle{empty}\begin{document}$ \frac{{8n}}{3} $\end{document} – 3 dicycles.     

Projective limits of finite decomposition systems

Special finite topological decomposition systems were used to get compactifications of topological spaces in [6]. In this paper the notion of finite decomposition systems is applied for topological measure spaces. We get two canonical topological measure spaces X_∞ and X_∞^d being projective limits of (discrete) finite decomposition systems for each topological measure space X = (X, O, A, P) and each net (A_α) α ? I of upward filtering finite σ-algebras in A. X_∞ is a compact topological measure space and the idea to construct is the same as used in [6]. The compactifications of [6] are cases of some special X_∞. Further on we obtain that each measurable set of the remainder of X_∞ has measure zero with respect to the limit measure P_∞ (Theorem 1). X_∞^d is the STONE representation space X(\documentclass{article}\pagestyle{empty}\begin{document}$ \mathop \cup \limits_{\alpha \in I} A\alpha $\end{document}) of \documentclass{article}\pagestyle{empty}\begin{document}$ \mathop \cup \limits_{\alpha \in I} A\alpha $\end{document} A_α, hence a Boolean measure space with regular Borel measure. Some measure theoretical and topological relations between X, X(\documentclass{article}\pagestyle{empty}\begin{document}$ \mathop \cup \limits_{\alpha \in I} A\alpha $\end{document}) and x(A) where x(A) is the Stone representation space of A, are given in Theorem 2. and 4. As a corollary from Theorem 2. we get a measure theoretical-topological version to the Theorem of Alexandroff Hausdorff for compact T₂ measure spaces x with regular Borel measure (Theorem 3.).     

Fourth-order accurate one-step integration methods with large imaginary stability limits

One-step integration methods of fourth-order accuracy using an odd number of function evaluations K, to solve dy/dt = A · y, are proposed. These methods have an imaginary stability limit \documentclass{article}\pagestyle{empty}\begin{document}$ S_{1\;} = \sqrt {(K - 1)^2 - 4} $\end{document}. In the case K = 5 the Kutta-Merson method results.     

Factorization theorems for some scales of operator ideals

Let ??_p and ?_p denote the operator ideals generated by the approximation numbers and entropy numbers, respectively. If 0<p,q< ∞ and \documentclass{article}\pagestyle{empty}\begin{document}$ \frac{1}{p} + \frac{1}{q} = \frac{1}{r}, $\end{document} then ??_p °_q=??_r and ?_p °_q=?_r.     

Strong uniform consistency of nonparametric regression function estimates

Let (X, Y) be a ^dx-valued random vector and let r(t)=E(Y/X=t) be the regression function of Y on X that has to be estimated from a sample (X _i, Y_i), i=1,..., n. We establish conditions ensuring that an estimate of the form

A Note on Bounds for the Supremum Metric for Discrete Random Variables

For positive or negative integer-valued random variables X and Y with finite second moments the inequality sup \documentclass{article}\pagestyle{empty}\begin{document}$ \mathop {\sup }\limits_n |\Pr \{ X \le n\} - \Pr \{ Y \le n\} |\, \le \,|EX - EY| + \frac{1}{2}(EX(EX - 1) + (EY(Y - 1)) $\end{document} is established by elementary manipulation, and shown to be tight. Use of generating functions and an inversion formula yields the larger bound with 1/2 replaced by 2/π.     

Almost sure invariance principles when EX 1 2 =∞

Summary Let {Y _i} be iid with EY ₁=0, EY ₁ ² =1. Let {X_i} be iid normal mean zero, variance one random variables. According to Strassen's first almost sure invariance principle {X _i} and {Y _i} can be reconstructed on a new probability space without changing the distribution of each sequence such that a.s., thus improving on the trivial bound obtainable from the law of the iterated logarithm: a.s. In this work we establish analogous improvements for symmetric {Y _i} in the domain of normal attraction to a symmetric stable law with index 0<<2. (We make this assumption of symmetry in order to avoid messy details concerning centering constants.) Let {X _i} be iid symmetric stable random variables with index 0<<2. Then, for example, hypotheses are stated which imply for a given satisfying 2> that a.s., thus improving on the trivial bound: a.s., >0.This research was supported in part by a National Science Foundation grant, USA     

A generalized jacobi theta function

The delta function initial condition solution v^*(x,t;y) at x = y ≥ 0 of the generalized Feller equation is used to define a generalized Jacobi Theta function \documentclass{article}\pagestyle{empty}\begin{document}$ \Theta (x,t) = \upsilon *(x,t;0) + 2\sum\limits_{n = 1}^\infty {v*(x,t;y_n)} $\end{document} for a sufficiently rapidly increasing and unbounded positive sequence {y_y}. It is shown that Θ(x,t) is analytic in each variable in certain regions of the complex x and t planes and that it is a solution of the generalized Feller equation. For those parameters for which this equation reduces to the heat equation, Θ(x,t) reduces to the third Jacobi Theta function.     

STRONG UNIFORM CONSISTENCY FOR DENSITY ESTIMATOR FROM RANDOMLY CENSORED DATA

Let X_1,…,X_n be a sequence of independent identically distributed random variableswith distribution function F and density function f.The X_are censored on the right byY_i,where the Y_i are i.i.d.r.v.s with distribution function G and also independent of theX_i.One only observesLet S=1-F be survival function and S be the Kaplan-Meier estimator,i.e.,where Z_are the order statistics of Z_i and δ_((i))are the corresponping censoring indicatorfunctions.Define the density estimator of X_i by where =1-and h_n(>0)↓0.     

A Weak Convergence Theorem for Functionals of Sums of Martingale Differences

For each n, let (S_nk), 1 ≦ k ≦ k_n, be a mean zero square — integrable martingale adapted to increasing s?-fields (b_nk), 0 ≦ k ≦ k_n, and let (b_nk), 0 ≦ k ≦ k_n, be a system of random variables such that b_n0 = 0 < b_n1 <…< b = 1 and such that b_nk is b_{n, k?1} measurable for each k. We present sufficient conditions under which \documentclass{article}\pagestyle{empty}\begin{document}$ \sum\limits_{i = 0}^{k_n - 1} {f_n (b_{ni,\;} S_{ni})\;(S_{n,i + 1} \; - \;S_{ni})\; \to \int\limits_0^1 {f(t,\;W(t))\;d{\rm W(t)}} } $\end{document} as n → ∞, where {W(t) : 0 ≦ t ≦ 1} is a standard WIENER process.     

Empirical Processes with a Bounded Ψ 1 Diameter

We study the empirical process ${{\rm sup}_{f \in F}|N^{-1}\sum_{i=1}^{N}\,f^{2}(X_i)-\mathbb{E}f^{2}|}We study the empirical process sup_{f ? F}|N^-1?_i=1^N f²(X_i)-\mathbbEf²|{{\rm sup}_{f \in F}|N^{-1}\sum_{i=1}^{N}\,f^{2}(X_i)-\mathbb{E}f^{2}|}, where F is a class of mean-zero functions on a probability space (Ω, μ), and (X_i)_{i = 1}^N{(X_{i})_{i =1}^N} are selected independently according to μ.     

Numerical implementation of the Sinc-Galerkin method for second-order hyperbolic equations

A fully Galerkin method in both space and time is developed for the second-order, linear hyperbolic problem. Sinc basis functions are used and error bounds are given which show the exponential convergence rate of the method. The matrices necessary for the formulation of the discrete system are easily assembled. They require no numerical integrations (merely point evaluations) to be filled. The discrete problem is formulated in two different ways and solution techniques for each are described. Consideration of the two formulations is motivated by the computational architecture available. Each has advantages for the appropriate hardware. Numerical results reported show that if 2N + 1 basis functions are used then the exponential convergence rate \documentclass{article}\pagestyle{empty}\begin{document}$ 0\left[{\exp \left({- \kappa \sqrt N} \right)} \right] $\end{document}, κ > 0, is attained for both analytic and singular problems.     

Strong convergence in nonparametric estimation of regression functions

The nonparametric regression problem has the objective of estimating conditional expectation. Consider the model $$Y = R(X) + Z$$ , where the random variableZ has mean zero and is independent ofX. The regression functionR(x) is the conditional expectation ofY givenX = x. For an estimator of the form $$R_n (x) = \sum\limits_{i = 1}^n {Y_i K{{\left[ {{{\left( {x - X_i } \right)} \mathord{\left/ {\vphantom {{\left( {x - X_i } \right)} {c_n }}} \right. \kern-\nulldelimiterspace} {c_n }}} \right]} \mathord{\left/ {\vphantom {{\left[ {{{\left( {x - X_i } \right)} \mathord{\left/ {\vphantom {{\left( {x - X_i } \right)} {c_n }}} \right. \kern-\nulldelimiterspace} {c_n }}} \right]} {\sum\limits_{i = 1}^n {K\left[ {{{\left( {x - X_i } \right)} \mathord{\left/ {\vphantom {{\left( {x - X_i } \right)} {c_n }}} \right. \kern-\nulldelimiterspace} {c_n }}} \right]} }}} \right. \kern-\nulldelimiterspace} {\sum\limits_{i = 1}^n {K\left[ {{{\left( {x - X_i } \right)} \mathord{\left/ {\vphantom {{\left( {x - X_i } \right)} {c_n }}} \right. \kern-\nulldelimiterspace} {c_n }}} \right]} }}} $$ , we obtain the rate of strong uniform convergence $$\mathop {\sup }\limits_{x\varepsilon C} \left| {R_n (x) - R(x)} \right|\mathop {w \cdot p \cdot 1}\limits_ = o({{n^{{1 \mathord{\left/ {\vphantom {1 {(2 + d)}}} \right. \kern-\nulldelimiterspace} {(2 + d)}}} } \mathord{\left/ {\vphantom {{n^{{1 \mathord{\left/ {\vphantom {1 {(2 + d)}}} \right. \kern-\nulldelimiterspace} {(2 + d)}}} } {\beta _n \log n}}} \right. \kern-\nulldelimiterspace} {\beta _n \log n}}),\beta _n \to \infty $$ . HereX is ad-dimensional variable andC is a suitable subset ofR ^d .     

Cutting a graph into two dissimilar halves

Given a graph G and a subset S of the vertex set of G, the discrepancy of S is defined as the difference between the actual and expected numbers of the edges in the subgraph induced on S. We show that for every graph with n vertices and e edges, n < e < n(n ? 1)/4, there is an n/2-element subset with the discrepancy of the order of magnitude of \documentclass{article}\pagestyle{empty}\begin{document}$\sqrt {ne}$\end{document} For graphs with fewer than n edges, we calculate the asymptotics for the maximum guaranteed discrepancy of an n/2-element subset. We also introduce a new notion called “bipartite discrepancy” and discuss related results and open problems.     

Laplace approximations for sums of independent random vectors

Summary LetX _i,iN, be i.i.d.B-valued random variables whereB is a real separable Banach space, and a mappingB R. Under some conditions an asymptotic evaluation of is possible, up to a factor (1+o(1)). This also leads to a limit theorem for the appropriately normalized sums under the law transformed by the density exp .     

A note on the normal approximation error for randomly weighted self-normalized sums

Let X = {X _n } _n≥1 and Y = {Y _n } _n≥1 be two independent random sequences. We obtain rates of convergence to the normal law of randomly weighted self-normalized sums $$\psi _n \left( {X,Y} \right) = \sum\nolimits_{i = 1}^n {{{X_i Y_i } \mathord{\left/ {\vphantom {{X_i Y_i } {V_n , V_n }}} \right. \kern-\nulldelimiterspace} {V_n , V_n }}} = \sqrt {Y_1^2 + \cdots + Y_n^2 } .$$ . These rates are seen to hold for the convergence of a number of important statistics, such as for instance Student’s t-statistic or the empirical correlation coefficient.     

The Extended Probability Generating Functional,with Application to Mixing Properties of Cluster Point Processes

An extended probability generating functional (p. g. fl.) \documentclass{article}\pagestyle{empty}\begin{document}$ \bar{G}[h]\; = \;E\;\left({\exp \;\int\limits_x {\log } \;h(x)\; \times \;N(dx)} \right) $\end{document} is well-defined for any point process N on the complete separable metric space χ over the space V?₀ of measurable functions h: χ → (0, 1) such that inf x cH h(x) > 0. The distribution of N is determined uniquely by the p.g.fl. G[h] ≡ ?[h] over the smaller space V₀ of functions h ε V?₀ for which 1 — h has bounded support. Continuity results for ?[·] involving pointwise convergent sequences {h_n} V₀ or V?₀ or V? ≡ {measurable h: χ → [0, 1]} or V = {h ε V?: 1 — h has bounded support} are reviewed, and used in furnishing a complete p. g. fl. proof of the mixing property of certain stationary cluster processes.