Géométrie des espaces de Banach ayant des approximations de l'identité contractantes

Let a, c ≥ 0 and let B be a compact set of scalars. We introduce property M^* (a, B, c) of Banach spaces X which is a geometric property of Banach spaces generalizing property (M^*) due to Kalton. Using M^*(a, B, c) with max ¦B¦ + c > 1, we characterize intrinsically a large class of shrinking approximations of the identity, including those related to M-, u-, and h-ideals of compact operators. We also show that the existence of these approximations of the identity is separably determined. As an application, we study ideals of compact and approximable operators. In particular, this provides an alternative unified and easier approach to the theories of M-, u-, and h-ideals of compact operators.     

An accelerated Newton method for equations with semismooth Jacobians and nonlinear complementarity problems

We discuss local convergence of Newton’s method to a singular solution x ^* of the nonlinear equations F(x) =  0, for $F:{\mathbb{R}}^n \rightarrow {\mathbb{R}}^n$ . It is shown that an existing proof of Griewank, concerning linear convergence to a singular solution x ^* from a starlike domain around x ^* for F twice Lipschitz continuously differentiable and x ^* satisfying a particular regularity condition, can be adapted to the case in which F′ is only strongly semismooth at the solution. Further, Newton’s method can be accelerated to produce fast linear convergence to a singular solution by overrelaxing every second Newton step. These results are applied to a nonlinear-equations reformulation of the nonlinear complementarity problem (NCP) whose derivative is strongly semismooth when the function f arising in the NCP is sufficiently smooth. Conditions on f are derived that ensure that the appropriate regularity conditions are satisfied for the nonlinear-equations reformulation of the NCP at x ^*.     

Algebras of incidence structures: representations of regular double p-algebras

An incidence structure is a standard geometric object consisting of a set of points, a set of lines and an incidence relation specifying which points lie on which lines. This concept generalises, for example, both graphs and projective planes. We prove that the lattice of point-preserving substructures of an incidence structure naturally forms a regular double p-algebra. A double p-algebra A is regular if for all \({x, y \,\in \, A}\), we have that x⁺ =  y⁺ and x* =  y* together imply x = y.     

Isoliertheit und Stabilität von Flächen konstanter mittlerer Krümmung

In the Sobolev space H^m(B,?³), B the open unit disc in ?², we consider the set M_n of all conformally parametrized surfaces of constant mean curvature H with exactly n simple interior branch points (and no others). We denote by M*_n the set of all xεM_n with the following properties:

1. in every branch point the geometrical condition K_G¦x_Z¦≡O holds (K_G is the Gauss curvature and x_z is the complex gradient of the surface x).
2. the corresponding boundary value problem Δh+×_z{²(2H²-K_G)h=O,^hδB=O, is uniquely solvable.

We prove then, that the manifold M*=UM*_n is open and dense in the set of all surfaces of constant mean curvature H and that all x εM*_n are isolated and stable solutions of the Plateau problem corresponding to their boundary curves. In addition, the submanifold M*_n contains exactly all surfaces x for which the space of Jacobi fields is transversal (with exception of the 3-dimensional space of conformai directions) to the tangent space T_xM_n.     

Cantor-Lebesgue theorem for double trignometric series

Let ∥·∥ be a norm in R² and let γ be the unit sphere induced by this norm. We call a segment joining points x,y ε R² rational if (x₁ ? y₁)/(x₂ ? y₂) or (x₂ ? y₂)/(x₁ ? y₁) is a rational number. Let γ be a convex curve containing no rational segments. Satisfaction of the condition $$T_\nu (x) = \sum\nolimits_{\parallel n\parallel = \nu } {c_n e^{2\pi i(n_1 x_1 + n_2 x_2 )} } \to 0(\nu \to \infty )$$ in measure on the set e? [- 1/2,1/2)×[- 1/2, 1/2) =T² of positive planar measure implies ∥T _v ∥L₄ (T²) → 0(v → ∞). if, however, γ contains a rational segment, then there exist a sequence of polynomials {T _v } and a set E ? T², ¦E¦ > 0, such that T _v (x) → 0(v → ∞) on E; however, ¦c_n¦ ? 0 for ∥n∥ → ∞.     

Некоторые точные инт егральные оценки про изводных рациональных и алгеб раических функций. Пр иложения

Exact estimates are obtained for integrals of absolute values of derivatives and gradients, for integral moduli of continuity and for major variations of piecewise algebraic functions (in particular, for polynomials, rational functions, splines, etc.). These results are applied to the problems of approximation theory and to the estimates of Laurent and Fourier coefficients. Typical results:

1. IfE is a measurable subset of the circle or of a line in thez-plane andR(z) is a rational function of degree ≦n, ¦R(z)¦≦ (z∈E), then ∝_E ¦R′(z)¦dz¦≦ 2πn; the latter estimate is exact forn=0, 1, ... and everyE with positive measure;
2. Iff(x ₁,x ₂, ...,x _m) is a real valued piecewise algebraic function of order (n, k) on the unit ballD?R ^m (in particular, a real valued rational function of order ≦n), and ¦f¦≦1 onD, then ∝_D¦gradf¦dx≦2π ^m/2n/Π(m/2); herem≧1, n≧0, 1≦k<∞;
3. LetE=Π={z∶¦z¦=1}, and letc _m(R) be the mth Laurent coefficient ofR onΠ,C _m(n)=sup{¦cm(R)¦}, where sup is taken over allR from 1), then 1/7 min {n/¦m¦, 1} ≦C _m(n) ≦ min {n/¦m¦, 1}.

     

Stability of the solutions of one-sided variational problems with applications to the theory of plasticity

In the paper one investigates the one-sided variational problem of the form $$/u - u_0 / = min, u \in M,$$ where ¦·¦ is the norm in some Hilbert space H₀, ? is a nonempty convex set, closed in the metric of H₀, and u₀ is a given element of this space. The fundamental results are: 1) The solution of the problem (1) is stable relative to small perturbations of the data of this problem: the element u₀, the norm ¦·¦, and the set ? the concept of small perturbation is precisely formulated. 2) Assume that the set ? is defined by the formula where g is an element of some Hilbert space containing the space H₀, ||| · ||| is some seminorm, and a is a positive constant. Let H⁽ⁿ⁾ be a subspace of the space H₀ on all of whose elements the seminorm ||| · ||| is finite. If u_n is the approximate solution of the problem (1), obtained as the solution of the problem ¦u?u₀¦=min, u_n ∈ M ∩ H⁽ⁿ⁾, then ¦u_*?u_n¦=0 (e_n(u_*)), where u_* is the exact solution of the problem (1), while e_n(u_*) is the best approximation of u_* by the elements of the subspace H⁽ⁿ⁾. The given results are used in a series of problems regarding the elastoplastic state according to the Saint-Venant-Mises theory; one assume that for these problems the Haar-Karman variational principle holds.     

Regularity and integrability of rotating shallow-water equations

We consider classical shallow-water equations for a rapidly rotating fluid layer. The Poincaré/Kelvin linear propagator describes fast oscillating waves for the linearized system. We show that solutions of the full nonlinear shallow-water equations can be decomposed as U(t,x₁,x₂) + Ũ(t,x₁,x₂) + W’(t,x₁,x₂) + r, where Ũ is a solution of the quasigeostrophic (QG) equation. Here r is a remainder, which is uniformly estimated from above by a majorant of order 1/f₀. The vector field W’(t,x₁,x₂) describes the rapidly oscillating ageostrophic (AG) component. This component is exactly solved in terms of Poincaré/Kelvin waves with phase shifts explicitly determined from the nonlinear quasigeostrophic equations. The mathematically rigorous control of the error r, based on estimates of small divisors, is used to prove the existence, on a long time interval T^*, of regular solutions to classical shallow-water equations with general initial data (T^* → +∞, as 1/f₀ → 0).     

On the multi-dimensional renewal theorem

Let X be a random variable on Rⁿ, n ? 2, having a density. Assume X has a finite exponential moment and non-zero mean vector, μ. Let ν be the corresponding renewal measure, and Q a cube. We obtain an asymptotic formula for ν(x + Q) as x → ∞ which is uniform in a small cone about the mean vector. This formula depends on moments of arbitrarily high order but depends only on the first and second moments of X in a region $\text{x · μ > ¦x¦¦μ¦(1 ? o(¦x¦}{}^{\mathrm{<mtext>?2</mtext><mtext>3</mtext>}}\text{))}$.     

METRIC ENTROPY OF HOMEOMORPHISM ON NON-COMPACT METRIC SPACE

Let T : X → X be a uniformly continuous homeomorphism on a non-compact metric space (X, d). Denote by X* = X ∪ {x*} the one point compactification of X and T * : X* → X* the homeomorphism on X* satisfying T *|X = T and T *x* = x*. We show that their topological entropies satisfy hd(T, X) ≥ h(T *, X*) if X is locally compact. We also give a note on Katok’s measure theoretic entropy on a compact metric space.     

Error-free computer solution of certain systems of linear equations

In this article we propose a procedure which generates the exact solution for the system Ax = b, where A is an integral nonsingular matrix and b is an integral vector, by improving the initial floating-point approximation to the solution. This procedure, based on an easily programmed method proposed by Aberth [1], first computes the approximate floating-point solution x^* by using an available linear equation solving algorithm. Then it extracts the exact solution x from x^* if the error in the approximation x^* is sufficiently small. An a posteriori upper bound for the error of x^* is derived when Gaussian Elimination with partial pivoting is used. Also, a computable upper bound for |det(A)|, which is an alternative to using Hadamard's inequality, is obtained as a byproduct of the Gaussian Elimination process.     

Bounds for the spectral abscissa of an element in a Banach algebra

For an arbitrary element x with spectrum sp(x) in a Banach algebra with identity e ≠ 0 we define the upper (lower) spectral abscissa \(\mathop {\sigma + (x)}\limits_{( - )} = \mathop {\max }\limits_{(\min )} \operatorname{Re} \lambda ,\lambda \in sp(x)\) . With the aid of the spectral radius \(\rho (x) = \mathop {\max }\limits_{\lambda \in sp(x)} \left| \lambda \right| = \mathop {\lim }\limits_{n \to + \infty } \parallel x^n {{1 - } \mathord{\left/ {\vphantom {{1 - } n}} \right. \kern-0em} n}\) we prove the following bounds: γ_?(x)?σ_?(x)?Γ_?(x)?₊(x)?σ₊(x)?γ₊(x), Γ_(±)(x)=(2δ_(±))^?1 (ρ _δ ² )_(±)?δ _(±) ² ?ρ ₀ ² )(δ_(±)≠0), γ_(±)(x)= (±)ρ_δ(±)?δ_(±), δ₊?0, δ_??0 ρ _(±) ^δ = ρ(x+eδ_(±)). We mention a case where equality is achieved, some corollaries,and discuss the sharpness of the bounds: for every ? > 0 there is a δ: ¦δ¦ ≥ρ ₀ ² /2?, such that Δ: = ¦γ_(±) x?Γ_(±) x¦?ε and conversely, if the bounds are computed for some δ ≠ 0, then △ ≤ρ ₀ ² /2 ¦δ¦. An example is considered.     

A Banach space admits a locally uniformly rotund norm if its dual is a Vašák space

LetV be a Banach space whose dualV ^* is Va?ák, that is, weakly countably determined. Then an equivalent locally uniformly rotund norm onV is constructed. According to a recent example of Mercourakis, this is a real extension of an earlier result of Godefroy Troyanski, Whitfield and Zizler, whereV ^* has been a subspace of a weakly compactly generated Banach space.     

Commuting Involution

We are given a semiprime unital ring A with * such that x*x = xx* for all elements x of A. We will show that both elements x + x* and xx* are central elements. In the case in which A is a quaternion algebra over a field F in the sense given by Albert, we show that * is unique and coincides with the canonical involution. We also provide specific constructions of quaternion division algebras A with canonical involution over a field F of one of the following types: (i) F is a function field in two variables over a ground field of unspecified characteristic; (ii) F is a function field over the Galois field GF(2ⁿ); and (iii) F is a function field over the Galois field GF(pⁿ) where p is an odd prime number and n is a natural number.     

Green's function,Jensen measures,and bounded point evaluations

A compactly supported measure μ on the complex plane C is called a Jensen measure for 0 if $\text{log}\text{¦P(0)¦ ? ∝}\text{log}\text{¦P(z)¦dμ(z)}$ for every polynomial P. H²(μ) denotes the closure of the polynomials in L²(μ). We obtain the result that if μ is not the point mass at 0, then the functions in H²(μ) are analytic on an open set which contains 0 and whose closure contains the support of μ. The primary tool used to obtain this result is a generalized Green's function for a measure, and we also derive some of its properties.     

Existence and continuity of an implicit function in the neighborhood of an abnormal point

Solutions to the equation F(x, ??) = 0 with unknown x and the parameter ?? in the neighborhood of the solution (x _*, ??_*) under the additional constraint x ?? U, where U is a closed convex set, are studied. The sufficient conditions for existence of an implicit function without prior assumption of the normalcy of point x _* are given. The obtained result is used to investigate the local solvability of controlled systems with mixed constraints.     

Weak Compactness and Variational Characterization of the Convexity

We prove that if X is a Banach space and ${f : X \rightarrow \mathbb{R} \cup \{+\infty\}}$ is a proper function such that f ? ? attains its minimum for every ? ε X ^*, then the sublevels of f are all relatively weakly compact in X. As a consequence we show that a Banach space X where there exists a function ${f : X \rightarrow \mathbb{R}}$ such that f ? ? attains its minimum for every ? ε X ^* is reflexive. We also prove that if ${f : X \rightarrow \mathbb{R} \cup \{+\infty\}}$ is a weakly lower semicontinuous function on the Banach space X and if for every continuous linear functional ? on X the set where the function f ? ? attains its minimum is convex and non-empty then f is convex.     

Everywhere divergent Φ-means of Fourier series

For a function ? ∈, L ¹( $\mathbb{T}$ ), we investigate the sequence (C, 1) of mean values Φ(¦S _k (x, ?) ? ?(x)¦), where Φ(t): [0, +∞) → [0,+∞), Φ(0) = 0, is a continuous increasing function. We prove that if Φ increases faster than exponentially, then these means can diverge everywhere. Divergence almost everywhere of such means was established earlier.     

On the sum-of-ranks winner when losers are removed

Suppose that m alternatives are linearly ranked from best to worst by each of a number of judges, and that alternative x is the unique winner on the sum-of-ranks basis. It is shown that it is possible to construct a situation (with an appropriate number of judges) such that the initial winner x will be a sum-of-ranks loser within every proper subset of the original set of alternatives that contains x and at least one other alternative, except that x is the winner in exactly one subset that contains x and one other alternative.     

Functional calculi of second-order elliptic partial differential operators with bounded measurable coefficients

Consider a second-order elliptic partial differential operatorL in divergence form with real, symmetric, bounded measurable coefficients, under Dirichlet or Neumann conditions on the boundary of a strongly Lipschitz domain Ω. Suppose that 1 <p < ∞ and μ > 0. ThenL has a bounded H_∞ functional calculus in L^p(Ω), in the sense that ¦¦f (L +cI)u¦¦_p ≤C sup_¦arλ¦<μ ^¦f¦ ¦‖u¦‖_p for some constantsc andC, and all bounded holomorphic functionsf on the sector ¦ argλ¦ < μ that contains the spectrum ofL +cI. We prove this by showing that the operatorsf(L + cI) are Calderón-Zygmund singular integral operators.