Multiscale convergence and reiterated homogenization of parabolic problems

Reiterated homogenization is studied for divergence structure parabolic problems of the form u /t–div (a(x,x/,x/²,t,t/ ^k)u )=f. It is shown that under standard assumptions on the function a(x, y ₁,y ₂,t,) the sequence {u } of solutions converges weakly in L ² (0,T; H ₀ ¹ ()) to the solution u of the homogenized problem u/t– div(b(x,t)u)=f.This revised version was published online in April 2005 with a corrected missing date string.     

Rational Chebyshev approximation on the unit disk

Summary In a recent paper we showed that error curves in polynomial Chebyshev approximation of analytic functions on the unit disk tend to approximate perfect circles about the origin [23]. Making use of a theorem of Carathéodory and Fejér, we derived in the process a method for calculating near-best approximations rapidly by finding the principal singular value and corresponding singular vector of a complex Hankel matrix. This paper extends these developments to the problem of Chebyshev approximation by rational functions, where non-principal singular values and vectors of the same matrix turn out to be required. The theory is based on certain extensions of the Carathéodory-Fejér result which are also currently finding application in the fields of digital signal processing and linear systems theory.It is shown among other things that iff(z) is approximated by a rational function of type (m, n) for >0, then under weak assumptions the corresponding error curves deviate from perfect circles of winding numberm+n+1 by a relative magnitudeO( ^{m + n + 2} as 0. The CF approximation that our method computes approximates the true best approximation to the same high relative order. A numerical procedure for computing such approximations is described and shown to give results that confirm the asymptotic theory. Approximation ofe ^z on the unit disk is taken as a central computational example.     

Construction of ɛ-nets

LetS be a set ofn points in the plane and let be a real number, 0<<1. We give a deterministic algorithm, which in timeO(n ^–2 log(1/)+ ^–8) (resp.O(n ^–2 log(1/)+ ^–10) constructs an-netNS of sizeO((1/) (log(1/))²) for intersections ofS with double wedges (resp. triangles); this means that any double wedge (resp. triangle) containing more thatn points ofS contains a point ofN. This givesO(n logn) deterministic preprocessing for the simplex range-counting algorithm of Haussler and Welzl [HW] (in the plane).We also prove that given a setL ofn lines in the plane, we can cut the plane intoO( ^–2) triangles in such a way that no triangle is intersected by more thann lines ofL. We give a deterministic algorithm for this with running timeO(n ^–2 log(1/)). This has numerous applications in various computational geometry problems.     

Approximating the ball by a minkowski sum of segments with equal length

It is proved that forn 2 the Euclidean ballB _n can be approximated up to (in the Hausdorff distance) by a zonotope havingN summands of equal length withN c(n)( ^–2|log|)^{(n–1)/(n+2)}.Research supported in part by the U.S.-Israeli Binational Science Foundation. [Please see the Editors' note on the first page of the preceding paper.]     

Die ɛ-Störung linearer und quadratischer Optimierungsaufgaben und ihre Anwendung auf das Hildreth-Verfahren

Zusammenfassung Durch eine -Störung in der Diagonalen der quadratischen Form kann man eine lineare oder quadratisch semidefinite Optimierungsaufgabe zu einer streng definiten quadratischen Aufgabe machen, so daß Lösungsverfahren, die die Formmatrix als nichtsingulär voraussetzen müssen, anwendbar werden. Bekanntlich konvergiert die Lösungx der -gestörten Aufgabe für 0 gegen den Lösungsvektorx ^m von minimalem Betrag der ursprünglichen Aufgabe. Wir zeigen darüber hinaus, daß im linearen Fall immer und im eigentlich quadratischen in gewissen Fällen schon für 0<<* die beiden Lösungenx undx ^m übereinstimmen. Im linearen Fall ist die obere Grenze * durch die Lösung eines linearen Ungleichungssystems gegeben.Im zweiten Abschnitt wenden wir dasHildreth-Verfahren mittels der -Störung auf lineare und quadratisch semidefinite Aufgaben an, diskutieren Konvergenz- und Genauigkeitsfragen und kommen zu dem Schluß, daß man in der Praxis sowohl bei Rechnung von Hand als auch bei maschineller Rechnung zu befriedigenden Ergebnissen kommt.

---

Summary Linear and quadratic semidefinite programming problems may be transformed into strongly definite quadratic problems by means of an -perturbation of the quadratic form so that procedures which presuppose the matrix of the form to be nonsingular, may be applied. As is well known, the solutionx of the -perturbated problem converges to the solutionx ^m of minimal length of the original problem as 0. We show that always in the linear case and in the quadratic case under certain circumstances, both solutionsx andx ^m are equal if 0 <<*. In the linear case, the upper limit * is given by the solution of a system of linear inequalities.In the second part of this paper we apply the method ofHildreth to linear and quadratic semidefinite programming problems by the -perturbation. We discuss questions of convergence and exactness, and conclude that in practice calculation by hand as well as by computer leads to satisfying results.

---

---

Der Verfasser ist Herrn Prof. Dr.W. Vogel, Bonn, für einen Hinweis zu Dank verpflichtet.

---

Vorgel. v.:H. P. Künzi     

A correction theorem and the dyadic space H1, ∞

It is shown that for every l-function f and for every , >0, there exists a function g such that mes {t=g} <, while the partial sums of the Fourier and Fourier-Walsh series of the function g are uniformly bounded by the number C log (^–1)f. In the proof we make use of the characterization of the dyadic space H^1, in terms of atomic decompositions (it is, apparently, new).Translated from Zapiski Nauchnykh Seminarov Leningradskogo Otdeleniya Matematicheskogo Instituta im. V. A. Steklova AN SSSR, Vol. 149, pp. 67–75, 1986.     

A result from diophantine approximation which has applications in economics

We say that a real number allows poor approximations if we can find 0<=()<1 and a sequence of integers n₁2<... such that for all rationals p/q with qn. we have |–.p/q| > Kn _j ^–l– where K is a constant depending only on .In this note we prove that the set of numbers which allow poor approximations are precisely the very well-approximable numbers.The existence of numbers with poor approximations has been used by Cheng [1] to show the existence of a dense set of economies whose cone converges to the Walras equilibrium as slowly as 0(n^–1/2–) after n replications.     

Errors in approximate solutions of Cauchy's problem for a first-order quasilinear equation

The proximity is investigated of the solution of Cauchy's problem for the equation u _t +((u))_x= u _xx ((u) > 0) to the solution of Cauchy's problem for the equation u_t+ ((u))_x= 0, when the solution of the latter problem has a finite number of lines of discontinuity in the strip 0 t T. It is proved that, everywhere outside a fixed neighborhood of the lines of discontinuity, we have |u–u| C, where the constant C is independent of. Similar inequalities are derived for the first derivatives of u–u.Translated from Matematicheskie Zametki, Vol. 8, No. 3, pp. 309–320, September, 1970.In conclusion we express our gratitude to L. A. Chudov for his valuable advice concerning this work.     

Functions with a finite dirichlet integral in a domain with cusp points on the boundary

Let be a domain in ⁿ, n >2, the boundary of which has a cusp point, pointing inside or outside the domain. The purpose of the paper is to characterize the traces on of the elements of the space H¹() of functions with a finite Dirichlet integral. As a consequence one establishes the existence of a linear continuous extension operator H¹ () H¹(ⁿ) under the presence of an interior cusp point on . Theorems on domains with cusps are proved with the aid of results on cylindrical domains. In the space of functions with a finite Dirichlet integral in the exterior or the interior of the cylinder one introduces the norm, depending on a small parameter and generating a norm of the trace on as an element of the quotient space. The latter is placed in correspondence with an explicitly described norm of functions on the boundary, uniformly equivalent relative to . One constructs an operator of extension of functions from the exterior of the cylinder to Rⁿ, preserving H¹, whose norm is uniformly bounded relative to . For the optimal operator of extension from the inside of the cylinder one finds the asymptotic behavior of the norm as 0. From these results there follow similar theorems on functions with a finite Dirichlet integral inside and outside a thin closed tube (of width ).Translated from Zapiski Nauchnykh Seminarov Leningradskogo Otdeleniya Matematicheskogo Instituta im. V. A. Steklova AN SSSR, Vol. 126, pp. 117–137, 1983.     

A remark on the strong law of large numbers

The strong law of large numbers for independent and identically distributed random variablesX _i ,i=1, 2, 3,... with finite expectationE|X ₁| can be stated as, for any >0, the number of integersn such that \varepsilon $$ " align="middle" border="0"> ,N is finite a. s. It is known thatEN < iffEX ₁ ² < and that ² EN var X₁ as 0, ifE X ₁ ² <. Here we consider the asymptotic behaviour ofEN (n) asn, whereN (n) is the number of integerskn such that \varepsilon $$ " align="middle" border="0"> andE N ₁ ² =.