Approximation Methods in the Computer Numerically Controlled Fabrication of Optical Surfaces, Part 1: Finite Dimensional Material Removal Profile Spaces

An analysis of the process of grinding and polishing of lensesand mirrored surfaces using a computer numerically controlledmachine is presented. This analysis provides the motivationfor the study of L₁ approximation of continuous functions bynonstandard approximants. Under suitable conditions, L₁ convergenceis established for a space of material removal profiles basedon a piecewise continuous material removal rate. A linear programmingalgorithm is used to find the best discrete L₁ approximation.Numerical examples are included.     

Error Control with Polynomial Approximations

A detailed analysis is given of the accumulation of errors whichmay occur in evaluating a polynomial approximation to a givenfunction. Both backward recursion using untransformed Chebyshevexpansions and the much faster nested multiplication using thetransformed simple polynomial form are treated. Two types ofarithmetic are dealt with covering most current machines. Forthe case of polynomials with coefficients of the same sign orstrictly alternating signs, a situation which is of considerablepractical importance in polynomial approximation of mathematicalfunctions, we show that controlling relative error requiresthe ratios |y|_max/|y|_min and |y'|_max/|y|_min to be kept small.Experimental verification of these effects is given based onexpansions available in the literature or produced by the authorsfor the Bessel function I₀(x).     

Discrete l1 Approximation by Rational Functions

The problem of finding a best approximation by a rational functionto discrete data, using the l₁ norm, is considered. An algorithmis developed which is frequently convergent in a finite numberof steps, and failing this usually has a second-order convergencerate. Details are given of the application of the algorithmto a number of rational approximation problems.     

A fully discrete stabilized finite-element method for the time-dependent Navier-Stokes problem

A fully discrete stabilized finite-element method is presentedfor the two-dimensional time-dependent Navier–Stokes problem.The spatial discretization is based on a finite-element spacepair (X_h, M_h) for the approximation of the velocity and thepressure, constructed by using the Q₁ – P₀ quadrilateralelement or the P₁ – P₀ triangular element; the time discretizationis based on the Euler semi-implicit scheme. It is shown thatthe proposed fully discrete stabilized finite-element methodresults in the optimal order error bounds for the velocity andthe pressure.     

A pointwise estimate on the 1-order approximation of G{varepsilon}x0

On the basis of Avellaneda & Hua-Lin (1991, Commun. PureAppl. Math., 44 897–910), a pointwise error estimate onthe 1-order approximation of Green function G_x0 defined in R²is shown at first. Then based on this estimate and using asymptoticexpansion method, an improved approximation of G_x0 and its pointwiseerror estimate are obtained.     

A note on least-squares mixed finite elements in relation to standard and mixed finite elements

^** Email: brandts{at}science.uva.nl The least-squares mixed finite-element method for second-orderelliptic problems yields an approximation u_h V_h H₀¹() of thepotential u together with an approximation p_{h h} H(div ; )of the vector field p = – Au. Comparing u_h with the standardfinite-element approximation of u in V_h, and p_h with the mixedfinite-element approximation of p, it turns out that they arehigher-order perturbations of each other. In other words, theyare ‘superclose’. Refined a priori bounds and superconvergenceresults can now be proved. Also, the local mass conservationerror is of higher order than could be concluded from the standarda priori analysis.     

Generalized Alternating Polynomials, some Properties and Numerical Applications

In the present paper we introduce the generalized alternatingpolynomials , the coefficients of which are defined together with the parameter W_n by the linear system where T_n(x) = cos (n arc cos x), is a set of n+2 distinct points in the interval [–1, 1], and fis a continuous function on [–1, 1], For the set of nodesx_k = cos [k/(n+1)] the w_n-polynomials coincide with the polynomialsof equiamplitude alternation introduced by de La Vall?e-Poussinand discussed in the literature earlier (Eterman, Malozemov,Meinardus, Cheney & Rivlin, Phillips & Taylor, Brutman,and others). It is shown that the generalized alternating polynomials arerelated to the polynomials of interpolation through the Lanczoseconomization process. Some approximation properties of w_n -polynomialsand W_n -parameters are studied. The application of w_n -polynomialsto function approximation and to the estimate of remainder termsfor quadrature formulas is discussed.     

The Stefan Problem with a Non-monotone Constitutive Relation

A finite difference approximation of the non-linear diffusionequation e_t = ((e))_xx' where is non-monotone, is shown to convergeto a measure valued solution. Some numerical results are described.     

Best Approximation in an Asymmetrically Weighted L1 Measure

Let f(x) be a given, real-valued, continuous function definedon an interval [a,b]of the real line. Given a set of m real-valued,continuous functions _j(x) defined on [a,b], a linear approximatingfunction can be formed with any real setA = {a₁, a₂,..., a_m}. We present results for determining A sothat F(A, x) is a best approximation to(x) when the measureof goodness of approximation is a weighted sum of |F(A, x)–f(x)|,the weights being positive constants, w, when F(A, x) f(x)and w2 otherwise (when w, = w2 = 1, the measure is the L₁, norm).The results are derived from a linear programming formulationof the problem. In particular, we give a theorem which shows when such bestapproximations interpolate the function at fixed ordinates whichare independent of f(x). We show how the fixed points can becalculated and we present numerical results to indicate thatthe theorem is quite robust.     

Properties of a Semi-discrete Approximation to the Beam Equation

The solution of the equation w(x)u_tt+[p(x)u_xx]_xx = 0, 0 < x < L, t > 0, where it is assumed that w and p are positive on the interval[O, L], is approximated by using the method of straight lines.The resulting approximation is a linear system of differentialequations with coefficient matrix S. The matrix S is studiedunder very general boundary conditions which result in a conservativesystem. In all cases the matrix S is either an oscillation matrixor possesses nearly all the properties of an oscillation matrix.     

Oscillation Properties of a Semi-discrete Approximation to the Beam Equation with a Second Order Term

The solution of the equation w(x)u_tt+[p(x)u_xx]_xx–[p(x)u_x]_x=0, 0< x < L, t > 0, where it is assumed that w, p,and q are positive on the interval [0, L], is approximated bythe method of straight lines. The resulting approximation isa linear system of differential equations with coefficient matrixS. The matrix S is studied under a variety of boundary conditionswhich result in a conservative system. In all cases the matrixS is shown to be similar to an oscillation matrix.     

Optimal Approximation of the Electrostatic Capacity Matrix of Two Conducting Spheres by a Short-Distance Asymptotic Expansion

An integral representation for the electrostatic capacity matrixC=[c_ij]_i,j=1,2 of two conducting spheres of radii R₁, and R₂is obtained. A short-distance asymptotic expansion is then derivedand its approximation properties for fixed (surface) distancer between the spheres are investigated. An error function is defined for c_ij(r) and its nthorder asymptotic approximant it has the property following from the divergence of the expansion, and thereby shows thatthe optimal approximation of c_ij(r) is achieved by an approximantof finite order n = n_ij(r) depending possibly on r and the indicesi,j. The value gives the quality of approximation of c_ij by the asymptotic expansion for a givendistance r between the spheres. The point sets and are introduced in order to describe the distance ranges where c_ij can be approximatedwithin a given error >0 by an asymptotic approximant of given order n, or at least by theoptimal approximant, respectively. The optimal order n_ij(r)and the -approximation sets and D() are investigated numerically.     

矩阵方程A1X1B1 + A2X2B2 + · · · + AlXlBl = C的中心对称解及其最佳逼近

设矩阵X=(x_ij) ∈^Rn×n, 如果xi_j=x_{n+1-i, n+1-j} (i,j=1,2, …,n), 则称X是中心对称矩阵. 该文构造了一种迭代法求矩阵方程A₁X₁B₁+A₂X₂B₂+…+A_lX_lB_l=C的中心对称解组(其中[X₁, X₂, …, X_l]是实矩阵组). 当矩阵方程相容时, 对任意初始的中心对称矩阵组[X₁⁽⁰⁾, X₂⁽⁰⁾, …, X_l⁽⁰⁾], 在没有舍入误差的情况下,经过有限步迭代,得到它的一个中心对称解组, 并且, 通过选择一种特殊的中心对称矩阵组, 得到它的最小范数中心对称解组. 另外, 给定中心对称矩阵组[X₁, X₂, …, X_l], 通过求矩阵方程A₁X₁B₁+A₂X₂B₂+…+A_lX_lB_l=C(其中C=C-A₁X₁B₁-A₂X₂B₂－…－A_lX_lB_l)的中心对称解组, 得到它的最佳逼近中心对称解组. 实例表明这种方法是有效的.     

Near-best Lp Approximations by Real and Complex Chebyshev Series

The expansion of a real or complex function in a series of Chebyshevpolynomials of the first and second kinds is discussed in thecontext of near-best approximation. The discussion covers realand complex approximation on the real interval [–1, 1]as a special example of the complex elliptical contour , as well as complex approximationon an elliptical domain, an ellipse exterior, and an ellipticalannulus (including special cases in which part of the boundarycollapses into a "crack"). Two distinct types of function spacesare considered, namely appropriately weighted L_p measure spacesand analytic function spaces, and resulting approximations areshown in all cases to be near-best in the L_p norm within a relativedistance asymptotic to (4^-2 log n)^2p-1–1 for all p (1p ), where relates to the order of approximation.     

Methods for the approximation of the matrix exponential in a Lie-algebraic setting

Discretization methods for ordinary differential equations basedon the use of matrix exponentials have been known for decades.This set of ideas has come off age and acquired greater interestrecently, within the context of geometric integration and discretizationmethods on manifolds based on the use of Lie-group actions. In the present paper we study the approximation of the matrixexponential in a particular context: given a Lie group G andits Lie algebra g, we seek approximants F(t B) of exp(t B) suchthat F(t B) G if B g. Having fixed a basis V₁, ..., V_d ofg, we write F(t B) as a composition of exponentials of the typeexp(_i (t) V_i), where _i for i = 1, 2, ..., d are scalar functions.In this manner it becomes possible to increase the order ofthe approximation without increasing the number of exponentialsto evaluate and multiply together. We study order conditionsand implementation details and conclude the paper with somenumerical experiments. Received 24 March 1999. Accepted 22 November 1999.     

Solving Linear Partial Differential Equations by Exponential Splitting

Let A₁, A₂,...,A_N be square matrices which do not commute. Weconsider approximations to the matrix exponential M = exp [t(A₁+ A₂ + ... + A_N)] of the form where each Y_k is a positive multiplying factor, and each E_kis a product of terms having the form exp (tA_n) for some >0 and 1 nN. This form is relevant to semi-discretization methodsfor the solution of linear partial differential equations andit produces systems which are easy to solve. The accuracy andstability of the splitting approximation are studied. It isshown that, even if the number of terms and the value of K arechosen to be large, the highest order of a stable approximationis two. Numerical examples are given.     

The A-Complexity of a Space

Suppose that A is a pointed CW-complex. The paper looks at howdifficult it is to construct an A-cellular space B from copiesof A by repeatedly taking homotopy colimits; this is determinedby an ordinal number called the complexity of B. Studying thecomplexity leads to an iterative technique, based on resolutions,for constructing the A-cellular approximation CW_A(X) of an arbitraryspace X.     

On the Approximation Numbers for the Finite Sections of Block Toeplitz Matrices

In this paper we discuss the asymptotic distribution of theapproximation numbers of the finite sections for a Toeplitzoperator and µ R, witha continuous matrix-valued generating function a. We prove thatthe approximation numbers of the finite sections T_n(a) = P_nT(a)P_nhave the k-splitting property, provided T(a) is a Fredholm operatoron . 2000 Mathematics SubjectClassification 47B35 (primary), 15A18, 47A58, 47A75, 65F20 (secondary).     

The Fourier-finite-element method with Nitsche mortaring

^** Email: bernd.heinrich{at}mathematik.tu-chemnitz.de The paper deals with a combination of the Fourier-finite-elementmethod with the Nitsche-finite-element method (as a mortar method).The approach is applied to the Dirichlet problem for the Poissonequation in 3D axisymmetric domains with non-axisymmetric data.The approximating Fourier method yields a splitting of the 3Dproblem into 2D problems on the meridian plane of the givendomain. For solving these 2D problems, the Nitsche-finite-elementmethod with non-matching meshes is applied. Some important propertiesof the approximation scheme are derived and the rate of convergencein an H¹-like norm as well as in the L₂-norm is estimated fora regular solution. Finally, some numerical results are presented.     

On approximation theorems for controllability of non-linear parabolic problems

^** Email: anil{at}math.iitb.ac.in^*** Email: mcj{at}math.iitb.ac.in^**** Email: akp{at}math.iitb.ac.in In this paper, we consider the following control system governedby the non-linear parabolic differential equation of the form: [graphic: see PDF] where A is a linear operator with dense domain and f(t, y)is a non-linear function. We have proved that under Lipschitzcontinuity assumption on the non-linear function f(t, y), theset of admissible controls is non-empty. The optimal pair (u^*,y^*) is then obtained as the limit of the optimal pair sequence{(u_n^*, y_n^*)}, where u_n^* is a minimizer of the unconstrainedproblem involving a penalty function arising from the controllabilityconstraint and y_n^* is the solution of the parabolic non-linearsystem defined above. Subsequently, we give approximation theoremswhich guarantee the convergence of the numerical schemes tooptimal pair sequence. We also present numerical experimentwhich shows the applicability of our result.