On the stability of rational approximations to the cosine with only imaginary poles

Letr(z) be a rational approximation to cosz with only imaginary poles ±i ₁ ^–1/2 , ±i ₂ ^–1/2 , ..., ±i _m ^–1/2 such that |cozz –r(z)| C|z|^2m+2 as |z| 0. If the degree of the numerator ofr(z) is less than or equal to 2m and _i m/4,i=1, ...,m, then we show that |r(z)|1 for all realz.     

Attainable order of rational approximations to the exponential function with only real poles

Rational approximations of the form _i=0 ^m a _i q ⁱ / _i=1 ⁿ (1+ _i q) to exp(–q),qC, are studied with respect to order and error constant. It is shown that the maximum obtainable order ism+1 and that the approximation of orderm+1 with least absolute value of the error constant has ₁=₂=...= _n . As an application it is shown that the order of av-stage semi-implicit Runge-Kutta method cannot exceedv+1.     

A stability result for general linear methods with characteristic function having real poles only

It is shown that there exist A-stable multistep formulae, with a characteristic function havings poles, all of which are real, with orderp satisfyingp>s+1. This contradicts the widely held belief thatp=s+1 is the maximum possible order of such a method.     

Real pole approximations to the exponential function

Rational approximations to the exponential function with real, not necessarily distinct poles are studied in this paper. The orthogonality relation is established in order to show that the zeros of the collocation polynomial of the corresponding Runge-Kutta method are all real, simple and positive. It is proven, that approximants with the smallest error constant are the Restricted Padé approximants of Nørsett. Some results concerning acceptability properties are given.This work was supported by RSS, Ljubljana while the author was at Division of Mathematical Sciences, Norwegian Institute of Technology, Trondheim.     

On implicit Runge-Kutta methods with a stability function having distinct real poles

Because of their potential for offering a computational speed-up when used on certain multiprocessor computers, implicit Runge-Kutta methods with a stability function having distinct poles are analyzed. These are calledmultiply implicit (MIRK) methods, and because of the so-calledorder reduction phenomenon, their poles are required to be real, i.e., only real MIRK's are considered. Specifically, it is proved that a necessary condition for aq-stage, real MIRK to beA-stable with maximal orderq+1 is thatq=1, 2, 3 or 5. Nevertheless, it is shown that for every positive integerq, there exists aq-stage, real MIRK which is stronglyA ₀-stable with orderq+1, and for every evenq, there is aq-stage, real MIRK which isI-stable with orderq. Finally, some useful examples of algebraically stable real MIRK's are given.This work was supported by the National Aeronautics and Space Administration under NASA Contract No. NAS1-18107 while the author was in residence at the Institute for Computer Applications in Science and Engineering (ICASE), NASA Langley Research Center, Hampton, VA 23665-5225.     

Error control of rational approximations to the exponential function

Several methods for the numerical solution of stiff ordinary differential equations require approximation of an exponential of a matrix. In the present paper we present a technique for estimating the error incurred in replacing a matrix exponential by a rational approximation. This estimation is done by introducing another approximation, of superior order, whose aposteriori evaluation is cheap. Properties of the new approximation pertaining to both its stability and the behavior of the error for matrices with negative eigenvalues are analyzed.     

On the simultaneous rational approximations to real and p-adic numbers

Lithuanian Mathematical Journal - We consider simultaneous rational approximations to real and p-adic numbers. We prove that for any irrational number α0 and p-adic number α, there are...     

On the construction of restricted-denominator exponential W-methods

In this paper we consider the practical construction of exponential W-methods for the solution of large stiff nonlinear initial value problems, based on the restricted-denominator rational approach for the computation of the functions of matrices required. This approach is employed together with the Krylov subspace method based on the Arnoldi algorithm. Two integrators are constructed and tested on some classical stiff equations arising from the semidiscretization of parabolic problems.     

On the stability of spline-collocation methods of multivalue type

In this paper the general classV of spline-collocation methods for first order systems of ordinary differential equations is investigated. The methods can in part be regarded as so-called multivalue methods. This type contains the generalized singly-implicit methods treated by Butcher.It is shown here, how any multivalue type representative ofV yields a matrix valued function, for the characterization of stability at infinity. It is shown in particular, that the structure of allows us to construct infinity-stable methods by an appropriate choice of the collocation points.     

On the q-Bernstein polynomials of rational functions with real poles

The paper aims to investigate the convergence of the q  -Bernstein polynomials _Bn,q(f;x)$B_{n,q}(f;x)$ attached to rational functions in the case q>1$q>1$. The problem reduces to that for the partial fractions (x−α)^−j${(x-\alpha )}^{-j}$, j∈N$j\in \mathbb{N}$. The already available results deal with cases, where either the pole α   is simple or α≠q^−m$\alpha \ne q^{-m}$, m∈_N0$m\in {\mathbb{N}}_0$. Consequently, the present work is focused on the polynomials _Bn,q(f;x)$B_{n,q}(f;x)$ for the functions of the form f(x)=(x−q^−m)^−j$f(x)={(x-q^{-m})}^{-j}$ with j?2$j?2$. For such functions, it is proved that the interval of convergence of {_Bn,q(f;x)}$\{B_{n,q}(f;x)\}$ depends not only on the location, but also on the multiplicity of the pole – a phenomenon which has not been considered previously.     

A note on a recent result of rational approximations to the exponential function

In a recent paper by Nørsett and Wolfbrandt [1] it is shown that the maximum attainable order ofN-approximationsR _m,n(u) to exp (u) ism + 1. The purpose of this note is to present an alternative proof of this result.     

A lower bound for the number of stages of an explicit continuous Runge-Kutta method to obtain convergence of given order

In this paper, we show that in order to construct an explicit continuous Runge-Kutta method of orderp, we need at leastp+d – 1 stages, whered is the dimension of the space generated by the vectors(), for rooted trees of orderp. In many particular cases, this lower bound is precisely 2p – 2.Partially supported by C.I.C.Y.T.     

On RD‐rational Krylov approximations to the core‐functions of exponential integrators

The paper deals with the application of the restricted‐denominator rational Krylov method, recently discussed in (BIT 2004; 44 (3):595–615; SIAM J. Sci. Comput. 2005; 27 :1438–1457), to the computation of the action of the so‐called φ‐functions, which play a fundamental role in several modern exponential integrators. The analysis here presented is devoted in particular to the construction of error estimates of easy practical use. Copyright © 2007 John Wiley & Sons, Ltd.     

Low and high frequency approximations to eigenvibrations of string with double contrasts

We study eigenvibrations for inhomogeneous string consisting of two parts with strongly contrasting stiffness and mass density. In this work we treat a critical case for the high frequency approximations, namely the case when the order of mass density inhomogeneity is the same as the order of stiffness inhomogeneity, with heavier part being softer. The limit problem for high frequency approximations depends nonlinearly on the spectral parameter. The quantization of the spectral semiaxis is applied in order to get a close approximations of eigenvalues as well as eigenfunctions for the prime problem under perturbation.     

Barycentric rational interpolation with no poles and high rates of approximation

It is well known that rational interpolation sometimes gives better approximations than polynomial interpolation, especially for large sequences of points, but it is difficult to control the occurrence of poles. In this paper we propose and study a family of barycentric rational interpolants that have no real poles and arbitrarily high approximation orders on any real interval, regardless of the distribution of the points. These interpolants depend linearly on the data and include a construction of Berrut as a special case.     

Stability and oscillations of numerical solutions for differential equations with piecewise continuous arguments of alternately advanced and retarded type

This paper is concerned with the numerical properties of θ-methods for the solution of alternately advanced and retarded differential equations with piecewise continuous arguments. Using two θ-methods, namely the one-leg θ-method and the linear θ-method, the necessary and sufficient conditions under which the analytic stability region is contained in the numerical stability region are obtained, and the conditions of oscillations for the θ-methods are also obtained. It is proved that oscillations of the analytic solution are preserved by the θ-methods. Furthermore, the relationships between stability and oscillations are revealed. Some numerical experiments are presented to illustrate our results.