Summary
We first discuss the solution of a fixed point equationxΦx of a Fréchet differentiable self-mapping Φ by iterative methods of the general form
defined by two infinite nonsingular upper triangular matricesB=(β j, m ) andC=(γ j, m ) with column sums 1. We show that two such methods, defined byB, C and\(\tilde B,\tilde C\), respectively, are in a certain sense equivalent if and only if\(C^{ - 1} B = \tilde C^{ - 1} \tilde B\). In particular,\(\hat B: = C^{ - 1} \) and the unit matrix (replacingC) define an equivalent so-called semiiterative method. We introduce (k, l)-step methods as those whereB andC have upper bandwidthk andl−1, respectively. They require storing max {k, l} previous iterates only. For stationary methodsB andC have Toeplitz structure except for their first row, which is chosen such that the column sum condition holds. An Euler method, which may require to store all iterates, is equivalent to a (stationary) (k, l)-step method if and only if the underlying conformal mapg is a rational function of the formg(w)=w(μ 0+...+μ k −k)/ (v 0+...+v l−1 w −+1). By choosingg withg(1)=1 such that for some ρ<1 it maps |w|>ρ onto the exterior of some continuumS known to contain the eigen values of the Fréchet derivative of Φ, one obtains a feasible procedure for designing a locally converging stationary (k, l)-step method custom-made for a set of problems.
In the case Φx≔Tx+d, with a linear operatorT, where one wants to solve a linear system of equations, we show that the residual polynomials of a stationary semiiterative method are generalized Faber polynomials with respect to a particular weight function. Using another weight function leads to what we call almost stationary methods. (The classical Chebyshev iteration is an example of such a method.) We define equivalent almost stationary (k, l)-step methods and give a corresponding convergence result.
Similar content being viewed by others
References
Bittner, L.: Über ein mehrstufiges Iterationsverfahren zur Lösung von linearen Gleichungen. Numer. Math.6, 161–180 (1964)
Eiermann, M., Li, X., Varga, R.S.: On hybrid semiiterative methods. SIAM J. Numer. Anal. (to appear)
Eiermann, M., Niethammer, W.: On the construction of semiiterative methods. SIAM J. Numer. Anal.20, 1153–1160 (1983)
Eiermann, M., Niethammer, W., Varga, R.S.: A study of semiiterative methods for nonsymmetric systems of linear equations. Numer. Math.47, 505–533 (1985)
Engeli, M., Ginsburg, T., Rutishauser, H., Stiefel, E.: Refined iterative methods for computation of the solution and the eigenvalues of self-adjoint boundary value problems. Mitt. Inst. Angew. Math. ETH Zürich, No. 8. Basel: Birkhäuser 1959
Faddeev, D.K., Faddeeva, V.N.: Computational methods of linear algebra, 1st Ed. San Francisco: Freeman 1963
Gekeler, E.: Über ein mehrstufiges Iterationsverfahren und die Lösung der Hammersteinschen Gleichung. Numer. Math.19, 351–360 (1972)
Gutknecht, M.H.: An iterative method for solving linear equations based on minimum norm Pick-Nevanlinna interpolation. In: Chui, C.K., Schumaker, L.L., Ward, J.D. (eds.) Approximation theory V, pp. 371–374. New York: Academic Press 1986
Gutknecht, M.H.: Numerical conformal mapping methods based on function conjugation. J. Comput. Appl. Math.14, 31–77 (1986)
Gutknecht, M.H., Kaiser, A.: Iterativek-step methods for computing possibly repulsive fixed points in Banach spaces. J. Math. Anal. Appl.125, 104–123 (1987)
Gutknecht, M.H., Niethammer, W., Varga, R.S.:k-step iterative methods for solving nonlinear systems of equations. Numer. Math.48, 699–712 (1986)
Kitchen, J.W. Jr.: Concerning the convergence of iterates to fixed points. Studia Math.27, 247–249 (1966)
Mann, W.R.: Averaging to improve convergence in iterative processes. In: M.Z. Nashed (ed.) Functional analysis methods in numerical analysis, pp. 169–179. Berlin Heidelberg New York: Springer 1979
Mann, W.R.: Mean value methods in iteration. Proc. Amer. Math. Soc.4, 506–510 (1953)
Niethammer, W.: Iterationsverfahren und allgemeine Euler-Verfahren. Math. Z.102, 288–317 (1967)
Niethammer, W., Schempp, W.: On the construction of iteration methods for linear equations in Banach spaces by summation methods. Aequationes Math.5, 273–284 (1970)
Niethammer, W., Varga, R.S.: The analysis ofk-step iterative methods for linear systems from summability theory. Numer. Math.41, 177–206 (1983)
Opitz, G.: Steigungsmatrizen. Z. Ang. Math. Mech.44, T52-T54 (1964)
Ortega, J.M., Rheinboldt, W.C.: Iterative solution of nonlinear equations in several variables. 1st Ed. New York: Academic Press 1970
Parsons, B.N.: Generalk-part stationary iterative solutions to linear systems. SIAM J. Numer. Anal.24, 188–198 (1987)
Schempp, W.: Iterative solution of linear operator equations in Hilbert space and optimal Euler methods. Arch. Math.21, 390–395 (1970)
Smirnov, V.I., Lebedev, N.A.: Functions of a complex variable 1st Ed. Cambridge: The M.I.T. Press 1968
Toeplitz, O.: Über allgemeine Mittelbildungen. Prace Mat.-Fiz.22, 113–119 (1911)
Trefethen, L.N.: Rational Chebyshev approximation on the unit disk. Numer. Math.37, 297–320 (1981)
Varga, R.S.: Matrix iterative analysis. Englewood Cliffs, N.J.: Prentice-Hall 1962
Warner, D.D.: Hermite interpolation with rational functions. PhD thesis, University of California at San Diego 1974
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Gutknecht, M.H. Stationary and almost stationary iterative (k, l)-step methods for linear and nonlinear systems of equations. Numer. Math. 56, 179–213 (1989). https://doi.org/10.1007/BF01409784
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF01409784