Abstract
Model problem ADI iteration is discussed for three distinct classes of problems. The first is discretized elliptic systems with separable coefficients so that difference equations may be split into two commuting matrices. The second is where the model ADI problem approximates the actual nonseparable problem and serves as a preconditioner. The third is an entirely different class of problems than initially considered. These are Lyapunov and Sylvester matrix equations in which commuting operations are inherent.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Nowadays, a variety of names are attached to variants of the Lanczos recursion formulas derived by minimizing different functionals. Forty years ago Gabe Horvay (a GE mechanics expert and one of my associates at KAPL) introduced me to this new approach developed by his friend Lanczos and, influenced strongly by Gabe, I became accustomed to referring to all these schemes as “Lanczos algorithms.” Hence, the method of “conjugate gradients” is often referred to as “Lanczos’ method” in my early works.
- 2.
My definition of positive definite in those days implied symmetry. More recently, the term has been used by some with a different definition so that it is now customary to impose symmetry and denote A as “SPD” for “symmetric and positive definite.” I still prefer the old definition in Wachspress, 1966 , but approve wholeheartedly of the use of SPD to resolve any doubt.
- 3.
Al Schatz (Cornell) advised me when I was preparing work on this preconditioner for publication that he had considered a related approximation for solving finite element problems but I have not yet seen a published reference to this work. His effort was devoted more to approximating equations over nonrectangular grids by preconditioning equations over rectangular grids.
- 4.
While at the University of Tennessee in Knoxville I interacted with Al Geist at Oak Ridge and awakened his interest in gaussian reduction to tridiagonal form. Our work stimulated renewed interest by several mathematicians with whom we communicated.
References
Bartels RH, Stewart GW (1972) Solution of the matrix equation AX + XB = C. Commun. ACM 15:820–826
Buzbee BL, Golub GH, Nielson CW (1970) On direct methods for solving Poisson’s equation. SIAM J Numer Anal 7:627–656
Byers R (1983) Hamilton and symplectic algorithms for the algebraic Riccati equation. Ph.D. Thesis, Cornell University
Cesari L (1937) Sulla risoluzione dei sistemi di equazioni lineari par aprossimazioni successive. Atti Accad. Naz. Lincei. Rend. Cl. Sci. Fis. Mat. Nat. VI S–25:422–428
Concus P, Golub G (1973) Use of fast direct methods for the efficient numerical solution of nonseparable elliptic equations. SIAM J Numer Anal 20:1103–1120
Dax A, Kaniel S (1981) The ELR method for computing the eigenvalues of a general matrix. SIAM J Numer Anal 18:597–605
Diliberto SP, Strauss EG (1951) On the approximation of a function of several variables by a sum of functions of fewer variables. Pacific J Math 1:195–210
D’Yakonov EG (1961) An iteration method for solving systems of linear difference equations. Soviet Math (AMS translation) 2:647
Engeli M, Ginsburg T, Rutishauser H, Stiefel E (1959) Refined iterative methods for computation of the solution and the eigenvalues of self-adjoint boundary value problems. Mitteilungen aus dem Institut für angewandte Mathematik, vol 8. Birkhäuser, Basel
Geist GA (1989) Reduction of a general matrix to tridiagonal form. ORNL/TM-10991
Geist GA, Lu A, Wachspress EL (1989) Stabilized Gaussian reduction of an arbitrary matrix to tridiagonal form. ORNL/TM-11089
Golub GH, O’Leary DP (1987) Some history of the conjugate gradient and Lanczos algorithms. UMIACS-TR-87-20, CS-TR-1859, University of Maryland Institute for Advanced Computer Studies and Computer Science Department
Golub GH, Nash S, Van Loan C (1979) A Hessenberg-Shur method for solution of the problem AX + XB = C. IEEE Trans Automat Contr AC24:909–913
Hare DE, Tang WP (1989) Toward a stable tridiagonalization algorithm for general matrices. University of Waterloo technical report CS-89-03, 1989
Howell GW (1994) Toward an efficient and stable determination of spectra of a general matrix and a more efficient solution of the Lyapunov equation. First world congress of nonlinear analysis, 1994
Hestenes M, Stieffel E (1952) Methods of conjugate gradients for solving linear systems. J Res NBS 49(6):409–436
Howell G, Diaa N (2005) Algorithm 841: BHESS: Gaussian reduction to a similar banded Hessenberg form, ACM transactions on Mathematical Software vol. 31 (1)
Hurwitz H (1984) Infinitesimal scaling: A new procedure for modeling exterior field problems. IEEE Trans 20:1918–1923
Istace MP, Thiran JP (1993) On the third and fourth Zolotarev problems in the complex plane. Math Comput
Opfer G, Schober G (1984) Richardson’s iteration for nonsymmetric matrices. Linear Algebra Appl 58:343–361
Peaceman DW, Rachford HH Jr (1955) The numerical solution of parabolic and elliptic differential equation. J SIAM 3:28–41
Smith RA (1968) Matrix equation XA + XB = C. SIAM J Appl Math 16(1):198–201
Starke G (1989) Rational minimization problems in connection with optimal ADI-parameters for complex domains. Ph.D. thesis, Institute für Praktische Mathematik, Universität Karlsruhe, FRG
Tang WP (1988) A stabilized algorithm for tridiagonalization of an unsymmetric matrix. University of Waterloo Technical Report CS-88-14
Varga RS (1962) Matrix iterative analysis. Prentice Hall, Englewood Cliffs
Wachspress EL (1957) CURE: A generalized two-space-dimension multigroup coding for the IBM-704. Knolls Atomic Power Report KAPL-1724, AEC R&D Physics and Mathematics TID-4500, 13th edn. Knolls Atomic Power Lab, Schenectady
Wachspress EL (1963) Extended application of alternating-direction-implicit iteration model problem theory. J Soc Indust Appl Math 11:994–1016
Wachspress EL (1966) Iterative solution of elliptic systems. Prentice Hall, Englewood Cliffs
Wachspress EL (1984) Generalized ADI preconditioning. Comput Math Appl 10(6):457–461
Wachspress EL (1988a) ADI solution of the Lyapunov equation. MSI Workshop on practical iterative methods for large-scale computations, Minneapolis, MN, 1988
Wachspress EL (1988b) Iterative solution of the Lyapunov matrix equation. Appl Math Lett 1: 87–90
Wachspress EL (1991) Optimum alternating direction implicit iteration parameters for rectangular regions in the complex plane. Center for Mathematical Analysis of the Australian National University Report CMA-R33-90, 1991
Watkins D (1988) Use of the LR algorithm to tridiagonalize a general matrix. SIAM annual meeting, Minneapolis, MN, 1988
Wilkinson JH (1965) The algebraic eigenvalue problem. Oxford University Press, Oxford
Young DM, Wheeler MF (1964) Alternating direction methods for solving partial difference equations. Nonlinear problems in engineering. Academic, New York, pp 220–246
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2013 Springer Science+Business Media New York
About this chapter
Cite this chapter
Wachspress, E. (2013). Model Problems and Preconditioning. In: The ADI Model Problem. Springer, New York, NY. https://doi.org/10.1007/978-1-4614-5122-8_3
Download citation
DOI: https://doi.org/10.1007/978-1-4614-5122-8_3
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-1-4614-5121-1
Online ISBN: 978-1-4614-5122-8
eBook Packages: EngineeringEngineering (R0)