On the theory of equivalent operators and application to the numerical solution of uniformly elliptic partial differential equations

  • Authors:

  • V Faber;Thomas A Manteuffel;Seymour V Parter

  • Affiliations:

  • Computing and Communications Division, Los Alamos National Laboratory, Los Alamos, New Mexico, U.S.A.;University of Colorado at Denver, Denver, Colorado and Computing and Communications Division, Los Alamos National Laboratory, Los Alamos, New Mexico, U.S.A.;Department of Mathematics, University of Wisconsin, Madison, Wisconsin, U.S.A.

  • Venue:
  • Advances in Applied Mathematics
  • Year:
  • 1990

Quantified Score

Hi-index 0.00

Visualization

Abstract

This work is motivated by the preconditioned iterative solution of linear systems that arise from the discretization of uniformly elliptic partial differential equations. Iterative methods with bounds independent of the discretization are possible only if the preconditioning strategy is based upon equivalent operators. The operators A, B: W - V are said to be V norm equivalent if @?Au@?"v@?Bu@?"v is bounded above and below by positive constants for u@e D, where D is ''sufficiently dense.'' If A is V norm equivalent to B, then for certain discretization strategies one can use B to construct a preconditioned iterative scheme for the approximate solution of the problem Au = F. The iteration will require an amount of work that is at most a constant times the work required to approximately solve the problem Bu@^ = \@^tf to reduce the V norm of the error by a fixed factor. This paper develops the theory of equivalent operators on Hubert spaces. Then, the theory is applied to uniformly elliptic operators. Both the strong and weak forms are considered. Finally, finite element and finite difference discretizations are examined.