Optimal Block Preconditioner for an Efficient Numerical Solution of the Elliptic Optimal Control Problems Using GMRES Solver

Optimal Block Preconditioner for an Efficient Numerical Solution of the Elliptic Optimal Control Problems Using GMRES Solver

Year:    2023

Author:    K. Muzhinji

International Journal of Numerical Analysis and Modeling, Vol. 20 (2023), Iss. 1 : pp. 47–66

Abstract

Optimal control problems are a class of optimisation problems with partial differential equations as constraints. These problems arise in many application areas of science and engineering. The finite element method was used to transform the optimal control problems of an elliptic partial differential equation into a system of linear equations of saddle point form. The main focus of this paper is to characterise and exploit the structure of the coefficient matrix of the saddle point system to build an efficient numerical process. These systems are of large dimension, block, sparse, indefinite and ill conditioned. The numerical solution of saddle point problems is a computational task since well known numerical schemes perform poorly if they are not properly preconditioned. The main task of this paper is to construct a preconditioner the mimic the structure of the system coefficient matrix to accelerate the convergence of the generalised minimal residual method. Explicit expression of the eigenvalue and eigenvectors for the preconditioned matrix are derived. The main outcome is to achieve optimal convergence results in a small number of iterations with respect to the decreasing mesh size $h$ and the changes in $δ$ the regularisation problem parameters. The numerical results demonstrate the effectiveness and performance of the proposed preconditioner compared to the other existing preconditioners and confirm theoretical results.

You do not have full access to this article.

Already a Subscriber? Sign in as an individual or via your institution

Journal Article Details

Publisher Name:    Global Science Press

Language:    English

DOI:    https://doi.org/10.4208/ijnam2023-1003

International Journal of Numerical Analysis and Modeling, Vol. 20 (2023), Iss. 1 : pp. 47–66

Published online:    2023-01

AMS Subject Headings:    Global Science Press

Copyright:    COPYRIGHT: © Global Science Press

Pages:    20

Keywords:    Partial differential equations (PDEs) PDE-optimal control problems saddle point problem block preconditioners preconditioned generalised minimal residual method (PGMRES).

Author Details

K. Muzhinji