Preconditioned iterative methods for Navier–Stokes control problems

Abstract

PDE-constrained optimization problems are a class of problems which have attracted much recent attention in scientific computing and applied science. In this paper we discuss preconditioned iterative methods for a class of (time-independent) Navier–Stokes control problems, one of the main problems of this type in the field of fluid dynamics. Having detailed the Picard-type iteration we use to solve the problems and derived the structure of the matrix system to be solved at each step, we utilize the theory of saddle point systems to develop efficient preconditioned iterative solution techniques for these problems. We also require theory of solving convection–diffusion control problems, as well as a commutator argument to justify one of the components of the preconditioner.