1: \begin{abstract} In this work we construct multigrid preconditioners to accelerate the solution process of a 

2: linear-quadratic optimal control problem constrained by the Stokes system. 

3: The first order optimality conditions of the control problem form  a linear system (the KKT system)

4: connecting the state, adjoint, and control variables. Our approach is to 

5: eliminate the state and adjoint variables by essentially solving two Stokes systems, 

6: and to construct efficient multigrid preconditioners for the 

7: Schur-complement of the block associated with the state and adjoint variables.

8: These multigrid preconditioners are shown to be of optimal order

9: with respect to the convergence properties of the discrete methods used to

10: solve the Stokes system. In particular, the number of conjugate gradient iterations

11: is shown to decrease as the resolution increases, a feature shared by similar

12: multigrid preconditioners for elliptic constrained optimal control problems.

13: \end{abstract}

14: