1: \begin{abstract}

2: The solution of matrices with $2\times 2$ block structure arises in

3: numerous areas of computational mathematics, such as PDE

4: discretizations based on mixed-finite element methods, constrained

5: optimization problems, or the implicit or steady state treatment of

6: any system of PDEs with multiple dependent variables. Often, these

7: systems are solved iteratively using Krylov methods and some form of

8: block preconditioner. Under the assumption that one diagonal block

9: is inverted exactly, this paper proves a direct equivalence between

10: convergence of $2\times2$ block preconditioned Krylov or fixed-point

11: iterations to a given tolerance, with convergence of the underlying

12: preconditioned Schur-complement problem. In particular, results

13: indicate that an effective Schur-complement preconditioner is a

14: necessary and sufficient condition for rapid convergence of

15: $2\times 2$ block-preconditioned GMRES, for arbitrary

16: relative-residual stopping tolerances. A number of corollaries and

17: related results give new insight into block preconditioning, such as

18: the fact that approximate block-LDU or symmetric block-triangular

19: preconditioners offer minimal reduction in iteration over

20: block-triangular preconditioners, despite the additional

21: computational cost. Theoretical results are verified numerically on

22: a nonsymmetric steady linearized Navier--Stokes discretization,

23: which also demonstrate that theory based on the assumption of an

24: exact inverse of one diagonal block extends well to the more

25: practical setting of inexact inverses.

26: \end{abstract}

27: