1: \begin{abstract}

2: The convergence of GMRES for solving linear systems can be influenced heavily by the structure of the right hand side. Within the solution of

3: eigenvalue problems via inverse iteration or subspace iteration, the right hand side is generally related to an approximate invariant subspace of the linear

4: system. We give detailed and new bounds on (block) GMRES that take the special behavior of the right hand side into account and explain the initial sharp

5: decrease of the GMRES residual. The bounds motivate the use of specific preconditioners for these  eigenvalue problems, e.g. tuned and polynomial

6: preconditioners, as we describe.  The numerical results show that the new (block) GMRES bounds are much sharper than conventional bounds and that preconditioned

7: subspace iteration with either a tuned or polynomial preconditioner should be used in practice.

8: \end{abstract}

9: