1: \begin{abstract}

2:         We present an algorithm for the solution of Sylvester equations 

3:         with right-hand side of low rank. The method is based on projection

4:         onto a block rational Krylov subspace, with two key contributions 

5:         with respect to the state-of-the-art. First, we show how to maintain 

6:         the last pole equal to infinity throughout the iteration, by means 

7:         of pole reordering, which allows for a cheap evaluation of the 

8:         true residual at every step. Second, we extend the convergence 

9:         analysis in [Beckermann B., \emph{An error analysis for rational {G}alerkin projection applied to the

10:         {S}ylvester equation}, SINUM, 2011] to the block case. 

11:         This extension allows us to link the convergence with the problem of minimizing 

12:         the norm of a small rational matrix over the spectra or field-of-values 

13:         of the involved matrices. This is in contrast with the non-block case, 

14:         where the minimum problem is scalar, instead of 

15:         matrix-valued. Replacing the norm of the objective function 

16:         with a more easily evaluated

17:         function yields several adaptive pole selection strategies, providing 

18:         a theoretical analysis for known heuristics, as well as effective novel 

19:         techniques. 

20:     \end{abstract}

21: