1: \begin{abstract}

2:   In recent years, contour-based eigensolvers have emerged as a

3:   standard approach for the solution of large and sparse eigenvalue

4:   problems. Building upon recent performance improvements through

5:   non-linear least square optimization of so-called rational filters,

6:   we introduce a systematic method to design these filters by

7:   minimizing the worst-case convergence ratio and eliminate the

8:   parametric dependence on weight functions. Further, we provide an

9:   efficient way to deal with the box-constraints which play a central

10:   role for the use of iterative linear solvers in contour-based

11:   eigensolvers. Indeed, these parameter-free filters consistently

12:   minimize the number of iterations and the number of FLOPs to reach

13:   convergence in the eigensolver. As a byproduct, our rational filters

14:   allow for a simple solution to load balancing when the solution of

15:   an interior eigenproblem is approached by the slicing of the sought

16:   after spectral interval.

17: \end{abstract}

18: