1: \begin{abstract} \noindent

2: We present a systematic study  on the linear convergence rates of

3: the powers of (real or complex) matrices. We derive a characterization

4: when the optimal convergence rate is attained. This

5: characterization is given in terms of semi-simpleness of all

6: eigenvalues having the second-largest modulus after 1.

7: We also provide applications of our general results to analyze

8: the optimal convergence rates for several relaxed alternating projection

9:  methods and the generalized Douglas-Rachford splitting methods for

10:  finding the projection on the intersection of two subspaces. Numerical experiments

11:  confirm our convergence analysis.

12: \end{abstract}

13: