1: \begin{abstract}

2: In this paper, we propose, analyze and demonstrate 

3: a dynamic momentum method to accelerate power and 

4: inverse power iterations with minimal computational overhead. 

5: The method is appropriate for real, diagonalizable matrices, and does not require

6: a priori spectral knowledge.

7: We review and extend background results on previously developed static momentum 

8: accelerations for the power iteration through the connection between the momentum 

9: accelerated iteration and the standard power iteration applied to an augmented matrix.

10: We show that the augmented matrix is defective for the optimal parameter choice.

11: We then present our dynamic method which updates the momentum parameter at each

12: iteration based on the Rayleigh quotient and two previous residuals. 

13: We present convergence and stability theory for the method by considering a power-like

14: method consisting of multiplying an initial vector by a sequence of augmented matrices.

15: We demonstrate the developed method on a number of benchmark problems, and see that

16: it outperforms both the power iteration and often the static momentum 

17: acceleration with optimal parameter choice. Finally, we present and

18: demonstrate an explicit extension of the algorithm to inverse power iterations.

19: \end{abstract}

20: