1: \begin{abstract}

2:     We present a novel method to estimate the dominant eigenvalue and eigenvector pair

3:     of any non-negative real matrix via graph infection.

4:     The key idea in our technique lies in approximating the solution to the first-order matrix ordinary differential equation (ODE) with the Euler method.

5:     Graphs, which can be weighted, directed, and with loops, are first converted to its adjacency matrix A.

6:     Then by a naive infection model for graphs, we establish the corresponding first-order matrix ODE,

7:     through which A's dominant eigenvalue is revealed by the fastest growing term.

8:     When there are multiple dominant eigenvalues of the same magnitude,

9:     the classical power iteration method can fail.

10:     In contrast, our method can converge to the dominant eigenvalue even when same-magnitude counterparts exist, be it complex or opposite in sign. 

11:     We conduct several experiments comparing the convergence between our method and power iteration.

12:     Our results show clear advantages over power iteration for tree graphs, bipartite graphs, directed graphs with periods, and Markov chains with spider-traps.

13:     To our knowledge, this is the first work that estimates dominant eigenvalue and eigenvector pair

14:     from the perspective of a dynamical system and matrix ODE.

15:     We believe our method can be adopted as an alternative to power iteration, especially for graphs.

16: 

17: \keywords{Dominant Eigenvalue \and Eigenvector Centrality \and First-order Matrix ODE \and Euler Method \and Graph Infection}

18: \end{abstract}

19: