1: \begin{abstract}

2: % \noindent  %\vspace{3mm}

3: 

4: Large scale tensors, including large scale Hankel tensors,

5: have many applications in science and engineering.

6: In this paper, we propose an inexact curvilinear search optimization method

7: to compute Z- and H-eigenvalues of $m$th order $n$ dimensional Hankel tensors, where $n$ is large.

8: Owing to the fast Fourier transform, the computational cost of each iteration

9: of the new method is about $\mathcal{O}(mn\log(mn))$.

10: Using the Cayley transform, we obtain an effective curvilinear search scheme.

11: Then, we show that every limiting point of iterates generated by the new algorithm is

12: an eigen-pair of Hankel tensors. Without the assumption of a second-order sufficient condition,

13: we analyze the linear convergence rate of iterate sequence by the Kurdyka-{\L}ojasiewicz property.

14:  Finally, numerical experiments for Hankel tensors,

15: whose dimension may up to one million, are reported to show the efficiency

16: of the proposed curvilinear search method.

17: 

18: \vspace{3mm}

19: \noindent {\bf Key words:}\hspace{2mm} Cayley transform, curvilinear search, eigenvalue,

20:   fast Fourier transform, Hankel tensor, Kurdyka-{\L}ojasiewicz property, large scale tensor.

21: 

22: \vspace{3mm}

23: \noindent {\bf AMS subject classifications (2010):}\hspace{2mm} 15A18, 15A69, 65F15, 65K05, 90C52.

24: \vspace{3mm}

25: \end{abstract}