1: \begin{abstract}

2: % Body of abstract:

3: \noindent

4: Non-smoothness at optimal points is a common phenomenon in many eigenvalue

5: optimization problems. We consider two recent algorithms to minimize the

6: largest eigenvalue of a Hermitian matrix dependent on one 

7: parameter, both proven to be globally convergent unaffected by

8: non-smoothness. One of these models the eigenvalue function with a piece-wise 

9: quadratic function, and effective in dealing with non-convex problems. The other projects

10: the Hermitian matrix into subspaces formed of eigenvectors, and effective in dealing

11: with large-scale problems. We generalize the latter slightly to cope with non-smoothness.

12: For both algorithms, we analyze the rate-of-convergence

13: in the non-smooth setting, when the largest eigenvalue is multiple at the minimizer

14: and zero is strictly in the interior of the generalized Clarke derivative, and prove that both algorithms

15: converge rapidly. The algorithms are applied to, and the deduced results are illustrated on 

16: the computation of the inner numerical radius, the modulus of the point on the boundary

17: of the field of values closest to the origin, which carries significance for instance for the 

18: numerical solution of a definite generalized symmetric eigenvalue problem.

19: \\[10pt]

20: % Keywords:

21: \noindent

22: \textbf{Key words.} Eigenvalue Optimization, Non-Smooth Optimization, Global Optimization,

23: Rate-of-Convergence, Subspace Projections, Field of Values, Definite Matrix Pairs, Inner Numerical Radius,

24: Generalized Eigenvalue Problem 

25: \\[5pt]

26: % AMS subject classifications:

27: \noindent

28: \textbf{AMS subject classifications.} 65F15, 90C26, 90C06, 15A60

29: % 65F15, 90C26, 93D09, 93D15, 49K35

30: %65F15, % eigenvalues, eigenvectors

31: %65D05, % interpolation

32: %15A22  % matrix pencils

33: \end{abstract}

34: