b19c4b69b89c80ec.tex
1: \begin{abstract}
2: We present the convergence rates and the explicit error bounds 
3: of Hill's method, which is a numerical method for computing the spectra of 
4: ordinary differential operators with periodic coefficients. 
5: This method approximates the operator by a finite dimensional matrix. 
6: On the assumption that the operator is self-adjoint,
7: it is shown that, under some conditions, we can obtain the convergence rates of eigenvalues with respect to
8: the dimension and the explicit error bounds.
9: Numerical examples demonstrate that we can verify these conditions 
10: using Gershgorin's theorem for some real problems.
11: Main theorems are proved using the Dunford integrals which project an eigenvector to the corresponding eigenspace.\\
12: 
13: \noindent
14: {\bf Keywords} Hill's method, Convergence rate, Error bound, Differential Operator, Eigenvalue Problem\\
15: 
16: \noindent
17: {\bf Mathematics Subject Classification (2000)} 65L15, 65L20, 65L70
18: \end{abstract}