1
$\begingroup$

There are asymptotic formulas for eigenvalues $\lambda_n$ and eigenfunctions $\phi(x, \lambda_n)$ of a Sturm-Liouville boundary value problem: $$ \rho_n = \sqrt{\lambda_n} = n + \frac{\omega}{\pi n} +\frac{\varkappa_n}{n} $$ $$ \phi(x, \lambda_n) = \cos nx + \frac{\xi_n(x)}{n} $$ where $$ \omega = h + H + \frac{1}{2}\int_{0}^{\pi}q(t)dt; \quad \{\varkappa_n\} \in l_2; \quad |\xi_n(x)|\leq C \quad (C>0) $$ The boundary value problem itself is: $$ -y''+q(x)y=\lambda y \quad 0

Question: What is the meaning of and the logic behind the above-mentioned asymptotic formulas? Why are they useful?

  • 1
    You probably have an error in $\phi(x,\lambda_n)$ because $\lambda_n$ is missing on the right. The idea is that, for large $\lambda$, the function $q$ is overwhelmed by $\lambda$ in $-y''=(\lambda-q)y$ and the eigenfunctions/eigenvalues behave very much like those for the problem for $q=0$ with the same endpoint conditions. The pointwise convergence of the Fourier series in the new eigenfunctions is then related to classical cases of trigonometric expansions associated with $q=0$ (they're equivalent.)2017-01-15

0 Answers 0