0
$\begingroup$

I have Sturm-Liouville problem: $x^2u''+xu'+\lambda u=0$

With conditions

$1<x<e^{\pi}$

$u'(1)=0$

$u'({e^\pi})=0$

I wrote this SL problem is self adjoint form:

$\frac{d}{dx}[x \frac{du}{dx}]=-\lambda w(x)u$, where $w(x)=\frac{1}{x}$

Now I have to show on the given interval and subject to given boundary conditions, operator $L$ is Hermitian and find eigenvalues and eigenfunction to verify that the eigenfunctions are orthogonal on the interval $1<x<e^{\pi}$. At last, I have to find expansion function $\delta(x-e)(1<x<e^{\pi})$ in a series of eigenfunctions. I don't know anyting abou expansion function. Also i cannot show orthogonality and operator is Hermitian. How can I do that?

$\endgroup$
1
$\begingroup$

I won't do the entire problem, but I'll provide some help. First, you need to write your equation in the form of an eigenvalue problem, that is $Lu=\lambda u$. To do this, divide your equation by $-w(x)$, so $$ Lu=\frac{-1}{w(x)}\left[\frac{d}{dx}\left(x\frac{du}{dx}\right)\right] =\lambda u.$$ So you now have your $L$, which should be viewed as an operator in a Hilbert space with inner product $$(\psi, \phi)= \int_{1}^{e^{\pi}}\psi^{*}(x)\phi(x) \ w(x)dx,$$ where $\phi(x)$ and $\psi(x)$ are smooth functions satisfying the boundary conditions of your problem. You wanna show $L$ is hermitian, meaning $(L\psi, \phi)=(\psi, L\phi)$ for arbitrary functions of the aforementioned Hilbert space. Indeed you can use integration by parts twice to show \begin{align}(L\psi, \phi) &= -\int_{1}^{e^\pi}\frac{d}{dx}\left(x\frac{d\psi^*}{dx}\right)\phi dx=\int_{1}^{e^\pi}\left(x\frac{d\psi^*}{dx}\right)\frac{d\phi}{dx} dx \\&=\int_{1}^{e^\pi}\psi^*\left[-\frac{1}{w(x)}\frac{d}{dx}\left(x\frac{d\phi}{dx} \right)\right] w(x)dx =(\psi,L\phi), \end{align} where the boundary terms vanish by virtue of the boundary conditions, and in the last step I multiplied and divided by $w(x)$. Thus, $L$ is hermitian. To verify the eigenfunctions are orthogonal you are gonna have to solve this differential equation. You should then find a set of permissible $\lambda$ (this equation may not be solvable for every possible $\lambda$), and a related solution (the eigenfunction) for each $\lambda$. When you obtain this set of functions (let's call it $\{\phi_n \}$), you must verify that $(\phi_n,\phi_m)=0$ for $n\neq m$, and that $(\phi_n,\phi_n)= c^2$ for some number $c$.

$\endgroup$
1
  • $\begingroup$ Tomorrow I'll try these. Thanks a lot! $\endgroup$ May 23 at 0:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.