1
$\begingroup$

Let $-\Delta: W^{2,2} \subset L^2(\mathbb{S}^2) \rightarrow L^2(\mathbb{S}^2)$. Then it is "easy" to show that $-\Delta $ is self-adjoint. Now, I am looking for closed operators $T$ and $T^*$ of order $1$ and $-1$ respectively such that we have $-\Delta = T^*T$. Notice, that the canonical square root by the functional calculus does not work, as this gives me again a self-adjoint operator of order 0.

$\endgroup$
7
  • 1
    $\begingroup$ The index of any selfadjoint operator is $0$. $\endgroup$ Jan 20, 2015 at 20:49
  • $\begingroup$ ah, you are right, now I see it. Do you know if there is a way to decompose the Laplacian on this sobolev space into operators that have an index 1? I have edited my question, cause this is essentially what I wanted to find out, just had the wrong idea. $\endgroup$
    – Ethan Co
    Jan 20, 2015 at 21:07
  • $\begingroup$ @LiviuNicolaescu I just played around with a decomposition $T^*T= -\Delta$ and I think I just showed that then in any case, we would have $ind(T)=0$. Now this is extremely strange, as this would mean that for ANY self-adjoint operator, the Witten index would vanish. $\endgroup$
    – Ethan Co
    Jan 20, 2015 at 21:39
  • $\begingroup$ I don't know what you meant by witten index. $\endgroup$ Jan 20, 2015 at 22:24
  • $\begingroup$ @LiviuNicolaescu actually, I still mean Fredholm index ( was talking physicists' language) $\endgroup$
    – Ethan Co
    Jan 20, 2015 at 22:44

1 Answer 1

1
$\begingroup$

Let $M$ be closed Riemannian manifold (e.g, $M={\mathbb S}^2$). Let $\{\varphi_k\}_{k\in\mathbb N_0}$ be an orthonormal basis for $L^2=L^2(M)$ consisting of eigenfunctions of $-\Delta$, where $-\,\Delta \varphi_k =\lambda_k^2\varphi_k$, $0=\lambda_0<\lambda_1\leq \lambda_2\leq \dots$ Then the operator $T\colon H^1\subset L^2 \to L^2$ defined by $$ T\varphi_k = \begin{cases} 0, & k=0, \\ \lambda_k\varphi_{k-1}, & k\geq1, \end{cases} $$ has the desired properties. Notice that $T^\ast \varphi_{k-1}=\lambda_k\varphi_k$ for $k\geq1$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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