I am reading the following theorem from my lecture notes (English translation of German text). But I don't understand exactly what is meant from this theorem and the proof.


Let $H$ be a Hilbert space and $T \in K(H)$ where $K(H)$ denotes the set of compact operators from $H$ to $H$. Let $T$ be normal or self adjoint depending on whether the underlying field is $\mathbb R$ or $\mathbb C$, then there exists an orthonormal set $\{e_i \mathrel| i\in I\}$ where $I$ is either $\mathbb N$ or $\{1,2,\dotsc,k\}$ and a sequence $(\lambda_i)_{i\in I} \in \mathbb K$ which converges to $0$ such that

$$\operatorname{span}{(x_i : i ∈ I})^{⊥} = \ker(T)$$

and also $\forall x \in H$ : $Tx= \sum_{i\in I} \lambda_i \langle x, e_i \rangle e_i$ with unconditional convergence.

I am sorry here because I don't understand the theorem completely which means the translation is wrong, but can someone point out this theorem in some textbook or somewhere.


  • $\begingroup$ @GitGud : what do you mean ? Actually i am looking for the proof of the theorem ? $\endgroup$ – Theorem Mar 30 '13 at 14:02
  • $\begingroup$ @Theorem Why do you assume the translation is wrong? I see only very minor typos ($x_i$ vs. $e_i$, and $\mathbb N \cup \{1,2,..... k\}$ is an incorrect way of saying that $I$ is a countable index set). $\endgroup$ – Erick Wong Mar 30 '13 at 14:07
  • $\begingroup$ @ErickWong : it should be $e_i$ . this seems to be a typo on the script also . thanks $\endgroup$ – Theorem Mar 30 '13 at 14:10

This is about the spectral decomposition of compact operators.

According to the theorem, a (normal / self-adjoint) compact operator can have countably many different eigenspaces, each of finite dimension, which are mutually orthogonal (due to self-adjointness / being normal). Let the eigenvectors be denoted by $e_1,e_2,e_3,...$ and the corresponding eigenvalues $\lambda_1,\lambda_2,..$ Moreover, if it has infinitely many eigenvalues, then these must tend to $0$.

Additionally, the operator must vanish outside the span of its eigenvectors.


I guess the most important thing to realize here is that compact operators have only point spectrum formed of eigenvalues, which tend to 0 (and 0 is in the spectrum as well). Then we can write the operator as a countable sum over projections multiplied by eigenvalues (in such way that the eigenvalues tend to 0) opposed to the need of writing some operators as integrals over spectral measures.


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.