Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Suppose $A$ and $B$ are linear compact operators on a Hilbert space with $\sigma(A)$ and $\sigma(B)$ as their spectrum.

Is it possible to obtain some continuity result of $\sigma(A+\epsilon B)$ as $\epsilon\downarrow 0$ towards $\sigma(A)$? Is the limiting behavior of the form "$\sigma(A)+\epsilon \sigma(B)$''?

Thanks in advance!

share|improve this question
add comment

2 Answers 2

up vote 1 down vote accepted

The first question has an answer and is "yes", see Dunford-Schwartz, Linear operators, part II, XI-9, Lemma 5; it reads:

"Let $A_n, A$ be compact operators and $A_n\rightarrow A$ in the uniform operator topology. Let $\lambda_m(A)$ be an enumeration of the non-zero eigenvalues of $A$, each repeated according to its multiplicity. Then, there exist enumerations $\lambda_m(A_n)$ of the non-zero eigenvalues of $A_n$, with repetitions according to multiplicity, such that $$\lim_{n\rightarrow\infty} \lambda_m(A_n)=\lambda_m(A), \quad m\geq 1,$$ the limit being uniform in $m$."

So, under compact perturbations $\epsilon B$ of a compact operator $A$, the perturbed spectrum moves continuously. I don't know about your second question, the order of convergence might be a little bit tricky to obtain, some Rellich result might be useful. Consider checking Kato's book "Perturbation Theory for Linear Operators".

share|improve this answer
add comment

A partial answer. I think it is true if $\sigma(A)=\{0\}$. For $\lambda\neq 0$, $A-\lambda I$ is invertible. However, the set of isomorphisms is open in $\mathcal{B}(H)$, so there exist $\varepsilon>0$ such that $A-\lambda I+\varepsilon B$ is also invertible. This means that $\lambda\notin\sigma(A+\varepsilon B)$, for $\varepsilon$ small enough. This should imply that $\sigma(A+\varepsilon B)\to 0$.

share|improve this answer
add comment

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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