MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Suppose $H$ is a Hilbert space, $B(H)$ is the algebra of bounded linear operators on it, $K(H)$ is ideal of compact operators in $B(H)$, $Inv(B(H)/K(H))$ is the topological group of invertible operators in $B(H)/K(H)$, $Inv(B(H)/K(H))_0$ --- connected component of $id$ in $Inv(B(H)/K(H))$. $ind\colon Inv(B(H)/K(H))\to \mathbb{Z}$ --- Fredholm index.

I want to find a reference for the following fact:

Fact 1. If $H$ is infinite-dimensional and separable then $ind$ is locally-constant and provides an isomorphism between $Inv(B(H)/K(H))/Inv(B(H)/K(H))_0$ and $\mathbb{Z}$.

In Murphy's textbook, where I've read almost all I know about Fredholm index, there is Atkinson's theorem, the fact, that Fredholm index is locally constant and $ind(ab)=ind(a)+ind(b)$. But in order to check the fact 1, I need also the fact, stated in the head of the question, or, equivalently, the following:

Fact 2. If $H$ is infinite-dimensional and separable then the set $\{a\in B(H)/K(H)\mid ind(a)=0\}$ is connected.

Where can I find the references? Where can I read that?

share|cite|improve this question
The proof of 1 as e.g. in the quoted paper by Kuiper is really elementary fact; it's a linear algebra version of Hilbert's Grand Hotel argument (aka Hotel California). As to 2, recall that any Fredholm operator $T:X\to Y$ of can be written as a direct sum $0\oplus T_1: X_0\oplus X_1\to Y_0\oplus Y_1$, with $T_1:X_1\to Y_1$ invertible, $X_0$ and $Y_0$ finite dimensional, $\mathrm{ind}(T)=\dim(X_0)-\dim(Y_0)$; if $\mathrm{ind}(T)=0$ there is an isomorphism $A:X_0\to Y_0$ that produces a path $[0,1]\ni s \mapsto (sA)\oplus T_1$ from $T$ to an invertible operator. – Pietro Majer Sep 27 '10 at 15:27
@ Pietro Majer: You have only prooved, that any connected component of $Inv(B(H)/K(H))$ with 0 Fredholm index contains an invertible operator. Fact 2 doesn't follow from this. – Fiktor Sep 27 '10 at 19:07
up vote 4 down vote accepted

You can use some spectral theory to show that the set of unitary operators in $B(H)$ is path-connected. Then the (path-)connectedness of the invertibles follows easily from the polar decomposition. See Theorem 5.29 and Corollary 5.30 in Douglas's Banach Algebra Techniques (1st edition).

share|cite|improve this answer
Thank you, this is very good reference, because 1. it is a good-written book, 2. I'm already referencing it. – Fiktor Sep 27 '10 at 16:32

Kuiper's theorem gives the contractibility of the invertible operators in $B(H)$. Start with's_theorem

Check out papers of Boris Mitjagin [Mitiagin] in the 1970s for more information.

EDIT 9/27: In "A course in functional analysis" (Springer TTM 96), Conway proves in Theorem XI.4.1 that the components of the semi-Fredhlom operators on $\ell_2$ are determined by the index.

share|cite|improve this answer

I recommend the paper by P. de la Harpe, {\it Initiation `a l'alg`ebre de Calkin}, pp. 180-219 in Alg`ebres d'Op\'erateurs, Springer Lect. Notes in Math. 725, 1979; all these facts (and much more) are available there.

share|cite|improve this answer

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.