I am trying to understand Araki's proof of the statement that the restricted orthogonal group of a Hilbert space with a unitary structure is simply connected. This proof starts on page 114 of these conference proceedings (the result is stated on page 77, Theorem 6.3 (4)).

I know that there are some alternative proofs of this statement, but I am specifically interested in this version. I will start with some prerequisites, which may or may not be necessary.

Let $V$ be a complex Hilbert space equipped with a real structure, that is, a complex anti-linear involution $\alpha: V \rightarrow V$. Then we define a complex bilinear form \begin{equation} \beta: V \times V \rightarrow \mathbb{C}, (v,w) \mapsto \langle v, \alpha (w) \rangle. \end{equation} Let $L$ be a Lagrangian subspace with respect to $\beta$. Then we define a so-called unitary structure on $V = L \oplus \alpha(L)$ by \begin{align} J:L \oplus \alpha(L) &\rightarrow L \oplus \alpha(L) \\ (v,w) &\mapsto i(v,-w). \end{align} The restricted orthogonal group is then defined to be \begin{equation} \operatorname{O}_{J}(V) = \{ U \in \operatorname{U}(V) \mid \| [U,J] \|_{HS} < \infty \}, \end{equation} where $\| \cdot \|_{HS}$ is the Hilbert-Schmidt norm. The restricted orthogonal group is a topological (actually Banach-Lie) group, when equipped with the so-called $J$-norm, \begin{equation} \|U \|_{J} = \|U\| + \|[U,J]\|_{HS}, \end{equation} where $\| \cdot \|$ is the usual operator norm. Let $P_{L}$ be the projection onto $L$. The condition that $\|[U,J]\|_{HS} < \infty$ is then equivalent to saying that $\|[U,P_{L}]\|_{HS} < \infty$.

Now, the claim is that (the identity component of) $\operatorname{O}_{J}(V)$ is simply connected. Araki proves this by starting with an arbitrary loop and arguing that it is contractible in $\operatorname{O}_{J}(V)$.

So let $U_{t}$, $0, \leqslant t \leqslant 1$ be a loop in $\operatorname{O}_{J}(V)$. Then we define $P(t) = U_{t}P_{L} U_{t}^{*}$. Araki then claims that $P$ is continuous with respect to the Hilbert-Schmidt norm.

Q1 Why is this so?

My problem here is that it is not clear to me that $U_{t}P_{L}U_{t}^{*}$ has finite Hilbert-Schmidt norm. Because of the fact that the Hilbert-Schmidt operators form an ideal we see that $[U_{t},P_{L}] U_{t}^{*} = U_{t} P_{L} U_{t}^{*} - P_{L}$ is Hilbert-Schmidt. Maybe it is meant that the map $t\mapsto U_{t} P_{L} U_{t}^{*} - P_{L}$ is continuous with respect to the Hilbert-Schmidt norm?

Next, we split the circle into a finite number of intervals $[t_{j},t_{j+1}], j = 1,...,n$ such that $\| P(t_{j}) - P(t) \| <1$. All we require here is that $P$ is norm-continuous.

For any $t \in [t_{j},t_{j+1}]$ we can find an operator $H(t)$ with finite Hilbert-Schmidt norm, such that \begin{equation} e^{H(t)}P(t_{j})e^{-H(t)} = P(t). \end{equation} The fact that $H(t)$ has finite Hilbert-Schmidt norm implies that $H(t) \in \operatorname{O}_{J}(H)$.

Q2 Why does such $H(t)$ exist? Presumably it uses the condition $\|P(t_{j}) - P(t) \| <1$.

We then define a path $W: [0,1] \rightarrow \operatorname{U}(V)$ by \begin{equation} W(t) = e^{H(t)} e^{H(t_{j})} ... e^{H(t_{2})}, \text{ for } t \in [t_{j},t_{j+1}]. \end{equation} I guess that it is somehow clear that $H(t_{j} + \epsilon)$ can be made arbitrarily close to $0$. Then we use the relation \begin{equation} e^{H(t_{j+1})} P(t_{j}) = P(t_{j+1}) e^{H(t_{j+})}, \end{equation} to prove that $W(1)$ commutes with $P_{L}$.

Next, Araki claims that $\|W(1) - 1\|_{HS} < \infty$, which implies that there exists an operantor $H = -H^{*}$ such that $P_{L}H = HP_{L} = H$, and furthermore $e^{H}P = W(1)P$ and finally $\|H\|_{HS} < \infty$.

Q3 Why does this $H$ exist?

For this step I am really lost.

The proof doesn't end here, but I think these are already way too many questions.


Your Answer

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

Browse other questions tagged or ask your own question.