Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $(A,H)$ be a von Neumann algebra in standard form (which means that $H=L^2A$), and recall that the automorphism group $Aut(A)$ acts on both $A$ and $H$. Let $$N:=\{u\in U(H): uAu^*=A\}$$ be the normalizer of $A$, equipped with the strong topoloy (the subspace topology from $U(H)$). The group $N$ is canonically isomorphic to the semidirect product $$N\cong Aut(A)\ltimes U(A').$$

Equip $Aut(A)$ with Haagerup's u-topology, i.e. the topology of pointwise convergence on $A_*$, equivlalently, it is the topology of pointwise convergence on $H$.

Question: is the projection map $$N\to Aut(A)$$ continuous?

share|improve this question

2 Answers 2

up vote 4 down vote accepted

Julien's comment to my previous answer leads to an even easier solution.

Consider the isomorphism map $\psi:Aut(A)\ltimes U(A^\prime)\rightarrow N$. The $u$-topology on $Aut(A)$ is the topology of pointwise convergence on $H$, which is the same as strong convergence when we consider the elements of $Aut(A)$ as unitaries on $H$. Since $N$ is a topological group, it follows that $\psi$ is continuous. Julien's comment shows that it is in fact a homeomorphism. So certainly the quotient $N\rightarrow Aut(A)$ is continuous.

share|improve this answer
Wonderful. This is very elegant! –  André Henriques Feb 7 '12 at 17:21

The map is continuous, see the argument below. This also means that the group isomorphism is a homeomorphism.

Let $(u_n)_n$ be a sequence in $N$ that converges to $1$ in strong operator topology. Then we have to show that $\varphi(u_n^\ast\cdot u_n)$ converges to $\varphi$ for all $\varphi\in A_\ast$.

We can assume that $\varphi(x)=\langle\xi,x\eta\rangle$ for some $\xi,\eta\in L^2(A)$. But we know that $u_n\xi\rightarrow\xi$ and $u_n\eta\rightarrow\eta$. So $\langle u_n\xi, xu_n\eta\rangle$ converges to $\langle\xi,x\eta\rangle$ uniformly over $x\in (A)_1$. In other words, the functionals $\varphi(u_n^\ast\cdot u_n)=\langle u_n\xi, \cdot u_n\eta\rangle$ converge to $\langle\xi,\cdot\eta\rangle=\varphi$ in $A_\ast$.

share|improve this answer
To see that this is a homeomorphism don't you need to also show that the inverse is continuous? Is there an automatic continuity result for Polish groups like there is with locally compact groups? –  Jesse Peterson Feb 3 '12 at 17:08
@Jesse: Yes, a continuous, bijective group isomorphism between two Polish groups is necessarily a homeomorphism. This follows for instance from Pettis' theorem (I think the first to prove that result was Banach). –  Julien Melleray Feb 3 '12 at 19:02
@Julien: Thanks. That's useful to know. –  Jesse Peterson Feb 4 '12 at 0:34
@julien: thank you, this was a result i did not know, and it actually makes my argument redundant. To me it seems that the continuity of the inverse is obvious. –  Steven Deprez Feb 6 '12 at 13:32

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.