2
$\begingroup$

I'm having a lot of trouble to picture the elements of the AFD (hyperfinite) $II_1$ von Neumann algebra. I would like to see concrete examples of operators and projections belonging to the hyperfinite $II_1$ factor $R$ when the when it's viewed as a subalgebra of $B(H)$ (assuming that this inclusion is possible).

For now, I'd like to make concrete the fact that $II_1$ algebras are diffuse, i.e. have no minimal projections. I'm trying to see how a projection $p>0$ could be decomposed in two other projections $p_1,p_2<p$ with $p=p_1+p_2$ and also how these projections could be approximated by the finite subalgebras.

When I try to follow the $II_1$ factor constructions I get lost in the GNS procedure. Also, when trying to use the $M_{2^n}$ construction, I'm not sure how the finite subalgebras belong to the hyperfinite factor. The naive visualization of finite algebras of type $I_{n}$ in $L(H)$ takes me to finite matrix algebras which do have minimal projections. I don't know where I'm making the mistakes.

I am overwhelmed by the loads of new concepts in the von neumann algebra theory.

I would highly appreciate any hints or references on how the operators and projections in the hyperfinite factor could be made explicity in some $B(H)$, perhaps operators in $\ell_2(\mathbb N)$.

Thanks in advance!

$\endgroup$
2
$\begingroup$

The inclusion into $B(H)$ for some $H$ is always possible. Generally one defines a vN-algebra as a self-adjoint subalgebra of $B(H)$ which is WOT-closed. Alternatively, vN-algebras are C*-algebras, and these can be embedded into $B(H)$ via the GNS construction.

The fact that a $II_1$ factor is diffuse follows from the fact that a factor $M$ has a minimal projection if and only if $M \simeq B(H)$ for some Hilbert space $H$. I'll leave this as a fact, but feel free to ask about clarification about the proof.

Clearly any $II_1$ factor has a (faithful) tracial state and is infinite dimensional, so it cannot be $B(H)$ ($B(H)$ has no tracial state when $H$ is infinite-dimensional; in fact, not even the compacts do). So it has to be diffuse. Consider $A = M_{2^{\infty}}$ (the direct limit $\underset{\to}{\lim} M_{2^n}$ with connecting maps $a \mapsto a \otimes 1$). This is a uniformly hyperfinite C*-algebra, and has a unique faithful trace $\tau$. The hyperfinite $II_1$ factor comes from taking the GNS representation with respect to $\tau$: $M = \pi_\tau(A)'' = \overline{\pi_\tau(A)}^{\text{SOT}} \subseteq B(L^2(A,\tau))$, where $(a\xi,b\xi) = \tau(b^*a)$ for $a,b \in A$, and $\xi$ is a unit vector which is separating and cyclic. The trace on $M$ is given by $a \mapsto (a\xi,\xi)$.

I'll give an explicit example of a projection in $A \subseteq M$ which we can decompose in the way you want. $A$ can be thought of as the infinite tensor product of $M_2$. So $A = \otimes_1^{\infty}M_2$, which is really the norm closure of $\cup_n (M_2^{\otimes n}\otimes 1\otimes\cdots)$. Let $$ p = \begin{pmatrix} 1 & 0 \\ 0 & 0 \end{pmatrix} \otimes 1 \otimes\cdots. $$ We can write this projection in the following way. Let $$p_1 = \begin{pmatrix} 1 & 0 \\ 0 & 0 \end{pmatrix} \otimes \begin{pmatrix} 1 & 0 \\ 0 & 0 \end{pmatrix} \otimes 1 \otimes\cdots $$ and $$p_2 = \begin{pmatrix} 1 & 0 \\ 0 & 0 \end{pmatrix} \otimes \begin{pmatrix} 0 & 0 \\ 0 & 1 \end{pmatrix} \otimes 1 \otimes\cdots. $$ Now its clear that $p_1,p_2 < p$ and that $p = p_1 + p_2$. Edit: Note that you can keep on doing this, so these projections can't be minimal!

| cite | improve this answer | |
$\endgroup$
2
$\begingroup$

This is a "less naive" way of seeing the chain $M_2(\mathbb C)\subset M_4(\mathbb C)\subset\cdots\subset B(H)$ for infinite-dimensional separable $H$ (and it is was was done in PStheman's answer, just a bit more explicit here).

You see $M_2(\mathbb C)$ as $$ \begin{bmatrix} a&b\\ c&d\\ &&a&b\\ &&c&d\\ &&&&a&b\\ &&&&c&d\\ &&&&&&a&b\\ &&&&&&c&d\\ &&&&&&&&\ddots \end{bmatrix}, $$ then $M_4(\mathbb C)$ as $$ \begin{bmatrix} a_{11}&a_{12}&a_{13}&a_{14}\\ a_{21}&a_{22}&a_{23}&a_{24}\\ a_{31}&a_{32}&a_{33}&a_{34}\\ a_{41}&a_{42}&a_{43}&a_{44}\\ &&&&a_{11}&a_{12}&a_{13}&a_{14}\\ &&&&a_{21}&a_{22}&a_{23}&a_{24}\\ &&&&a_{31}&a_{32}&a_{33}&a_{34}\\ &&&&a_{41}&a_{42}&a_{43}&a_{44}\\ &&&&&&&&\ddots \end{bmatrix}. $$ So for example take $E_{11}^{(2)}\in M_2(\mathbb C)$, and let's find subprojections of it: $$ E_{11}^{(2)}=\begin{bmatrix} 1\\ &0\\ &&1\\ &&&0\\ &&&&1\\ &&&&&0\\ &&&&&&1\\ &&&&&&&0\\ &&&&&&&&\ddots \end{bmatrix}. $$ Now you can see that $E_{11}{(4)}$ is a subprojection of $E_{11}^{(2)}$: $$ E_{11}^{(4)}=\begin{bmatrix} 1\\ &0\\ &&0\\ &&&0\\ &&&&1\\ &&&&&0\\ &&&&&&0\\ &&&&&&&0\\ &&&&&&&&\ddots \end{bmatrix}. $$ Continuing this way you can get the proper chain of projections $$ E_{11}^{(2)}\geq E_{11}^{(4)}\geq E_{11}^{(8)}\geq\cdots $$

| cite | improve this answer | |
$\endgroup$

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.