I found the following assertion in my class notes, but i did not understood why doe's it holds:

"Suppose that in an infinite-dimensional von Neumann algebra there is an increasing net $\{p_j\}$ of projections with $p_j\nearrow 1$ and $p_j\ne 1$ for all $j$. Then the subspace $P=\overline{\operatorname{span}}\{p_j:\ j\}$ does not contain $1$."

It is a basic fact of von Neumann algebras that any bounded, totally ordered net of self-adjoints has a supremum, and said supremum is the sot (wot, etc) limit of the net. but I can't see why doesn't 1 belongs in P.

Thanks in advance!


Suppose $p_1,..., p_n$ are projections $≤p$ for some projection $p<1$. Then for any scalars $a_1,...,a_n$ you have:


The simplest way to see this is to consider a representation on some Hilbert space. Then since $p<1$ you have some $x\in\mathrm{im}(p)^\perp$, which is necessarily also in $\mathrm{im}(p_i)^\perp$ for all $i$ since $p_i≤p$. You then get:

$$(1-a_1p_1+...+a_np_n)(x)=x,$$ which implies the inequality above. Now in your situation if you have some linear combination of projections from the net there is a common supremum in the net of these projections. This supremum must $<1$ because $1$ is not in the net. Then the above inequality implies that the linear combination has distance $≥1$ from the operator $1$.

| cite | improve this answer | |

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.