Let $R$ be a non-type-I factor acting on a separable Hilbert space.

Let $P(R)$ be the set of $R$'s projections with the usual ordering ($x \leq y \iff$ range$(x) \subseteq$ range$(y)$) under which it forms a complete lattice; let $P^+(R)$ be the same excluding the null projection.

Let $C$ be the poset of finite sequences of ordinals $< 2^\omega$, ordered by reverse inclusion (note $C$ is the most common poset used in forcing to collapse $2^\omega$ to $\omega$).

Question: Must there exist an order-embedding $\phi : C \rightarrow P^+(R)$ whose range is order-dense in the sense relevant to forcing (namely for all $x \in P^+(R)$ there exists $c \in C$ such that $\phi(c) \leq x$)?

I'm not sure how much intrinsically operator-algebraic interest this question holds, but it is potentially interesting for set theory, in particular for forcing. If there is such a embedding then $P^+(R)$ is forcing-equivalent to the standard continuum-collapse forcing; if not, I suspect $P^+(R)$ would (at least for some types of factor $R$) be inequivalent to known forcing posets, and would presumably have some novel properties.

Note that there definitely exist order-embeddings of $C$ into $P^+(R)$ that map each unbounded descending chain in $C$ to an unbounded descending chain in $P^+(R)$, but this property by itself doesn't ensure the embedding is dense.


I believe I've shown (see answer to related stackoverflow question at https://math.stackexchange.com/a/4223869/250373 ) that there is such an embedding, on the supposition that every nontrivial lattice partition of $P(R)$ has continuum cardinality. This supposition is probably true in general but for now I can only prove this in the type III case, and with the assumption that CH holds.

[I'm deleting the sketch of a potential proof that had been in this answer previously, because it can't be salvaged.]


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.