this is a little bit of a dumb question, please be nice, I had some doubts about the type decomposition of von Neumann algebras.

I was reading Bruce Blackadar's "Operator algebras. Theory of C*-algebras and von Neumann algebras" and on page 246 he says that a von Neumann algebra $M$ can be decomposed as follows:

$M = Mz_I \oplus Mz_{II_1} \oplus Mz_{II_{\infty}} \oplus Mz_{III}$,

where $Mz_I = z_{I}Mz_{I}$, $Mz_{II_1} = z_{II_1}Mz_{II_1}$, $Mz_{II_\infty} = z_{II_\infty}Mz_{II_\infty}$ and $Mz_{III} = z_{III}Mz_{III}$.

Here if $\{ p_i\}$ is a family of abelian projections which is maximal with respect to the condition that the central carriers $z_{p_i}$ are pairwise orthogonal, then $z_I = \sum z_{p_i}$, with

$z_I = \bigvee \{p:$ p is a discrete projection$\} = \bigvee \{p:$ p is abelian$\}$,

and $z_{II} = 1 - z_{I} - z_{III}$ (there are other formulas for $z_{II_1}$ and $z_{II_\infty}$).

A von Neumann algebra is said to be of pure type $I$ if $z_{I} = 1$, with similar definitions for types $II_1$, $II_\infty$, and $III$.

I wanna make sure I'm understanding this right:

This means $Mz_{I}$, $Mz_{II_1}$, $Mz_{II_\infty}$ and $Mz_{III}$ are themselves von Neumann algebras of pure type?

That means that for example $Mz_{III}$, as a von Neumann algebra, has all the respective projections (let's call them $j_{I}$, $j_{II_1}$, $j_{II_{\infty}}$ and $j_{III}$), where $j_{III} = 1$ (which in this case $1 = z_{III}$ in $Mz_{III}$?), and $j_{I} = 0$, $j_{II_1} = 0$, $j_{II_{\infty}} = 0$?? Am I making sense here? If this is true or not true it is not obvious to me or just going over my head and I don't want there to be any room for confusion.


Yes. Those central projections exist for any von Neumann algebra. When the algebra is of a certain type, it means that the central projection corresponding to that type is $1$ and the ones corresponding to the other types are zero.

  • $\begingroup$ Thank you! Don't know if "type decomposition" is the correct term, I kept hearing and being referred to things like "The Murray-von Neumann type decomposition" and stuff but couldn't find the original source. $\endgroup$ – The K Jan 11 '14 at 1:32
  • $\begingroup$ You are welcome! $\endgroup$ – Martin Argerami Jan 11 '14 at 4:20

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.