The strong degree of von Neumann algebras and the structure of Lie and Jordan derivations (Q4824775)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 2113305
Language Label Description Also known as
default for all languages
No label defined
    English
    The strong degree of von Neumann algebras and the structure of Lie and Jordan derivations
    scientific article; zbMATH DE number 2113305

      Statements

      The strong degree of von Neumann algebras and the structure of Lie and Jordan derivations (English)
      0 references
      0 references
      0 references
      1 November 2004
      0 references
      Lie derivation
      0 references
      Jordan derivation
      0 references
      von Neumann algebra
      0 references
      strong degree
      0 references
      functional identity
      0 references
      Let \(A\) be an (associative) algebra, and let \(M\) be an \(A\)-bimodule. There are canonical notions of Lie and of Jordan derivations from \(A\) into \(M\), and a natural question is how much they can differ from associative derivations, that is, linear mappings \(d\colon A\to M\) with the property \(d(ab)=da\cdot b+a\cdot db\), \(a,b\in A\). (Here ``\(\,\cdot\,\)'' denotes the bimodule action.) The paper under review discusses this question in particular when \(A\) is a von Neumann algebra and \(M\) is a Banach \(A\)-bimodule. Since the results for Jordan derivations are at least as good as for Lie derivations, but in general easier to obtain, we shall exclusively focus on the Lie case. (It turns out that, in the cases considered, Jordan derivations are simply ordinary derivations.) A linear mapping \(D\colon A\to M\) is said to be a Lie derivation provided that \(D([a,b])=[Da,b]+[a,Db]\), \(a,b\in A\), where \([\,\cdot,\cdot]\) stands for the Lie bracket (in both \(A\) and \(M\)). Besides associative derivations, every linear mapping \(\tau\colon A\to M\) which takes its values in the centre of \(M\) and such that \(\tau([a,b])=0\) for all \(a,b\in A\) is a Lie derivation. The Lie derivation \(D\) is said to be in standard form if it can be written as \(D=d+\tau\) for some derivation \(d\) and some \(\tau\) of the above kind (such a \(\tau\) is called a centre-valued trace). \textit{B. E. Johnson} showed in [Math. Proc. Camb. Philos. Soc. 120, 455--473 (1996; Zbl 0888.46024)] that every continuous Lie derivation from a \textsl{C*}-algebra \(A\) into a Banach \(A\)-bimodule is in standard form. Johnson achieved this by introducing a new concept of amenability (called symmetric amenability). The main result of the paper under review, Theorem~6.10, extends Johnson's result to not necessarily continuous Lie derivations on von Neumann algebras. However, the efforts needed to do so are considerable. The argument starts with a reduction to unital modules (\(1\cdot x=x\cdot1=x\) for all \(x\in M\)), section~6.1, and a reduction to central summands, section 6.2. It turns out that finite type I von Neumann algebras have to be treated separately (section 6.3), in which the commutative case (which takes up the whole section 5 by its own) plays a central part. In fact, it suffices to rule out the case of type I\(_n\) summands with \(n\leq3\), since for all other cases the strong degree of the von Neumann algebra is sufficiently large to imply that every Lie derivation is in standard form (Lemma 3.12, Corollary 4.2 and Corollary 6.2). It is here that algebraic methods and, in particular, the concept of the strong degree come in. Rather than defining this somewhat elaborate concept, let us point out that a high enough strong degree allows to solve certain functional identities in a canonical way. In the situation under consideration, the functional identity is of the form \[ [D(x^2)-Dx\cdot x-x\cdot Dx,x]=0\qquad(x\in A), \] which follows directly from the defining property of the Lie derivation~\(D\). The idea of using such functional identities goes back to work by \textit{M. Brešar} [Trans. Am. Math. Soc. 335, 525--546 (1993; Zbl 0791.16028)] and has been turned into a machinery using a certain kind of ``freeness'' by Beidar and Chebotar in a series of papers. We only remark here that a properly infinite von Neumann algebra \(A\) has infinite strong degree (Corollary 3.9) and hence Lie derivations on \(A\) can be handled by these methods in this case. In fact, since a properly infinite von Neumann algebra is linearly spanned by commutators, it turns out that every Lie derivation from such von Neumann algebra is indeed an associative derivation (Corollary 6.9). It is clear from the arguments, though not explicitly stated, that the last mentioned result remains valid for properly infinite \(C^*\)-algebras. For arbitrary \(C^*\)-algebras the question whether every Lie derivation into a Banach bimodule has to be in standard form remains however open. (It is unfortunate that, in the statement of Theorem 6.11 which recovers Johnson's result from 1996 by the methods here, the assumption of continuity got lost.) In this context, it is worth mentioning that Lie derivations on symmetrically amenable Banach algebras were treated in a paper by \textit{J. Alaminos, A. R. Villena} and the reviewer in [Math. Proc. Camb. Philos. Soc. 137, 433--439 (2004; Zbl 1063.46033)] and in an article by \textit{A. R. Villena} and the reviewer it was shown that every Lie derivation from a \(C^*\)-algebra \(A\) into \(A\) is in standard form [J. Funct. Anal. 202, 504--525 (2003; Zbl 1032.46086)].
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references