Lie ideals in properly infinite C*-algebras

Abstract.

We show that every Lie ideal in a unital, properly infinite $C^{*}$-algebra is commutator equivalent to a unique two-sided ideal. It follows that the Lie ideal structure of such a $C^{*}$-algebra is concisely encoded by its lattice of two-sided ideals. This answers a question of Robert in this setting.

We obtain similar structure results for Lie ideals in unital, real rank zero $C^{*}$-algebras without characters. As an application, we show that every Lie ideal in a von Neumann algebra is related to a unique two-sided ideal, which solves a problem of Brešar, Kissin, and Shulman.

1. Introduction

A Lie ideal in a $C^{*}$-algebra $A$ is a linear subspace $L\subseteq A$ such that for every $a\in A$ and $x\in L$ the commutator $[a,x]:=ax-xa$ belongs to $L$. The study of Lie ideals in $C^{*}$-algebras has a long history, and we refer to Section 2 in [Mar10] for an overview of available results and directions of active research.

Immediate examples of Lie ideals in a $C^{*}$-algebra $A$ are all subspaces contained in the center $Z(A)$ and all subspaces containing the commutator subspace $[A,A]$. If $A$ is unital and simple, then this does in fact describes all Lie ideals – a consequence of Herstein’s description of Lie ideals in simple rings [Her55]. Since the center of a unital, simple $C^{*}$-algebra $A$ contains only scalar multiples of the identity, a subspace $L$ is a Lie ideal in $A$ if and only if $L=\{0\}$, $L={\mathbb{C}}1$, or $[A,A]\subseteq L$.

Pop [Pop02] showed that a unital $C^{*}$-algebra satisfies $A=[A,A]$ if and only if $A$ has no tracial states. Consequently, a simple, unital $C^{*}$-algebra $A$ without tracial states has exactly three Lie ideals: $\{0\}$, ${\mathbb{C}}1$ and $A$. Cuntz and Pedersen [CP79] gave a description of the closure $\overline{[A,A]}$ as the intersection of the kernels of all tracial states on $A$. In particular, a simple, unital $C^{*}$-algebra $A$ with a unique tracial state has exactly four closed Lie ideals: $\{0\}$, ${\mathbb{C}}1$, $\overline{[A,A]}$ and $A$. This was first observed by Marcoux and Murphy [MM98], generalizing the case of UHF-algebras, which was first proved by Marcoux [Mar95].

Ng and Robert [NR16] showed that $[A,A]$ is closed in every pure, exact $C^{*}$-algebra. It follows that a simple, unital, exact, pure $C^{*}$-algebra $A$ with a unique tracial state has exactly four Lie ideals: $\{0\}$, ${\mathbb{C}}1$, $[A,A]$ and $A$. This applies for example to UHF-algebras, to irrational rotation algebras, and to the Jiang-Su algebra (where it was first shown by Ng [Ng12]). It also applies to many, and possibly all, simple, reduced group $C^{*}$-algebras. Indeed, if $G$ is a countable, exact, acylindrically hyperbolic group with trivial finite radical and with the rapid decay property, then $C^{*}_{\mathrm{red}}(G)$ has strict comparison [AGEP24], and therefore is pure [APTV24], and has a unique tracial state [BKKO17].

Turning towards Lie ideals in non-simple $C^{*}$-algebras, it is clear that the structure of the lattice of two-sided ideals will play a role. A systematic study of the interplay between Lie and two-sided ideals was initiated by Brešar, Kissin, and Shulman [BKS08] We first introduce the necessary notation. Following the usual convention, given subspaces $M,N\subseteq A$, we use $[M,N]$ to denote the subspace of $A$ generated by the set of commutators $[x,y]$ for $x\in M$ and $y\in N$. Further, given a subspace $K\subseteq A$, we set

$$T(K):=\big{\{}a\in A:[A,a]\subseteq K\big{\}}.$$

This ‘derived subspace’ was first considered by Herstein [Her55], and it is denoted by $N(K)$ in [BKS08].

Given a two-sided ideal $I\subseteq A$, Brešar, Kissin, and Shulman observed that a subspace $L\subseteq A$ is a Lie ideal whenever

$$[A,I]\subseteq L\subseteq T([A,I]).$$

We say that a Lie ideal $L$ is embraced by $I$ if it satisfies these inclusions. The Lie ideals that are embraced by some two-sided ideal are thought of as ‘unsurprising’. A major question is if there exist any Lie ideals that do not arise this way:

Problem (Brešar, Kissin, and Shulman [BKS08, p. 74]).

Given a $C^{*}$-algebra $A$, is every Lie ideal of $A$ embraced by some two-sided ideal of $A$?

To study this problem, two further notions were introduced in [BKS08]: Given a Lie ideal $L\subseteq A$ and a two-sided ideal $I\subseteq A$, we say that $L$ is commutator equivalent to $I$ if

$$[A,I]=[A,L].$$

Further, we say that $L$ is related to $I$ if

$$[A,I]\subseteq L\subseteq T(I).$$

It is easy to see that these constitute a sufficient and a necessary conditions for embracement. Indeed, if $L$ is commutator equivalent to $I$, then $L$ is embraced by $I$. Further, if $L$ is embraced by $I$, then $L$ is related to $I$.

It was shown in [BKS08, Theorem 5.27] that every Lie ideal $L$ in a $C^{*}$-algebra $A$ is topologically commutator equivalent to a two-sided ideal $I$ in the sense that $[A,L]$ and $[A,I]$ have the same closure. This leads to a description of all closed Lie ideals in terms of closed two-sided ideals. However, this does not show that all closed Lie ideals in a $C^{*}$-algebra are commutator equivalent to (let alone embraced by or related to) a two-sided ideal. For general non-closed Lie ideals even less is known, and Section 1 as well as the following stronger problem remain open:

Problem (Robert [Rob16, Question 1.12]).

Given a $C^{*}$-algebra $A$, is every Lie ideal of $A$ commutator equivalent to some two-sided ideal of $A$?

The results mentioned at the beginning of the introduction show that every Lie ideal in a unital, simple $C^{*}$-algebra $A$ is commutator equivalent to a two-sided ideal, which gives positive solutions to Problems 1 and 1 in this setting. Indeed, the only two-sided ideals are $\{0\}$ and $A$, and a Lie ideal is commutator equivalent to $\{0\}$ if and only if it is contained in $Z(A)$, while a Lie ideal it is commutator equivalent to $A$ if and only if it contains $[A,A]$. (This uses that $[A,A]=[A,[A,A]]$, which follows for example from [GT24, Corollary 3.4].)

In [FMS82], Fong, Miers and Sourour showed that every Lie ideal in the von Neumann algebra of bounded, linear operators on a separable Hilbert space is embraced by a two-sided ideal. This was substantially extended by Brešar, Kissin, and Shulman, who showed in [BKS08, Theorem 5.19] that every Lie ideal in a von Neumann algebra is commutator equivalent to a two-sided ideal, thereby providing positive solutions to Problems 1 and 1 for von Neumann algebras.

Our main result is a positive solution to Section 1 for unital, properly infinite $C^{*}$-algebras.

Theorem (5).

Let $L$ be a Lie ideal in a unital, properly infinite $C^{*}$-algebra $A$. Then $L$ is commutator equivalent to the two-sided ideal $I:=A[A,L]A$, and hence also embraced by $I$, and related to $I$. Further, $I$ is the only two-sided ideal of $A$ to which $L$ is related.