Twisted Steinberg algebras
Becky Armstrong, Lisa Orloff Clark, Kristin Courtney, Ying-Fen Lin, Kathryn McCormick, Jacqui Ramagge
aa r X i v : . [ m a t h . R A ] O c t TWISTED STEINBERG ALGEBRAS
BECKY ARMSTRONG, LISA ORLOFF CLARK, KRISTIN COURTNEY, YING-FEN LIN,KATHRYN MCCORMICK, AND JACQUI RAMAGGE
Abstract.
We introduce twisted Steinberg algebras, which generalise complex Stein-berg algebras and are a purely algebraic notion of Renault’s twisted groupoid C ∗ -algebras.In particular, for each ample Hausdorff groupoid G and each locally constant 2-cocycle σ on G taking values in the complex unit circle, we study the complex ∗ -algebra A ( G, σ )consisting of locally constant compactly supported functions on G , with convolution andinvolution twisted by σ . We also introduce a “discretised” analogue of a twist Σ over aHausdorff ´etale groupoid G , and we show that there is a one-to-one correspondence be-tween locally constant 2-cocycles on G and discrete twists over G admitting a continuousglobal section. Given a discrete twist Σ arising from a locally constant 2-cocycle σ onan ample Hausdorff groupoid G , we construct an associated Steinberg algebra A ( G ; Σ),and we show that it coincides with A ( G, σ ). We also prove a graded uniqueness the-orem for A ( G, σ ), and under the additional hypothesis that G is effective, we prove aCuntz–Krieger uniqueness theorem and show that simplicity of A ( G, σ ) is equivalent tominimality of G . Introduction
Steinberg algebras have become a topic of great interest for algebraists and analystsalike since their independent introduction in [28] and [6]. Before Steinberg algebras werespecified by name, they appeared in the details of many groupoid C ∗ -algebra construc-tions, such as those in [9, 13, 14, 22]. Not only have these algebras provided useful insightinto the analytic theory of groupoid C ∗ -algebras, they give rise to interesting examplesof ∗ -algebras; for example, all Leavitt path algebras and Kumjian–Pask algebras canbe realised as Steinberg algebras. Moreover, Steinberg algebras have served as a bridgeto facilitate the transfer of concepts and techniques between the algebraic and analyticsettings; see [3] for one such case.Thirty years prior to the introduction of Steinberg algebras, Renault [24] initiatedthe study of twisted groupoid C ∗ -algebras. These are a generalisation of groupoid C ∗ -algebras in which multiplication and involution are twisted by a T -valued 2-cocycle onthe groupoid. Twisted groupoid C ∗ -algebras have since proved extremely valuable inthe study of structural properties for large classes of C ∗ -algebras. In particular, workof Renault [25], Tu [29], and Barlak and Li [2] has revealed deep connections betweentwisted groupoid C ∗ -algebras and the UCT problem from the classification program forC ∗ -algebras. For more work on twisted C ∗ -algebras associated to graphs and groupoids,see [1, 11, 12, 15, 16, 17, 18, 19, 27].Given the success of non-twisted Steinberg algebras and the far-reaching significanceof C ∗ -algebraic results relating to twisted groupoid C ∗ -algebras, we expect that a purelyalgebraic analogue of twisted groupoid C ∗ -algebras will supply several versatile classes of ∗ -algebras to the literature, as well as a new avenue to approach important problems inC ∗ -algebras. In this article, we introduce the notion of a twisted Steinberg algebra A ( G, σ )(or A C ( G, σ )) constructed from an ample Hausdorff groupoid G and a locally constant Date : October 30, 2019.2010
Mathematics Subject Classification.
Key words and phrases.
Steinberg algebra, topological groupoid, cohomology, graded algebra. T -valued 2-cocycle σ on G . Our construction generalises the Steinberg algebra A C ( G ),and provides a purely algebraic analogue of the twisted groupoid C ∗ -algebra C ∗ ( G, σ ).In the non-twisted setting, the Steinberg algebra and the C ∗ -algebra associated to anample Hausdorff groupoid G are both built from the convolution algebra C c ( G ). As avector space, C c ( G ) denotes the set of continuous compactly supported functions fromthe groupoid to the complex field C , with pointwise operations. The complex Steinbergalgebra A ( G ) of G is the ∗ -subalgebra of C c ( G ) consisting of locally constant functions,and the full (or reduced) groupoid C ∗ -algebra C ∗ ( G ) (or C ∗ r ( G )) is the closure of C c ( G )with respect to the full (or reduced) C ∗ -norm (see [26, Chapter 3]). It turns out (see[6, Proposition 4.2]) that A ( G ) sits densely inside of both the full and the reduced C ∗ -algebras. Therefore, the definition of a twisted Steinberg algebra should result in thesame inclusions; that is, the twisted, complex, involutive Steinberg algebra should sit ∗ -algebraically and densely inside the twisted groupoid C ∗ -algebra. However to evenmake sense of that goal, one must first choose between two methods of constructing atwisted groupoid C ∗ -algebra. The first involves twisting the multiplication on C ∗ ( G ) bya continuous T -valued 2-cocycle, whereas the second involves constructing a C ∗ -algebrafrom a twist over the groupoid itself.In [24], Renault observed that the structure of a twisted groupoid C ∗ -algebra withmultiplication incorporating a 2-cocycle σ could be realised instead by first twisting thegroupoid itself, and then constructing an associated C ∗ -algebra. This is achieved byforming a split groupoid extension G (0) × T ֒ → G × σ T ։ G, where multiplication and inversion on the groupoid G × σ T both incorporate a T -valued2-cocycle σ on G , and then defining the twisted groupoid C ∗ -algebra to be the completionof the algebra of T -equivariant functions on C c ( G × T ) under a C ∗ -norm. A few yearslater, while developing a C ∗ -analogue of Feldman–Moore theory, Kumjian [12] observedthe need for a more general construction arising from a locally split groupoid extension G (0) × T ֒ → Σ ։ G, where Σ is not necessarily homeomorphic to G × T . It turns out that when G is a second-countable, ample, Hausdorff groupoid, a folklore result (Theorem 4.10) tells us that everytwist over G does arise from a T -valued 2-cocycle on G .Therefore, our first task is to define twisted Steinberg algebras with respect to bothnotions of a twist, and then to show that they coincide when these twists are constructedusing the same 2-cocycle. This is the focus of Sections 3 and 4. In Section 3, we define thetwisted Steinberg algebra A ( G, σ ) by taking an ample Hausdorff groupoid G and twistingthe multiplication of the classical Steinberg algebra A ( G ) using a locally constant T -valued2-cocycle σ on G . We then show that A ( G, σ ) sits densely inside the twisted groupoidC ∗ -algebra C ∗ ( G, σ ). In Section 4.3, we give an alternative construction of a twistedSteinberg algebra built using a twist over G , and then verify that these two definitions oftwisted Steinberg algebras agree when the twist over G arises from a 2-cocycle.However, in order to construct a twisted Steinberg algebra using a twist over a group-oid, we are forced to first “discretise” our groupoid extension by replacing the standardtopology on T with the discrete topology. Though this may seem a little artificial to a C ∗ -algebraist, this change is indeed necessary, as we explain in Remarks 4.20. (Nonetheless,this should not come as too much of a surprise, given the purely algebraic nature of Stein-berg algebras.) Thus, Section 4.1 is dedicated to introducing these discretised groupoidtwists and establishing in this setting the aforementioned folklore result (Theorem 4.10).Then in Section 4.2, we flesh out the relationships between these twists over groupoidsand the cohomology theory of groupoids. WISTED STEINBERG ALGEBRAS 3
Section 5 provides several examples of twisted Steinberg algebras, including a notion of twisted Kumjian–Pask algebras . The final two sections of the paper are devoted to provingseveral important results in Steinberg algebras in the twisted setting. In Section 6 weprove a twisted version of the Cuntz–Krieger uniqueness theorem for effective groupoids(Theorem 6.1), and we show that when G is effective, simplicity of A ( G, σ ) is equivalentto minimality of G (Theorem 6.2). Finally, in Section 7, we show that twisted Steinbergalgebras inherit a graded structure from the underlying groupoid, and we prove a gradeduniqueness theorem for twisted Steinberg algebras (Theorem 7.2).In [28], and in much of the related literature, Steinberg algebras are defined moregenerally by replacing the set C of scalars with a unital ring R (which may not have aninvolution). Because our inspiration comes from twisted groupoid C ∗ -algebras, we havechosen to focus on the setting where the ring of scalars for the algebra is C . However, if R is a unital ring with involution r r ∈ R such that r = r − for every unit r ∈ R × ⊆ R ,then we expect that all of the results of Section 3 and much of Section 4 will still holdwhen C is replaced by R and T is replaced by R × .2. Preliminaries
In this section we introduce some notation, and we recall relevant background informa-tion on topological groupoids, continuous 2-cocycles, and twisted groupoid C ∗ -algebras.Throughout this article, G will always be a locally compact Hausdorff topological group-oid with unit space G (0) , composable pairs G (2) ⊆ G × G , and range and source maps r, s : G → G (0) . We will refer to such groupoids as Hausdorff groupoids . We evaluatecomposition of groupoid elements from right to left, which means that γγ − = r ( γ ) and γ − γ = s ( γ ), for all γ ∈ G . We write G (3) for the set of composable triples in G ; that is, G (3) := { ( α, β, γ ) : ( α, β ) , ( β, γ ) ∈ G (2) } . For each x ∈ G (0) , we define G x := s − ( x ) , G x := r − ( x ) , and G xx := G x ∩ G x . For any two subsets U and V of a groupoid G , we define U s × r V := ( U × V ) ∩ G (2) , U V := { αβ : ( α, β ) ∈ U s × r V } , and U − := { α − : α ∈ U } . We call a subset B of G a bisection if there exists an open subset U of G such that B ⊆ U , and r | U and s | U are homeomorphisms onto open subsets of G . We say that G is ´etale if r (or, equivalently, s ) is a local homeomorphism. If G is ´etale, then G (0) is open,and both G x and G x are discrete in the subspace topology for any x ∈ G (0) . We recallthat G is ´etale if and only if G has a basis of open bisections. We say that G is ample ifit has a basis of compact open bisections. If G is ´etale, then G is ample if and only if itsunit space G (0) is totally disconnected (see [10, Proposition 4.1]).If B and D are compact open bisections of an ample Hausdorff groupoid, then B − and BD are also compact open bisections. In fact, the collection of compact open bisectionsforms an inverse semigroup under these operations (see [22, Proposition 2.2.4]).The isotropy of a groupoid G is the setIso( G ) := { γ ∈ G : r ( γ ) = s ( γ ) } = [ x ∈ G (0) G xx . We say that G is principal if Iso( G ) = G (0) , and that G is effective if the topologicalinterior of Iso( G ) is equal to G (0) . We say that G is topologically principal if the set { x ∈ G (0) : G xx = { x }} is dense in G (0) . Every principal ´etale groupoid is effective andtopologically principal. If G is a Hausdorff ´etale groupoid, then G is effective if it istopologically principal, and the converse holds if G is additionally second-countable (see ARMSTRONG, CLARK, COURTNEY, LIN, MCCORMICK, AND RAMAGGE [3, Lemma 3.1]). We will often work with Hausdorff groupoids that are ´etale, ample, orsecond-countable, but we will explicitly state these assumptions.Before we describe algebras of functions defined on a groupoid, a few remarks on pre-liminary point-set topology and notation are in order. Given topological spaces X and Y , a function f : X → Y is said to be locally constant if every element of X has an openneighbourhood U such that f | U is constant. Every locally constant function is continuous(because the preimage of every singleton set under a locally constant function is open);moreover, if Y has the discrete topology, then every continuous function f : X → Y islocally constant. We write C d for the set of complex numbers endowed with the discretetopology, and T d for the complex unit circle endowed with the discrete topology. We willfrequently view locally constant C -valued (or T -valued) functions as continuous functionstaking values in C d (or T d ).Given a complex-valued function f on a topological space X , we define the support of f to be the set supp( f ) := { x ∈ X : f ( x ) = 0 } . If f is continuous, then its support is open, because supp( f ) = f − ( C \{ } ). If f islocally constant, then its support is clopen, because supp( f ) = f − ( C d \{ } ). If supp( f )is compact, then we say that f is compactly supported .As motivation for our definition of a twisted Steinberg algebra, it will be helpful tobriefly recall the construction of groupoid C ∗ -algebras and Steinberg algebras, and to de-scribe the ways in which twisted groupoid C ∗ -algebras have been defined in the literature.We begin by describing groupoid C ∗ -algebras, which were introduced by Renault in[24]. In the discussion that follows, it will suffice to restrict our attention to the settingin which the underlying Hausdorff groupoid G is second-countable and ´etale. Althoughthe ´etale assumption is not required, this setting is general enough to include a plethoraof examples, including the Cuntz–Krieger algebras of all compactly aligned topologicalhigher-rank graphs (see [30, Theorem 3.16]).Given a second-countable Hausdorff ´etale groupoid G , the convolution algebra C c ( G ) isthe complex ∗ -algebra C c ( G ) := { f : G → C : f is continuous and supp( f ) is compact } , equipped with multiplication given by the convolution product ( f ∗ g )( γ ) := X ( α,β ) ∈ G (2) ,αβ = γ f ( α ) g ( β ) = X η ∈ G s ( γ ) f ( γη ) g ( η − ) , and involution given by f ∗ ( γ ) := f ( γ − ). The full groupoid C ∗ -algebra C ∗ ( G ) is definedto be the completion of C c ( G ) in the full C ∗ -norm , and the reduced groupoid C ∗ -algebra C ∗ r ( G ) is defined to be the completion of C c ( G ) in the reduced C ∗ -norm (see [26, Chapter 3]for the details).The first conception of a twisted groupoid C ∗ -algebra was also introduced by Renaultin [24]. In this setting, the “twist” refers to a continuous T -valued 2-cocycle on G , whichis incorporated into the definitions of the multiplication and involution of the convolutionalgebra C c ( G ). A 2 -cocycle is a continuous function σ : G (2) → T that satisfies the 2 -cocycle identity : σ ( α, β ) σ ( αβ, γ ) = σ ( α, βγ ) σ ( β, γ ) , for all ( α, β, γ ) ∈ G (3) , and is normalised , in the sense that σ ( r ( γ ) , γ ) = 1 = σ ( γ, s ( γ )) , WISTED STEINBERG ALGEBRAS 5 for all γ ∈ G . We say that the 2-cocycles σ, τ : G (2) → T are cohomologous if there is acontinuous function b : G → T such that b ( x ) = 1 for all x ∈ G (0) , and σ ( α, β ) τ ( α, β ) = b ( α ) b ( β ) b ( αβ ) , for all ( α, β ) ∈ G (2) . Cohomology of continuous 2-cocycles on G is an equivalence rela-tion. The equivalence class of a continuous 2-cocycle σ under this relation is called its cohomology class .Given a 2-cocycle σ : G (2) → T , the twisted convolution algebra C c ( G, σ ) is the complex ∗ -algebra that is equal as a vector space to C c ( G ), but has multiplication given by the twisted convolution product ( f ∗ g )( γ ) := X ( α,β ) ∈ G (2) ,αβ = γ σ ( α, β ) f ( α ) g ( β ) = X η ∈ G s ( γ ) σ ( γη, η − ) f ( γη ) g ( η − ) , and involution given by f ∗ ( γ ) := σ ( γ, γ − ) f ( γ − ) . The 2-cocycle identity guarantees that the multiplication is associative, and the assump-tion that the 2-cocycle is normalised implies that the twist is trivial when either multiply-ing or applying the involution to functions supported on G (0) . The full twisted groupoid C ∗ -algebra C ∗ ( G, σ ) is defined to be the completion of C c ( G, σ ) in the full C ∗ -norm ,and the reduced twisted groupoid C ∗ -algebra C ∗ r ( G, σ ) is defined to be the completion of C c ( G, σ ) in the reduced C ∗ -norm (see [24, Chapter II.1] for the details). There is also a ∗ -algebra norm on C c ( G, σ ), called the I -norm , which is given by k f k I,σ := max n sup u ∈ G (0) (cid:8) X γ ∈ G u | f ( γ ) | (cid:9) , sup u ∈ G (0) (cid:8) X γ ∈ G u | f ( γ ) | (cid:9)o , for all f ∈ C c ( G, σ ). The I -norm dominates the full norm on C c ( G, σ ).Renault [24] also introduced an alternative construction of these twisted groupoid C ∗ -algebras involving twisting the groupoid itself, via a split groupoid extension G (0) × T ֒ → G × σ T ։ G, called a twist over G . In 1986, Kumjian generalised this construction to give twistedgroupoid C ∗ -algebras whose twists are not induced by T -valued 2-cocycles. In particular,the extension Σ of G by G (0) × T need not admit a continuous global section P : G → Σ.In Section 4.1 we develop a “discretised” version of this more general notion of a twist.Since our definition is almost identical to Kumjian’s (with the difference being the choiceof topology on T ), we refer the reader to Definition 4.1 for a more precise definition of atwist over a Hausdorff ´etale groupoid. Given a twist G (0) × T ֒ → Σ ։ G, over a Hausdorff ´etale groupoid G , one defines C c (Σ) with (untwisted) convolution andinvolution. The completion of the ∗ -subalgebra of C c (Σ) consisting of T -equivariant func-tions with respect to the full (or reduced) C ∗ -norm yields the full (or reduced) twistedgroupoid C ∗ algebra C ∗ ( G, Σ) (or C ∗ r ( G, Σ)). (See [25] or [26, Chapter 5] for more details.)We conclude this section with the definition of Steinberg algebras, which were originallyintroduced in [28, 6], and are a purely algebraic analogue of groupoid C ∗ -algebras. Let G be an ample Hausdorff groupoid and let 1 B denote the characteristic function of B from G to C . The (complex) Steinberg algebra associated to G is the complex ∗ -algebra A C ( G ) := span { B : G → C : B is a compact open bisection of G } = { f : G → C : f is locally constant and supp( f ) is compact } , ARMSTRONG, CLARK, COURTNEY, LIN, MCCORMICK, AND RAMAGGE equipped with multiplication given by the convolution product ( f ∗ g )( γ ) := X ( α,β ) ∈ G (2) ,αβ = γ f ( α ) g ( β ) = X η ∈ G s ( γ ) f ( γη ) g ( η − ) , and involution given by f ∗ ( γ ) := f ( γ − ). The Steinberg algebra A ( G ) := A C ( G ) is densein C c ( G ) with respect to the full and reduced C ∗ -norms (as shown in [28, 6]).3. Twisted Steinberg algebras arising from locally constant -cocycles In this section we introduce the twisted complex Steinberg algebra A ( G, σ ) (or A C ( G, σ ))associated to an ample Hausdorff groupoid G and a continuous 2-cocycle σ : G (2) → T d .As a vector space, the twisted Steinberg algebra is identical to the untwisted versiondefined in Section 2. That is A ( G, σ ) := span { B : G → C d : B is a compact open bisection of G } ;we now emphasise that we are viewing C with the discrete topology. Lemma 3.1.
Let G be an ample Hausdorff groupoid. Let C c ( G, C d ) denote the collectionof continuous, compactly supported functions f : G → C d . For any continuous -cocycle σ : G (2) → T d , we have the following: (a) A ( G, σ ) = C c ( G, C d ) = { f ∈ C c ( G ) : f is locally constant } as vector spaces; and (b) for any f ∈ A ( G, σ ) , there exist λ , . . . , λ n ∈ C \{ } and mutually disjoint compactopen bisections B , . . . , B n ⊆ G such that f = P ni =1 λ i B i .Proof. Part (a) follows from the characterisations of the (untwisted, complex) Steinbergalgebra A ( G ) given in [6, Definition 3.2 and Lemma 3.3], because A ( G, σ ) and A ( G ) agreeas sets. Similarly, part (b) follows from [6, Lemma 3.5]. (cid:3) From now on, we will use the characterisations of A ( G, σ ) given in Lemma 3.1 inter-changeably with the definition.As a vector space, A ( G, σ ) is identical to the usual (complex) Steinberg algebra A ( G )introduced in [28, 6]. However, we equip A ( G, σ ) with a multiplication and involutionthat both incorporate the 2-cocycle σ into their definitions, thereby distinguishing A ( G, σ )from A ( G ). Proposition 3.2.
Let G be an ample Hausdorff groupoid and σ : G (2) → T d be a contin-uous -cocycle. There is a multiplication (called (twisted) convolution ) on A ( G, σ ) givenby ( f ∗ g )( γ ) := X ( α,β ) ∈ G (2) ,αβ = γ σ ( α, β ) f ( α ) g ( β ) = X η ∈ G s ( γ ) σ ( γη, η − ) f ( γη ) g ( η − ) , and an involution given by f ∗ ( γ ) := σ ( γ, γ − ) f ( γ − ) . Under these operations, along with pointwise addition and scalar multiplication, A ( G, σ ) is a dense ∗ -subalgebra of the twisted convolution algebra C c ( G, σ ) with respect to the I -norm, and hence also with respect to the full and reduced C ∗ -norms.We call A ( G, σ ) the twisted Steinberg algebra associated to the pair ( G, σ ) .Remarks . (1) If the 2-cocycle σ is trivial (in the sense that σ (cid:0) G (2) (cid:1) = { } ), then A ( G, σ ) isidentical to A ( G ) as a complex ∗ -algebra.(2) We often write f g to denote the convolution product f ∗ g of functions f, g ∈ A ( G, σ ) if the intended meaning is clear.
WISTED STEINBERG ALGEBRAS 7 (3) If f, g ∈ A ( G, σ ), then supp( f g ) ⊆ supp( f ) supp( g ). If B and D are compact openbisections of G such that supp( f ) = B and supp( g ) = D , then supp( f g ) = BD and supp( f ∗ ) = B − .(4) From the 2-cocycle identity, one can readily verify that σ ( γ, γ − ) = σ ( γ − , γ ) forany γ ∈ G . Proof of Proposition 3.2.
Since A ( G, σ ) and A ( G ) agree as vector spaces, it follows from[22, Proposition 2.2.7] that A ( G, σ ) is dense in C c ( G, σ ) with respect to the I -norm. Weknow from [24, Proposition II.1.1] that C c ( G, σ ) is a ∗ -algebra, and so to see that A ( G, σ )is a ∗ -algebra, it suffices to show that A ( G, σ ) is closed under the twisted convolution andinvolution.Fix f, g ∈ A ( G, σ ). By Lemma 3.1(b), there exist mutually disjoint compact openbisections B , . . . , B m , C , . . . , C n ⊆ G and scalars λ , . . . , λ m , µ , . . . , µ n ∈ C \{ } suchthat f = m X i =1 λ i B i and g = n X j =1 µ j C j . We claim that f g ∈ A ( G, σ ). Since 0 / ∈ σ (cid:0) G (2) (cid:1) , [26, Proposition 3.1.1] implies that foreach γ ∈ G , the set (cid:8) ( α, β ) ∈ G (2) : αβ = γ and σ ( α, β ) f ( α ) g ( β ) = 0 (cid:9) is finite. Since σ is locally constant, we can assume that for all i ∈ { , . . . , m } and j ∈ { , . . . , n } , there exists ν i,j ∈ T d such that σ ( α, β ) = ν i,j for all ( α, β ) ∈ ( B i ) s × r ( C j )(because otherwise we can further refine the bisections to ensure that this is true). Thus,for all γ ∈ G , we have( f g )( γ ) = X ( α,β ) ∈ G (2) ,αβ = γ σ ( α, β ) f ( α ) g ( β )= X ( α,β ) ∈ G (2) ,αβ = γ σ ( α, β ) m X i =1 λ i B i ( α ) ! n X j =1 µ j C j ( β ) ! = X ( α,β ) ∈ G (2) ,αβ = γ m X i =1 n X j =1 ν i,j λ i µ j B i ( α ) 1 C j ( β )= m X i =1 n X j =1 ν i,j λ i µ j B i C j ( γ ) . Hence f g ∈ A ( G, σ ).We now show that f ∗ ∈ A ( G, σ ). Since σ is locally constant, we can assume that forall i ∈ { , . . . , m } , there exists κ i ∈ T d such that σ ( γ, γ − ) = κ i for all γ ∈ B i (becauseotherwise we can further refine the bisections to ensure that this is true). Thus, for all γ ∈ G , we have f ∗ ( γ ) = σ ( γ, γ − ) f ( γ − ) = σ ( γ, γ − ) m X i =1 λ i B i ( γ − ) ! = m X i =1 κ i λ i B − i ( γ ) . Hence f ∗ ∈ A ( G, σ ). (cid:3) Note that we used that σ is locally constant in order to show that A ( G, σ ) is closedunder the twisted convolution and involution.
ARMSTRONG, CLARK, COURTNEY, LIN, MCCORMICK, AND RAMAGGE
In the untwisted Steinberg algebra setting, given compact open bisections B and D of G , we have 1 B D = 1 BD . This is not the case in the twisted setting, due to the presenceof the 2-cocycle in the convolution formula. Instead, we have the following propertiesconcerning the generators 1 B of the twisted Steinberg algebra A ( G, σ ). Lemma 3.4.
Let G be an ample Hausdorff groupoid, σ : G (2) → T d be a continuous -cocycle, and B and D be compact open bisections of G . (a) For all ( α, β ) ∈ B s × r D , we have (1 B D )( αβ ) = σ ( α, β ) 1 B ( α ) 1 D ( β ) = σ ( α, β ) 1 BD ( αβ ) = σ ( α, β ) . (b) If B ⊆ G (0) or D ⊆ G (0) , then B D = 1 BD . (c) For all γ ∈ G , we have ∗ B ( γ ) = σ ( γ, γ − ) 1 B − ( γ ) . (d) We have B ∗ B = 1 r ( B ) and ∗ B B = 1 s ( B ) . (e) We have B ∗ B B = 1 B and ∗ B B ∗ B = 1 ∗ B .Proof. (a) This follows immediately from the definition of the twisted convolutionproduct because B and D are bisections.(b) Suppose that B ⊆ G (0) or D ⊆ G (0) , and fix γ ∈ G . If γ ∈ BD , then γ = αβ forsome pair ( α, β ) ∈ B s × r D . Since σ is normalised, we have σ ( α, β ) = 1, and so(1 B D )( γ ) = σ ( α, β ) 1 B ( α ) 1 D ( β ) = 1 B ( α ) 1 D ( β ) = 1 BD ( γ ) . If γ / ∈ BD , then (1 B D )( γ ) = 0 = 1 BD ( γ ). Thus 1 B D = 1 BD .(c) If γ ∈ B − , then we have1 ∗ B ( γ ) = σ ( γ, γ − ) 1 B ( γ − ) = σ ( γ, γ − ) 1 B − ( γ ) . If γ / ∈ B − = supp(1 ∗ B ), then1 ∗ B ( γ ) = 0 = 1 B − ( γ ) = σ ( γ, γ − ) 1 B − ( γ ) . (d) We know that supp(1 B ∗ B ) = BB − = r ( B ), and for all γ ∈ B , we have(1 B ∗ B ) (cid:0) r ( γ ) (cid:1) = (1 B ∗ B )( γγ − )= σ ( γ, γ − ) 1 B ( γ ) 1 ∗ B ( γ − )= σ ( γ, γ − ) 1 B ( γ ) σ ( γ − , γ ) 1 B − ( γ − ) (using part (c))= 1= 1 r ( B ) (cid:0) r ( γ ) (cid:1) . Similarly, we have supp(1 ∗ B B ) = B − B = s ( B ), and so for all γ ∈ B , we have(1 ∗ B B ) (cid:0) s ( γ ) (cid:1) = (1 ∗ B B )( γ − γ )= σ ( γ − , γ ) 1 ∗ B ( γ − ) 1 B ( γ )= σ ( γ − , γ ) σ ( γ − , γ ) 1 B − ( γ − ) 1 B ( γ ) (using part (c))= 1= 1 s ( B ) (cid:0) s ( γ ) (cid:1) . (e) Parts (b) and (d) imply that1 B ∗ B B = 1 r ( B ) B = 1 r ( B ) B = 1 B , and 1 ∗ B B ∗ B = 1 s ( B ) ∗ B . Hence supp(1 ∗ B B ∗ B ) = s ( B ) B − = B − . For all γ ∈ B , we have(1 ∗ B B ∗ B )( γ − ) = σ ( s ( γ ) , γ − ) 1 s ( B ) (cid:0) s ( γ ) (cid:1) ∗ B ( γ − ) = 1 ∗ B ( γ − ) , and so 1 ∗ B B ∗ B = 1 ∗ B . (cid:3) WISTED STEINBERG ALGEBRAS 9
The proof of the following result is inspired by the proof of [24, Proposition II.1.2].
Lemma 3.5.
Let G be an ample Hausdorff groupoid, and σ, τ : G (2) → T d be two con-tinuous -cocycles whose cohomology classes coincide. Then A ( G, σ ) is ∗ -isomorphic to A ( G, τ ) .Proof. For this proof, we will use ∗ to denote convolution, in order to distinguish it fromthe pointwise product.Since σ is cohomologous to τ , there is a continuous function b : G → T d such that b ( x ) = 1 for all x ∈ G (0) , and σ ( α, β ) τ ( α, β ) = b ( α ) b ( β ) b ( αβ ) , (3.1)for all ( α, β ) ∈ G (2) .For each f ∈ A ( G, σ ) = C c ( G, C d ), let φ ( f ) denote the pointwise product bf . Since bf : G → C d is continuous and satisfies supp( bf ) = supp( f ), we have bf ∈ C c ( G, C d ) = A ( G, τ ). We claim that φ : A ( G, σ ) → A ( G, τ ) is a ∗ -isomorphism. It is clear that φ islinear, so we must show that it respects the twisted convolution and involution.For all γ ∈ G , letting α = γ and β = γ − in Equation (3.1) gives σ ( γ, γ − ) τ ( γ, γ − ) = b ( γ ) b ( γ − ) b ( γγ − ) = b ( γ ) b ( γ − ) , and hence b ( γ ) σ ( γ, γ − ) = τ ( γ, γ − ) b ( γ − ) . (3.2)Thus, for all f ∈ A ( G, σ ) and γ ∈ G , we have φ ( f ∗ )( γ ) = b ( γ ) f ∗ ( γ )= b ( γ ) σ ( γ, γ − ) f ( γ − )= τ ( γ, γ − ) b ( γ − ) f ( γ − ) (using Equation (3.2))= ( bf ) ∗ ( γ )= φ ( f ) ∗ ( γ ) . For all ( α, β ) ∈ G (2) , Equation (3.1) implies that σ ( α, β ) b ( αβ ) = τ ( α, β ) b ( α ) b ( β ) . (3.3)Hence, for all f, g ∈ A ( G, σ ) and γ ∈ G , we have (cid:0) φ ( f ) ∗ φ ( g ) (cid:1) ( γ ) = X ( α,β ) ∈ G (2) ,αβ = γ τ ( α, β ) φ ( f )( α ) φ ( g )( β )= X ( α,β ) ∈ G (2) ,αβ = γ τ ( α, β ) b ( α ) f ( α ) b ( β ) g ( β )= X ( α,β ) ∈ G (2) ,αβ = γ σ ( α, β ) b ( αβ ) f ( α ) g ( β ) (using Equation (3.3))= b ( γ ) X ( α,β ) ∈ G (2) ,αβ = γ σ ( α, β ) f ( α ) g ( β )= (cid:0) b ( f ∗ g ) (cid:1) ( γ )= φ ( f ∗ g )( γ ) . Therefore, φ is a ∗ -homomorphism. We now show that φ is a bijection. For each h ∈ A ( G, τ ), we have bh ∈ A ( G, σ ),and so φ ( bh ) = bbh = h . Hence φ is surjective. To see that φ is injective, suppose that f, g ∈ A ( G, σ ) satisfy φ ( f ) = φ ( g ). Then f = bbf = bφ ( f ) = bφ ( g ) = bbg = g . Therefore, φ is a ∗ -isomorphism. (cid:3) Proposition 3.6.
Let G be an ample Hausdorff groupoid and σ : G (2) → T d be a contin-uous -cocycle. The set { B : G → C d : B is a nonempty compact open subset of G (0) } forms a local unit for A ( G, σ ) . That is, for any finite collection f , . . . , f n ∈ A ( G, σ ) ,there exists a compact open subset E of G (0) such that E f i = f i = f i E , for each i ∈ { , . . . , n } .Proof. Since multiplication by 1 E for E ⊆ G (0) is not affected by the 2-cocycle, this followsfrom the analogous non-twisted result [5, Lemma 2.6]. (cid:3) Twisted Steinberg algebras arising from discrete twists
There is another (often more general) notion of a twisted groupoid C ∗ -algebra which isconstructed from a “twist” over the groupoid itself; that is, an algebra built from a locallysplit groupoid extension of an ample Hausdorff groupoid G by G (0) × T . In this section,we define a discretised analogue of this twist and its associated twisted Steinberg algebra.The primary modification is to replace the standard topology on T with the discretetopology. Many of the results in Section 4.1 and Section 4.2 have roots or inspiration inKumjian’s study of groupoid C ∗ -algebras built from groupoid extensions in [12].The results in Section 4.1 and Section 4.2 also hold in the non-discrete setting with thesame proofs. Replacing T d with T will not change any of the algebraic arguments therein,and the topological arguments carry through mutatis mutandis. As our ultimate focus isalgebraic, we present all of our results in terms of T d .4.1. Discrete twists over Hausdorff ´etale groupoids.
The definition of a twist overa Hausdorff ´etale groupoid, which we refer to as a classical twist , can be found in [26,Definition 5.1.1]. The following is our discretised version.
Definition 4.1.
Let G be a Hausdorff ´etale groupoid. A discrete twist over G is asequence G (0) × T d i ֒ → Σ q ։ G, where the groupoid G (0) × T d is regarded as a trivial group bundle with fibres T d , Σ isa Hausdorff groupoid with Σ (0) = i (cid:0) G (0) × { } (cid:1) , and i and q are continuous groupoidhomomorphisms that restrict to homeomorphisms of unit spaces, such that the followingconditions hold.(a) The sequence is exact, in the sense that i ( { x } × T d ) = q − ( x ) for every x ∈ G (0) , i is injective, and q is surjective. (b) The groupoid Σ is a locally trivial G -bundle, in the sense that for each α ∈ G , thereis an open bisection B α of G containing α , and a continuous map P α : B α → Σsuch that(i) q ◦ P α = id B α ;(ii) P α ( G (0) ∩ B α ) ⊆ Σ (0) ; and Although it is not explicitly stated in [26, Definition 5.1.1] that the groupoid homomorphism q : Σ → G is surjective and satisfies q ( i ( x, z )) = x for every ( x, z ) ∈ G (0) × T d , it is implicitly assumed. WISTED STEINBERG ALGEBRAS 11 (iii) the map ( β, z ) i ( r ( β ) , z ) P α ( β ) is a homeomorphism from B α × T d to q − ( B α ).(c) The image of i is central in Σ, in the sense that i ( r ( ǫ ) , z ) ǫ = ǫ i ( s ( ǫ ) , z ) for all ǫ ∈ Σ and z ∈ T d .We denote a discrete twist over G either by (Σ , i, q ), or simply by Σ. We identify Σ (0) with G (0) via i . A continuous map P α : B α → Σ is called a (continuous) local section ifit satisfies parts (i) and (ii) of condition (b). A (classical) twist over G has the samedefinition as above, with the exception that T d is replaced by T .In brief, we think of a twist over G as a locally split extension Σ of G by G (0) × T d ,where the image of G (0) × T d is central in Σ. If G is ample, then the open bisections fromcondition (b) can be chosen to be compact. Example . If G is a discrete group, then a twist over G as defined above is a centralextension of G .The following result is an immediate consequence of Definition 4.1. Lemma 4.3.
Let G be a Hausdorff ´etale groupoid, and (Σ , i, q ) be a discrete twist over G . Then i is a homeomorphism onto its image.Proof. Since i is injective and continuous by definition, we need only show that i is anopen map. Fix open sets U ⊆ G (0) and W ⊆ T d . For each x ∈ U , condition (b)(iii) ofDefinition 4.1 implies that there is an open bisection B x of G containing x , and a homeo-morphism ψ x : B x × T d → q − ( B x ) given by ψ x ( β, z ) := i ( r ( β ) , z ) P x ( β ). In particular, foreach y ∈ B x ∩ G (0) and z ∈ T d , we have ψ x ( y, z ) = i ( y, z ), since P x ( y ) ∈ Σ (0) . Therefore, i ( U × W ) = [ x ∈ U ψ x (cid:0) ( B x ∩ U ) × W (cid:1) , which is an open subset of Σ, because each ψ x is a homeomorphism onto the open set q − ( B x ), and each B x ∩ U is open. (cid:3) We define a notion of an isomorphism of discrete twists in an analogous way to thenon-discrete version.
Definition 4.4.
We say that two twists (Σ , i, q ) and (Σ ′ , i ′ , q ′ ) over a Hausdorff ´etalegroupoid G are isomorphic if there exists a groupoid isomorphism φ : Σ → Σ ′ that isequivariant for i ′ and q ′ ; or, equivalently, if the following diagram commutes. G (0) × T d Σ GG (0) × T d Σ ′ G i qφi ′ q ′ It is natural to ask whether there is a correspondence between twists over a groupoid andlocally constant 2-cocycles which can be used to “twist” the multiplication in Steinbergalgebras, given the shared terminology. As one familiar with the literature would expect,we can readily build a twist over a Hausdorff ´etale groupoid from a locally constant 2-cocycle. To demonstrate this, we adapt the construction outlined in [26, Example 5.1.5]to the setting where the continuous 2-cocycle maps into T d (rather than T ), which isequivalent to insisting that the 2-cocycle is locally constant. We say that φ : Σ → Σ ′ is a groupoid isomorphism if it is a homeomorphism such that φ ( δǫ ) = φ ( δ ) φ ( ǫ )for all ( δ, ǫ ) ∈ Σ (2) . Example . Let G be a Hausdorff ´etale groupoid, and σ : G (2) → T d be a continuous2-cocycle. Let G × σ T d be the set G × T d endowed with the product topology, withmultiplication given by ( α, z )( β, w ) := ( αβ, σ ( α, β ) zw ) , and inversion given by( α, z ) − := ( α − , σ ( α, α − ) z ) = ( α − , σ ( α − , α ) z ) , for all ( α, β ) ∈ G (2) and z, w ∈ T d . Then G × σ T d is a Hausdorff groupoid. In fact,unlike in the classical setting, G being ´etale implies that G × σ T d is ´etale, because foreach z ∈ T d and bisection U of G , r | U ×{ z } is a homeomorphism onto r ( U ) × { } . Define i : G (0) × T d → G × σ T d by i ( x, z ) := ( x, z ), and q : G × σ T d → G by q ( γ, z ) := γ . Then q is easily verified to be a surjective groupoid homomorphism, and since σ is normalised, i is an injective groupoid homomorphism. Just as in [26, Example 5.1.5], it is routine tothen check that ( G × σ T d , i, q ) is a discrete twist over G .Example 4.5 shows that any locally constant 2-cocycle on a Hausdorff ´etale groupoid G gives rise to a discrete twist over G . According to folklore, the converse is true when G is additionally second-countable and ample. The proof of this fact and its consequenceswill be the focus of the remainder of this subsection.Before we proceed, we need two technical results regarding the left and right groupactions of T d on Σ that are induced by the map i : G (0) × T d → Σ. Identifying Σ (0) with G (0) , these actions are given by z · ǫ := i ( r ( ǫ ) , z ) ǫ and ǫ · z := ǫ i ( s ( ǫ ) , z ) , for each z ∈ T d and ǫ ∈ Σ. Since the image of i is central in Σ, we have z · ǫ = ǫ · z , and( z · ǫ )( w · δ ) = ( zw ) · ( ǫδ ) for all ( ǫ, δ ) ∈ Σ (2) and z, w ∈ T d . Lemma 4.6.
Let G be a Hausdorff ´etale groupoid. Suppose that (Σ , i , q ) and (Σ , i , q ) are discrete twists over G , and φ : Σ → Σ is an isomorphism of twists, as defined inDefinition 4.4. Then φ respects the action of T d , in the sense that φ ( z · ǫ ) = z · φ ( ǫ ) , forall z ∈ T d and ǫ ∈ Σ .Proof. Since φ : Σ → Σ is an isomorphism of twists, we have i = φ ◦ i . Thus, for all z ∈ T d and ǫ ∈ Σ , we have φ ( z · ǫ ) = φ (cid:0) i ( r ( ǫ ) , z ) ǫ (cid:1) = i ( r ( ǫ ) , z ) φ ( ǫ ) = z · φ ( ǫ ) . (cid:3) The following result is inspired by [26, Proposition 5.1.3].
Lemma 4.7.
Let G be a Hausdorff ´etale groupoid. Suppose that (Σ , i, q ) is a discretetwist over G , and δ, ǫ ∈ Σ satisfy q ( δ ) = q ( ǫ ) . Then r ( δ ) = r ( ǫ ) , and there is a unique z ∈ T d such that ǫ = z · δ .Proof. Fix δ, ǫ ∈ Σ such that q ( δ ) = q ( ǫ ). Then q ( r ( δ )) = r ( q ( δ )) = r ( q ( ǫ )) = q ( r ( ǫ )),and hence r ( δ ) = r ( ǫ ), because q restricts to a homeomorphism of unit spaces. Thus q ( ǫδ − ) = q ( ǫ ) q ( ǫ ) − = r ( q ( ǫ )) ∈ G (0) , so there is a unique element z ∈ T d such that ǫδ − = i (cid:0) r ( q ( ǫ )) , z (cid:1) . By identifying Σ (0) with G (0) , we obtain ǫ = i ( r ( ǫ ) , z ) δ = z · δ . (cid:3) Notice that in the case where Σ is the twist G × σ T d described in Example 4.5, we cancheck Lemma 4.7 directly. Identifying Σ (0) = G (0) × { } with G (0) , we have z · ( α, w ) = i ( r ( α ) , z )( α, w ) = ( r ( α ) , z )( α, w ) = ( α, zw ) , for all z ∈ T d and ( α, w ) ∈ Σ. If q ( δ ) = q ( ǫ ) for some δ, ǫ ∈ Σ, then δ = ( α, w ) and ǫ = ( α, w ) for some α ∈ G and unique w , w ∈ T d . Clearly there is a unique z ∈ T d such that zw = w , and hence z · δ = ( α, zw ) = ǫ . WISTED STEINBERG ALGEBRAS 13
Our key tool in what follows will be a (continuous) global section ; that is, a continuousmap P : G → Σ, such that q ◦ P = id G and P ( G (0) ) ⊆ Σ (0) = i ( G (0) × { } ). Our nextresult shows that every discrete twist admitting a continuous global section is isomorphicto a twist coming from a locally constant 2-cocycle, as described in Example 4.5. Parts ofthis result are inspired by the analogous non-discrete versions in [12, Section 4] and [26,Chapter 5]. Proposition 4.8.
Let G be a Hausdorff ´etale groupoid, and (Σ , i, q ) be a discrete twistover G . Suppose that Σ is topologically trivial , in the sense that it admits a continuousglobal section P : G → Σ . Then the following conditions hold. (a) The continuous global section P preserves composability, and induces a continuous -cocycle σ : G (2) → T d satisfying P ( α ) P ( β ) P ( αβ ) − = i (cid:0) r ( α ) , σ ( α, β ) (cid:1) , for all ( α, β ) ∈ G (2) . (b) For all ( α, β ) ∈ G (2) , we have P ( α ) P ( β ) = σ ( α, β ) · P ( αβ ) and P ( α ) − = σ ( α, α − ) · P ( α − ) . (c) Let ( G × σ T d , i σ , q σ ) be the twist from Example 4.5. The map φ P : G × σ T d → Σ defined by φ P ( α, z ) := z · P ( α ) gives an isomorphism of the twists G × σ T d and Σ .Proof. For (a), fix ( α, β ) ∈ G (2) . Since q ◦ P = id G and q is a groupoid homomorphismthat restricts to a homeomorphism of unit spaces, we have q ( s ( P ( α ))) = s ( q ( P ( α ))) = s ( α ) = r ( β ) = r ( q ( P ( β ))) = q ( r ( P ( β ))) , and hence ( P ( α ) , P ( β )) ∈ Σ (2) . We have q (cid:0) P ( α ) P ( β ) P ( αβ ) − (cid:1) = q ( P ( α )) q ( P ( β )) q ( P ( αβ )) − = r ( α ) = q (cid:0) P ( r ( α )) (cid:1) , and so Lemma 4.7 implies that there is a unique value σ ( α, β ) ∈ T d such that P ( α ) P ( β ) P ( αβ ) − = σ ( α, β ) · P ( r ( α )) = i (cid:0) r ( α ) , σ ( α, β ) (cid:1) . (4.1)Therefore, σ ( α, β ) = ( π ◦ i − ) (cid:0) P ( α ) P ( β ) P ( αβ ) − (cid:1) , where π is the projection of G (0) × T d onto the second coordinate. Noting that i is an open map by Lemma 4.3, we deduce that σ is continuous because it is a composition of continuous functions.To check that σ satisfies the 2-cocycle identity, we fix ( α, β, γ ) ∈ G (3) and show that σ ( β, γ ) = σ ( α, β ) σ ( αβ, γ ) σ ( α, βγ ) . Since the image of i is central in Σ, we have i (cid:0) r ( α ) , σ ( β, γ ) (cid:1) P ( α ) = P ( α ) i (cid:0) s ( α ) , σ ( β, γ ) (cid:1) = P ( α ) i (cid:0) r ( β ) , σ ( β, γ ) (cid:1) . (4.2)Using Equation (4.2) for the first equality below and Equation (4.1) for the second andfourth equalities, we obtain i (cid:0) r ( α ) , σ ( β, γ ) (cid:1) = P ( α ) i (cid:0) r ( β ) , σ ( β, γ ) (cid:1) P ( α ) − = P ( α ) P ( β ) P ( γ ) P ( βγ ) − P ( α ) − = (cid:0) P ( α ) P ( β ) P ( αβ ) − (cid:1)(cid:0) P ( αβ ) P ( γ ) P ( αβγ ) − (cid:1)(cid:0) P ( αβγ ) P ( βγ ) − P ( α ) − (cid:1) = i (cid:0) r ( α ) , σ ( α, β ) (cid:1) i (cid:0) r ( αβ ) , σ ( αβ, γ ) (cid:1) i (cid:0) r ( α ) , σ ( α, βγ ) (cid:1) − = i (cid:0) r ( α ) , σ ( α, β ) σ ( αβ, γ ) σ ( α, βγ ) (cid:1) . Thus, by the injectivity of i , we deduce that σ satisfies the 2-cocycle identity.To see that σ is normalised, first note that for all α in G , q (cid:0) i (cid:0) r ( α ) , σ ( r ( α ) , α ) (cid:1)(cid:1) = q (cid:0) i (cid:0) r ( α ) , σ ( α, s ( α )) (cid:1)(cid:1) = q (cid:0) i ( r ( α ) , (cid:1) = r ( α ) , (4.3) and i ( r ( α ) , ∈ Σ (0) . Moreover, by Equation (4.1), we have i (cid:0) r ( α ) , σ ( r ( α ) , α ) (cid:1) = P ( r ( α )) P ( α ) P ( r ( α ) α ) − = P ( r ( α )) ∈ Σ (0) , and, since P ( s ( α )) ∈ Σ (0) , i (cid:0) r ( α ) , σ ( α, s ( α )) (cid:1) = P ( α ) P ( s ( α )) P ( αs ( α )) − = P ( α ) P ( α ) − = r ( P ( α )) ∈ Σ (0) . Since q restricts to a homeomorphism of unit spaces and i is injective, we deduce fromEquation (4.3) that σ ( r ( α ) , α ) = σ ( α, s ( α )) = 1 , for all α ∈ G .For (b), fix ( α, β ) ∈ G (2) . Then Equation (4.1) implies that P ( α ) P ( β ) = i (cid:0) r ( αβ ) , σ ( α, β ) (cid:1) P ( αβ ) = σ ( α, β ) · P ( αβ ) , and also that P ( α ) P ( α − ) P ( αα − ) − = i (cid:0) r ( α ) , σ ( α, α − ) (cid:1) . Since P ( αα − ) − = P ( r ( α )) ∈ Σ (0) , we deduce that P ( α ) − = P ( α − ) i (cid:0) r ( α ) , σ ( α, α − ) (cid:1) − = P ( α − ) · σ ( α, α − ) = σ ( α, α − ) · P ( α − ) . For (c), define φ P : G × σ T d → Σ by φ P ( α, z ) := z · P ( α ) = i ( r ( α ) , z ) P ( α ). Then φ P is continuous, because it is the pointwise product of the continuous maps i ◦ ( r × id) and P ◦ π from G × σ T d to Σ, where π is the projection of G × σ T d onto the first coordinate.To see that φ P is injective, suppose that ( α, z ) , ( β, w ) ∈ G (2) satisfy φ P ( α, z ) = φ P ( β, w ).Then α = q ( i ( r ( α ) , z )) q ( P ( α )) = q ( φ P ( α, z )) = q ( φ P ( β, w )) = q ( i ( r ( β ) , w )) q ( P ( β )) = β. Therefore, i ( r ( α ) , z ) = φ P ( α, z ) P ( α ) − = φ P ( β, w ) P ( α ) − = i ( r ( β ) , w ) P ( β ) P ( α ) − = i ( r ( α ) , w ) , and since i is injective, we have z = w . Thus φ P is injective. To see that φ P is surjective,fix ǫ ∈ Σ. Then q ( ǫ ) = q (cid:0) P ( q ( ǫ )) (cid:1) , and so by Lemma 4.7, there exists a unique z ǫ ∈ T d such that φ P (cid:0) P ( q ( ǫ )) , z ǫ (cid:1) = z ǫ · P ( q ( ǫ )) = i ( r ( ǫ ) , z ǫ ) P ( q ( ǫ )) = ǫ. Thus φ P is surjective, and we have z ǫ = π (cid:0) i − ( ǫ P ( q ( ǫ )) − ) (cid:1) , where π is the projectionof G (0) × T d onto T d . Since φ − P ( ǫ ) = (cid:0) P ( q ( ǫ )) , z ǫ (cid:1) and Lemma 4.3 implies that i − iscontinuous, we deduce that φ − P is continuous, because it is a composition of continuousmaps. Hence φ P is a homeomorphism.To see that φ P is also a homomorphism, fix ( α, β ) ∈ G (2) and z, w ∈ T d . Then, usingpart (b) for the third equality, we have φ P ( α, z ) φ P ( β, w ) = ( z · P ( α ))( w · P ( β ))= ( zw ) · ( P ( α ) P ( β ))= ( zw ) · (cid:0) σ ( α, β ) · P ( αβ ) (cid:1) = (cid:0) σ ( α, β ) zw (cid:1) · P ( αβ )= φ P (cid:0) αβ, σ ( α, β ) zw (cid:1) = φ P (cid:0) ( α, z )( β, w ) (cid:1) . Hence φ P is a groupoid isomorphism. WISTED STEINBERG ALGEBRAS 15
Recall from Example 4.5 that i σ : G (0) × T d → G × σ T d is the inclusion map and q σ : G × σ T d → G is the projection onto the first coordinate. Fix α ∈ G and w ∈ T d .Then( φ P ◦ i σ )( r ( α ) , w ) = φ P ( r ( α ) , w ) = i ( r ( α ) , w ) P ( r ( α )) = i ( r ( α ) , w ) i ( r ( α ) ,
1) = i ( r ( α ) , w ) , and ( q ◦ φ P )( α, w ) = q (cid:0) i ( r ( α ) , w ) P ( α ) (cid:1) = r ( α ) α = α = q σ ( α, w ) . Therefore, Σ and G × σ T d are isomorphic as twists over G . (cid:3) As one might expect, all twists constructed from locally constant 2-cocycles (as inExample 4.5) are topologically trivial, as we now prove.
Lemma 4.9.
Let G be a Hausdorff ´etale groupoid and σ : G (2) → T d be a continuous -cocycle. The twist ( G × σ T d , i, q ) described in Example 4.5 is topologically trivial, andthe map S : γ ( γ, is a continuous global section from G to G × σ T d that induces σ .Proof. It is clear that S is a continuous global section, and so G × σ T d is topologicallytrivial. By Proposition 4.8, S induces a 2-cocycle ω : G (2) → T d satisfying S ( α ) S ( β ) S ( αβ ) − = i (cid:0) r ( α ) , ω ( α, β ) (cid:1) = ( r ( α ) , ω ( α, β )) , for all ( α, β ) ∈ G (2) . To see that S induces σ , fix ( α, β ) ∈ G (2) . Then( r ( α ) , ω ( α, β )) = S ( α ) S ( β ) S ( αβ ) − = ( α, β, αβ, − = ( αβ, σ ( α, β )) (cid:0) ( αβ ) − , σ ( αβ, ( αβ ) − ) (cid:1) = (cid:0) r ( αβ ) , σ ( αβ, ( αβ ) − ) σ ( α, β ) σ ( αβ, ( αβ ) − ) (cid:1) = ( r ( α ) , σ ( α, β )) . Therefore, σ = ω , and so S induces σ . (cid:3) Together, Proposition 4.8 and Lemma 4.9 give us a one-to-one correspondence betweendiscrete twists over a Hausdorff ´etale groupoid G which admit a continuous global sectionand twists over G arising from locally constant 2-cocycles on G .As we shall see in Theorem 4.10, it turns out that all twists over a second-countable,ample, Hausdorff groupoid G admit a continuous global section. We are grateful toElizabeth Gillaspy for alerting us to this folklore fact, citing conversations with AlexKumjian. Because we know of no proofs in the literature, we give a detailed proof herein the discrete setting. Theorem 4.10.
Let G be a second-countable, ample, Hausdorff groupoid, and (Σ , i, q ) bea discrete twist over G . Then Σ is topologically trivial. In order to prove Theorem 4.10, we need the following lemma.
Lemma 4.11.
Let G be a second-countable, ample, Hausdorff groupoid, and suppose that U is an open cover of G . Then U has a countable refinement { B j } ∞ j =1 of mutually disjointcompact open bisections that form a cover of G .Proof. Let U be an open cover of G . By possibly passing to a refinement, we may assumethat U consists of compact open bisections. Since G is second-countable, it is Lindel¨of,and so we may assume that U = { D j } ∞ j =1 , where each D j is a compact open bisection of G . Define B := D , and for each n ≥
2, define B n := D n \ ∪ n − i =1 B i . Then each B j is acompact open bisection contained in D j , and { B j } ∞ i = j forms a disjoint cover of G . (cid:3) Proof of Theorem 4.10.
Since Σ is a twist over the ample groupoid G , for each α ∈ G ,there exists a compact open bisection D α ⊆ G and a continuous local section P α : D α → Σsuch that the map φ α : D α × T d → q − ( D α ) given by φ α ( β, z ) := i ( r ( β ) , z ) P α ( β ) = z · P α ( β )is a homeomorphism. By Lemma 4.11, { D α } α ∈ G has a countable refinement { B j } ∞ j =1 consisting of mutually disjoint compact open bisections that form a cover of G . For each j ≥
1, choose α j ∈ G such that B j ⊆ D α j , and define P j := P α j | B j . For each β ∈ G ,there is a unique j β ≥ β ∈ B j β , and hence the map P : G → Σ given by P ( β ) := P j β ( β ) is well-defined. Since q ( P ( β )) = q ( P j β ( β )) = β = id G ( β ) for all β ∈ G , and P j ( G (0) ∩ B j ) ⊆ Σ (0) for each j ≥ P is a global section. We claim that P is continuous.Let U be an open subset of Σ. Then P − ( U ) = ∪ ∞ j =1 P − j ( U ) = ∪ ∞ j =1 (cid:0) P − α j ( U ) ∩ B j (cid:1) .Since each P α j is continuous and each B j is open, P − ( U ) is open in G . Hence P is acontinuous global section, and Σ is topologically trivial. (cid:3) Twists and 2-cocycles.
In this section we restrict our attention to twists arisingfrom locally constant 2-cocycles, and we investigate the relationships between such twists.In particular, we prove the following theorem.
Theorem 4.12.
Let G be a Hausdorff ´etale groupoid, and σ, τ : G (2) → T d be continuous -cocycles. The following are equivalent: (1) G × σ T d ∼ = G × τ T d ; (2) σ is cohomologous to τ ; and (3) σ is induced by a continuous global section P : G → G × τ T d . We will split the proof of this theorem up into three lemmas. This proof has notableoverlap with [12, Section 4], particularly the equivalence of (2) and (3). However, the twoformulations are sufficiently different to warrant independent treatment here.The following lemma expands on an argument given in [26, Remark 5.1.6] showing thatthe cohomology class of a continuous 2-cocycle σ : G (2) → T d can always be recoveredfrom the twist G × σ T d . Lemma 4.13.
Let G be a Hausdorff ´etale groupoid and τ : G (2) → T d be a continuous -cocycle. Suppose that P : G → G × τ T d is a continuous global section, and σ : G (2) → T d is the induced continuous -cocycle satisfying i (cid:0) r ( α ) , σ ( α, β ) (cid:1) = P ( α ) P ( β ) P ( αβ ) − for all ( α, β ) ∈ G (2) , as in Proposition 4.8. Then σ is cohomologous to τ .Proof. To see that σ is cohomologous to τ , we will find a continuous function b : G → T d such that b ( x ) = 1 for all x ∈ G (0) , and σ ( α, β ) = τ ( α, β ) b ( α ) b ( β ) b ( αβ )for all ( α, β ) ∈ G (2) . For each γ ∈ G , let b ( γ ) be the unique element of T d such that P ( γ ) = ( γ, b ( γ )). Since P ( G (0) ) ⊆ G (0) × { } , we have b ( x ) = 1 for all x ∈ G (0) . Since b = π ◦ P , where π is the projection of G × τ T d onto the second coordinate, b is continuous.For all ( α, β ) ∈ G (2) , we have i (cid:0) r ( α ) , σ ( α, β ) (cid:1) = P ( α ) P ( β ) P ( αβ ) − = ( α, b ( α )) ( β, b ( β )) ( αβ, b ( αβ )) − = (cid:0) αβ, τ ( α, β ) b ( α ) b ( β ) (cid:1)(cid:0) ( αβ ) − , τ ( αβ, ( αβ ) − ) b ( αβ ) (cid:1) = (cid:0) r ( αβ ) , τ ( αβ, ( αβ ) − ) τ ( α, β ) b ( α ) b ( β ) τ ( αβ, ( αβ ) − ) b ( αβ ) (cid:1) = (cid:0) r ( α ) , τ ( α, β ) b ( α ) b ( β ) b ( αβ ) (cid:1) . WISTED STEINBERG ALGEBRAS 17
Thus, noting that i : G (0) × T d → G × σ T d is the inclusion map, we deduce that σ ( α, β ) = τ ( α, β ) b ( α ) b ( β ) b ( αβ ) , as required (cid:3) We now show that cohomologous locally constant 2-cocycles give rise to isomorphictwists.
Lemma 4.14.
Let G be a Hausdorff ´etale groupoid and σ, τ : G (2) → T d be continuous -cocycles. If σ is cohomologous to τ , then the discrete twists G × σ T d and G × τ T d areisomorphic.Proof. If σ is cohomologous to τ , then there is a continuous function b : G → T d such that b ( x ) = 1 for all x ∈ G (0) , and b ( αβ ) σ ( α, β ) = τ ( α, β ) b ( α ) b ( β ) (4.4)for all ( α, β ) ∈ G (2) . Define φ : G × σ T d → G × τ T d by φ ( α, z ) := ( α, b ( α ) z ). Then φ isbijective, with inverse given by φ − ( α, z ) := ( α, b ( α ) z ). Since φ ( α, z ) = ( r ( α ) , b ( α ))( α, z ), φ is continuous, because it is the pointwise product of the continuous map ( r × b ) ◦ π and the identity map, where π is the projection of G × σ T d onto the first coordinate. Asimilar argument shows that φ − is continuous, and thus φ is a homeomorphism.To see that φ is a homomorphism, fix ( α, β ) ∈ G (2) and z, w ∈ T d . Using Equation (4.4)for the third equality, we obtain φ (( α, z )( β, w )) = φ ( αβ, σ ( α, β ) zw )= ( αβ, b ( αβ ) σ ( α, β ) zw )= ( αβ, τ ( α, β ) b ( α ) b ( β ) zw )= ( α, b ( α ) z ) ( β, b ( β ) w )= φ ( α, z ) φ ( β, w ) , as required.Let i σ : G (0) × T d → G × σ T d and i τ : G (0) × T d → G × τ T d be the inclusion maps, and q σ : G × σ T d → G and q τ : G × τ T d → G be the projections onto the first coordinate. Since b ( x ) = 1 for all x ∈ G (0) , we have φ ( i σ ( x, z )) = ( x, b ( x ) z ) = ( x, z ) = i τ ( x, z ) , and q τ ( φ ( α, z )) = q τ ( α, b ( α ) z ) = α = q σ ( α ) , for all x ∈ G (0) , α ∈ G , and z ∈ T d . Therefore, φ is an isomorphism of the twists G × σ T d and G × τ T d . (cid:3) Finally, we show that if σ and τ are locally constant 2-cocycles on G giving rise toisomorphic twists G × σ T d and G × τ T d , then G × τ T d admits a continuous global sectionthat induces σ . Lemma 4.15.
Let G be a Hausdorff ´etale groupoid and σ, τ : G (2) → T d be continuous -cocycles. If ( G × σ T d , i σ , q σ ) and ( G × τ T d , i τ , q τ ) are isomorphic as twists, then σ isinduced by a continuous global section P : G → G × τ T d .Proof. Suppose that φ : G × σ T d → G × τ T d is an isomorphism of twists. By Lemma 4.9,the map S : γ → ( γ,
1) is a continuous global section from G to G × σ T d that induces σ ,in the sense that S ( α ) S ( β ) S ( αβ ) − = i σ (cid:0) r ( α ) , σ ( α, β ) (cid:1) , (4.5)for all ( α, β ) ∈ G (2) . Define P := φ ◦ S : G → G × τ T d . We claim that P is a continuous global section. Since S is a continuous global section and φ is a groupoid isomorphism, P is continuous and P ( G (0) ) ⊆ G (0) ×{ } . Recall from Example 4.5 that q σ : G × σ T d → G and q τ : G × τ T d → G are the projections onto the first coordinate. Since φ is an isomorphism of twists, we have q τ ◦ P = q τ ◦ ( φ ◦ S ) = ( q τ ◦ φ ) ◦ S = q σ ◦ S = id G . Hence P is a continuous global section.We now show that P induces σ . By Proposition 4.8(a), P induces a continuous 2-cocycle ω : G (2) → T d satisfying P ( α ) P ( β ) P ( αβ ) − = i τ (cid:0) r ( α ) , ω ( α, β ) (cid:1) , (4.6)for all ( α, β ) ∈ G (2) . Together, Equations (4.6) and (4.5) imply that i τ (cid:0) r ( α ) , ω ( α, β ) (cid:1) = P ( α ) P ( β ) P ( αβ ) − = φ (cid:0) S ( α ) S ( β ) S ( αβ ) − (cid:1) = φ (cid:0) i σ (cid:0) r ( α ) , σ ( α, β ) (cid:1)(cid:1) = i τ (cid:0) r ( α ) , σ ( α, β ) (cid:1) , for all ( α, β ) ∈ G (2) . Since i σ and i τ are both injective, we deduce that σ = ω , and hence σ is induced by P . (cid:3) We now combine these three lemmas to prove our main theorem for this section.
Proof of Theorem 4.12.
Lemma 4.15 gives (1) = ⇒ (3), Lemma 4.13 gives (3) = ⇒ (2), andLemma 4.14 gives (2) = ⇒ (1). (cid:3) We conclude this section with a corollary of Theorem 4.12.
Corollary 4.16.
Let G be a Hausdorff ´etale groupoid and Σ be a topologically trivialdiscrete twist over G . Suppose that σ , σ : G (2) → T d are continuous -cocycles inducedby two different continuous global sections P , P : Σ → G , as in Proposition 4.8(a). Then σ is cohomologous to σ .Proof. By Proposition 4.8(c), we have G × σ T d ∼ = Σ ∼ = G × σ T d , and hence Theorem 4.12implies that σ is cohomologous to σ . (cid:3) Twisted Steinberg algebras arising from discrete twists.
In this section wegive a construction of a twisted Steinberg algebra A ( G ; Σ) coming from a topologicallytrivial discrete twist Σ over an ample Hausdorff groupoid G . We prove that if two suchtwists are isomorphic, then they give rise to isomorphic twisted Steinberg algebras. Wealso prove that if Σ ∼ = G × σ T d for some continuous 2-cocycle σ : G (2) → T d , then thetwisted Steinberg algebras A ( G ; Σ) and A ( G, σ ) are ∗ -isomorphic. Definition 4.17.
Let G be an ample Hausdorff groupoid and (Σ , i, q ) be a topologicallytrivial discrete twist over G . We say that f ∈ C (Σ , C d ) is T d -equivariant if f ( z · ǫ ) = z f ( ǫ )for all z ∈ T d and ǫ ∈ Σ, and we define A ( G ; Σ) := { f ∈ C (Σ , C d ) : f is T d -equivariant and q (supp( f )) is compact } . We first show that A ( G ; Σ) is a vector space under the pointwise operations inheritedfrom C (Σ , C d ). Lemma 4.18.
Let G be an ample Hausdorff groupoid and (Σ , i, q ) be a topologically trivialdiscrete twist over G . Then A ( G ; Σ) is a linear subspace of C (Σ , C d ) . WISTED STEINBERG ALGEBRAS 19
Proof.
Fix f, g ∈ A ( G ; Σ) and λ ∈ C \{ } . Then λf + g is continuous and T d -equivariant.Since q (cid:0) supp( λf + g ) (cid:1) is contained in the compact set q (supp( f )) ∪ q (supp( g )), we deducethat q (cid:0) supp( λf + g ) (cid:1) has compact closure. Hence λf + g ∈ A ( G ; Σ). (cid:3) Since we are assuming that the twist Σ is topologically trivial, it necessarily admits acontinuous global section P : G → Σ. We now show that Definition 4.17 can be rephrasedin terms of any such P . Lemma 4.19.
Let G be an ample Hausdorff groupoid and (Σ , i, q ) be a topologically trivialdiscrete twist over G . Let P : G → Σ be any continuous global section. Then A ( G ; Σ) = { f ∈ C (Σ , C d ) : f is T d -equivariant and f ◦ P ∈ C c ( G, C d ) } . Proof.
Fix f ∈ C (Σ , C d ). Then f ◦ P is continuous. It suffices to show that q (supp( f )) =supp( f ◦ P ), because then q (supp( f )) is compact if and only if f ◦ P ∈ C c ( G, C d ). ByProposition 4.8(c), we know that Σ = { z · P ( α ) : ( α, z ) ∈ G × T d } . Therefore, we have q (supp( f )) = { q ( ǫ ) : ǫ ∈ Σ , f ( ǫ ) = 0 } = { q ( z · P ( α )) : ( α, z ) ∈ G × T d , f ( z · P ( α )) = 0 } = { α : ( α, z ) ∈ G × T d , z f ( P ( α )) = 0 } = { α ∈ G : ( f ◦ P )( α ) = 0 } = supp( f ◦ P ) , as required. (cid:3) Remarks . (1) It is crucial here that we are dealing with discrete twists. Suppose that σ is a 2-cocycle on an ample Hausdorff groupoid G that is continuous with respect to thestandard topology on T , and consider the classical twist G × σ T over G . Supposethat f ∈ C ( G × σ T ) is a T -equivariant function that is locally constant. Then,for any α ∈ G , there is an open subset V of G containing α and an open subset W of T containing 1 such that f is constant on V × W . Since W is open in thestandard topology on T , we have W = { } . For each z ∈ W \{ } , we have f ( α,
1) = f ( α, z ) = f ( z · ( α, z f ( α, , and hence f | G ×{ } ≡
0. But this implies that f ( β, w ) = 0 for all ( β, w ) ∈ G × σ T ,because f is T -equivariant. In other words, if singleton sets are not open in T , thenthe only locally constant T -equivariant function on G × σ T is the zero function.(2) Suppose that G is an ample Hausdorff groupoid and σ : G (2) → T d is a continuous2-cocycle. Since T d has the discrete topology, nonzero functions in A ( G ; G × σ T d )are not compactly supported. To see this, fix f ∈ A ( G ; G × σ T d ) such that f ( α, w ) = 0 for some ( α, w ) ∈ G × σ T d . Then, for all z ∈ T d , we have f ( α, z ) = f ( α, z w w ) = f (cid:0) ( z w ) · ( α, w ) (cid:1) = z w f ( α, w ) = 0 . Thus { α } × T d is a closed subset of supp( f ) which is not compact, and hence f isnot compactly supported.Note that Definition 4.17 differs from the C ∗ -algebraic analogue defined in [26,Definition 5.1.7 and Theorem 5.1.11], which is the completion of the subalgebra ofcontinuous compactly supported T -equivariant functions on a (non-discrete) twistover G with respect to a C ∗ -norm. Proposition 4.21.
Let G be an ample Hausdorff groupoid and (Σ , i, q ) be a topologicallytrivial discrete twist over G . There is a multiplication on A ( G ; Σ) given by ( f ∗ Σ g )( ǫ ) := X ( δ,η ) ∈ Σ (2) ,δη = ǫ f ( δ ) g ( η ) = X ζ ∈ Σ s ( ǫ ) f ( ǫζ ) g ( ζ − ) , (4.7) and an involution given by f ∗ ( ǫ ) := f ( ǫ − ) . Under these operations, along with pointwise addition and scalar multiplication, A ( G ; Σ) is a ∗ -algebra.We call A ( G ; Σ) the twisted Steinberg algebra associated to the pair ( G, Σ) .Proof. By Lemma 4.18, A ( G ; Σ) is a vector space. To see that A ( G ; Σ) is a ∗ -algebra, wewill just show that it is closed under the involution and convolution, as it is routine tocheck that the multiplication and involution satisfy all of the other necessary properties.We first prove that A ( G ; Σ) is closed under the involution. Fix f ∈ A ( G ; Σ) ⊆ C (Σ , C d ).Then f ∗ is a composition of continuous maps, so f ∗ ∈ C (Σ , C d ). For all z ∈ T d and ǫ ∈ Σ,we have f ∗ ( z · ǫ ) = f (( z · ǫ ) − ) = f ( z · ( ǫ − )) = z f ( ǫ − ) = z f ∗ ( ǫ ) , and so f ∗ is T d -equivariant. Since supp( f ∗ ) = (supp( f )) − and q is a continuous ho-momorphism, we have q (supp( f ∗ )) ⊆ (cid:0) q (supp( f )) (cid:1) − , and hence q (supp( f ∗ )) is compactbecause it is a closed subset of a compact set. Thus f ∗ ∈ A ( G ; Σ).We now prove that A ( G ; Σ) is closed under the convolution. To see this, first notethat since Σ is topologically trivial, it admits a continuous global section P : G → Σ.Moreover, by Proposition 4.8, P induces a continuous 2-cocycle σ : G (2) → T d such thatthe map φ P : G × σ T d → Σ given by φ P ( α, z ) := z · P ( α ) is an isomorphism of twists. Fix f, g ∈ A ( G ; Σ), and define f P := f ◦ P and g P := g ◦ P . By Lemma 4.19, f P and g P areelements of C c ( G, C d ), which is equal (as a vector space) to A ( G, σ ), by Lemma 3.1. Wewill express f ∗ Σ g in terms of f P ∗ g P , which we know is in A ( G, σ ) by Proposition 3.2.Fix ( α, z ) ∈ G × σ T d . ThenΣ s ( z · P ( α )) = { ǫ ∈ Σ : r ( ǫ ) = s ( z · P ( α )) } = { w · P ( β ) : ( β, w ) ∈ G × T d , r ( w · P ( β )) = s ( z · P ( α )) } = { w · P ( β ) : ( β, w ) ∈ G s ( α ) × T d } . Using T d -equivariance and Proposition 4.8(b) for the fourth equality below, we obtain( f ∗ Σ g )( z · P ( α )) = X ζ ∈ Σ s ( z · P ( α )) f (cid:0) ( z · P ( α )) ζ (cid:1) g ( ζ − )= X ( β,w ) ∈ G s ( α ) × T d f (cid:0) ( z · P ( α ))( w · P ( β )) (cid:1) g (cid:0) ( w · P ( β )) − (cid:1) = X ( β,w ) ∈ G s ( α ) × T d f (cid:0) ( zw ) · ( P ( α ) P ( β )) (cid:1) g (cid:0) w · P ( β ) − (cid:1) = X ( β,w ) ∈ G s ( α ) × T d z w f (cid:0) σ ( α, β ) · P ( αβ ) (cid:1) w g (cid:0) σ ( β, β − ) · P ( β − ) (cid:1) = X β ∈ G s ( α ) z σ ( α, β ) f (cid:0) P ( αβ ) (cid:1) σ ( β, β − ) g (cid:0) P ( β − ) (cid:1) = z X β ∈ G s ( α ) σ ( α, β ) σ ( β, β − ) f P ( αβ ) g P ( β − ) . (4.8) WISTED STEINBERG ALGEBRAS 21
We also have ( f P ∗ g P )( α ) = X β ∈ G s ( α ) σ ( αβ, β − ) f P ( αβ ) g P ( β − ) . (4.9)Since σ is normalised and satisfies the 2-cocycle identity, we have σ ( α, β ) σ ( αβ, β − ) = σ ( α, ββ − ) σ ( β, β − ) = σ ( β, β − ) , and hence σ ( α, β ) σ ( β, β − ) = σ ( αβ, β − ) , (4.10)for each β ∈ G s ( α ) . Together, Equations (4.8), (4.9), and (4.10) imply that( f ∗ Σ g )( φ P ( α, z )) = ( f ∗ Σ g )( z · P ( α )) = z ( f P ∗ g P )( α ) . (4.11)Define ψ f,gP : G × σ T d → C d by ψ f,gP ( α, z ) := z ( f P ∗ g P )( α ). Since f P , g P ∈ A ( G, σ ), we have f P ∗ g P ∈ A ( G, σ ) ⊆ C ( G, C d ). Thus ψ f,gP is continuous. Since φ P is a homeomorphism and f ∗ Σ g = ψ f,gP ◦ φ − P , we deduce that f ∗ Σ g ∈ C (Σ , C d ). Taking z = 1 in Equation (4.11)shows that ( f ∗ Σ g ) ◦ P = f P ∗ g P ∈ C c ( G, C d ), and Lemma 4.19 implies that this isequivalent to showing that q (supp( f ∗ Σ g )) is compact. Finally, to see that f ∗ Σ g is T d -equivariant, fix z ∈ T d and ǫ ∈ Σ. Then ǫ = w · P ( β ) for a unique pair ( β, w ) ∈ G × σ T d .Thus, Equation (4.11) implies that ( f ∗ Σ g )( ǫ ) = w ( f P ∗ g P )( β ), and hence( f ∗ Σ g )( z · ǫ ) = ( f ∗ Σ g ) (cid:0) ( zw ) · P ( β ) (cid:1) = z w ( f P ∗ g P )( β ) = z ( f ∗ Σ g )( ǫ ) . Therefore, f ∗ Σ g ∈ A ( G ; Σ), and so A ( G ; Σ) is a ∗ -algebra. (cid:3) We now show that isomorphic twists give rise to isomorphic twisted Steinberg algebras.
Proposition 4.22.
Let G be an ample Hausdorff groupoid, and (Σ , i , q ) and (Σ , i , q ) be topologically trivial discrete twists over G . If φ : Σ → Σ is an isomorphism of twists,then the map Φ : f f ◦ φ is a ∗ -isomorphism from A ( G ; Σ ) to A ( G ; Σ ) .Proof. We first show that f ◦ φ ∈ A ( G ; Σ ) for each f ∈ A ( G ; Σ ). Let P : G → Σ bea continuous global section, and define P := φ ◦ P : G → Σ . Then P is continuous, P ( G (0) ) ⊆ φ (cid:0) Σ (0)1 (cid:1) = Σ (0)2 , and q ◦ P = q ◦ ( φ ◦ P ) = ( q ◦ φ ) ◦ P = q ◦ P = id G . Hence P is a continuous global section. Fix f ∈ A ( G ; Σ ) ⊆ C (Σ , C d ). Since φ iscontinuous, f ◦ φ ∈ C (Σ , C d ). By Lemma 4.6, φ respects the action of T d , and hence the T d -equivariance of f implies that f ◦ φ is T d -equivariant. Moreover, Lemma 4.19 impliesthat f ◦ φ ◦ P = f ◦ P ∈ C c ( G, C d ), and thus f ◦ φ ∈ A ( G ; Σ ).Therefore, there is a map Φ : A ( G ; Σ ) → A ( G ; Σ ) given by Φ( f ) := f ◦ φ . Routinecalculations show that Φ is a ∗ -homomorphism. Furthermore, Φ is bijective with inversegiven by Φ − ( g ) := g ◦ φ − , and hence Φ is a ∗ -isomorphism. (cid:3) By Proposition 4.8, we know that for every topologically trivial twist Σ over an ampleHausdorff groupoid G , there is a continuous 2-cocycle σ : G (2) → T d such that Σ ∼ = G × σ T d .Hence A ( G ; Σ) is ∗ -isomorphic to A ( G ; G × σ T d ), by Proposition 4.22. We now prove that A ( G ; Σ) is also ∗ -isomorphic to A ( G, σ ). Theorem 4.23.
Let G be an ample Hausdorff groupoid and Σ be a topologically trivialdiscrete twist over G . Let P : G → Σ be a continuous global section and let σ : G (2) → T d bethe continuous -cocycle induced by P , as in Proposition 4.8(a). The map ψ : f f ◦ P is a ∗ -isomorphism from A ( G ; Σ) to A ( G, σ ) . Proof.
By Lemma 3.1(a), A ( G, σ ) and C c ( G, C d ) agree as sets, and hence Lemma 4.19implies that A ( G ; Σ) = { f ∈ C (Σ , C d ) : f is T d -equivariant and f ◦ P ∈ A ( G, σ ) } . (4.12)Thus there is a map ψ : A ( G ; Σ) → A ( G, σ ) given by ψ ( f ) := f ◦ P .To see that ψ is injective, suppose that ψ ( f ) = ψ ( g ) for some f, g ∈ A ( G ; Σ). Fix( α, z ) ∈ G × σ T d . Since f and g are T d -equivariant, we have f ( z · P ( α )) = z f ( P ( α )) = z ψ ( f )( α ) = z ψ ( g )( α ) = z g ( P ( α )) = g ( z · P ( α )) . (4.13)By Proposition 4.8(c), we have Σ = { z · P ( α ) : ( α, z ) ∈ G × σ T d } , and so Equation (4.13)implies that f = g , and thus ψ is injective.To see that ψ is surjective, fix h ∈ A ( G, σ ), and recall from Proposition 4.8(c) that themap φ P : G × σ T d → Σ given by φ P ( α, z ) := z · P ( α ) is an isomorphism of twists. Define f : Σ → C d by f ( z · P ( α )) := z h ( α ), and e f : G × σ T d → C d by e f ( α, z ) := z h ( α ). Since h ∈ C ( G, C d ), we have e f ∈ C ( G × σ T d , C d ), and hence f = e f ◦ φ − P ∈ C (Σ , C d ) because φ − P is continuous. For all α ∈ G and z, w ∈ T d , we have f (cid:0) z · ( w · P ( α )) (cid:1) = f (cid:0) ( zw ) · P ( α ) (cid:1) = z w h ( α ) = z f (cid:0) w · P ( α ) (cid:1) , and so f is T d -equivariant. We also have f ◦ P = h ∈ A ( G, σ ), and thus Equation (4.12)implies that f ∈ A ( G ; Σ). Since ψ ( f ) = f ◦ P = h , ψ is surjective.It is clear that ψ is linear. We claim that it is a ∗ -homomorphism. Fix f, g ∈ A ( G ; Σ)and α ∈ G . By Proposition 4.8(b), we have P ( α ) − = σ ( α, α − ) · P ( α − ), and hence ψ ( f ∗ )( α ) = f ∗ ( P ( α )) = f (cid:0) P ( α ) − (cid:1) = f (cid:0) σ ( α, α − ) · P ( α − ) (cid:1) . (4.14)We also have ψ ( f ) ∗ ( α ) = σ ( α, α − ) ψ ( f )( α − ) = σ ( α, α − ) f (cid:0) P ( α − ) (cid:1) = f (cid:0) σ ( α, α − ) · P ( α − ) (cid:1) . (4.15)Together, Equations (4.14) and (4.15) imply that ψ ( f ∗ ) = ψ ( f ) ∗ . In the notation de-fined in the proof of Proposition 4.21, we have ψ ( f ) = f P and ψ ( g ) = g P , and henceEquation (4.11) implies that for all α ∈ G , we have ψ ( f ∗ Σ g )( α ) = ( f ∗ Σ g )( P ( α )) = ( ψ ( f ) ∗ ψ ( g ))( α ) . So ψ ( f ∗ Σ g ) = ψ ( f ) ∗ ψ ( g ), and thus ψ is a ∗ -isomorphism. (cid:3) Corollary 4.24.
Let G be an ample Hausdorff groupoid and σ : G (2) → T d be a continuous -cocycle. There is a ∗ -isomorphism ψ : A ( G ; G × σ T d ) → A ( G, σ ) such that ψ ( f )( γ ) = f ( γ, for all γ ∈ G .Proof. By Lemma 4.9, the map S : γ ( γ,
1) is a continuous global section from G to G × σ T d that induces σ , so the result follows from Theorem 4.23. (cid:3) Remark . If G is an ample Hausdorff groupoid, then G × σ T d is also an ample Hausdorffgroupoid for any continuous 2-cocycle σ : G (2) → T d , and hence there is an associated(untwisted) Steinberg algebra A ( G × σ T d ). As a set, A ( G × σ T d ) is equal to { f ∈ C ( G × σ T d , C d ) : supp( f ) is compact } , and is dense in C ∗ r ( G × σ T d ), by [6, Proposition 4.2] and [28, Proposition 5.7]. Moreover,by Theorem 4.23, we have A ( G ; G × σ T d ) ∼ = A ( G, σ ), and we know from Proposition 3.2that A ( G, σ ) is dense in C ∗ r ( G, σ ). We saw in Remarks 4.20(2) that the only compactlysupported function in A ( G ; G × σ T d ) ⊆ C ( G × σ T d , C d ) is the zero function, and hence A ( G ; G × σ T d ) ∩ A ( G × σ T d ) = { } . WISTED STEINBERG ALGEBRAS 23
However, this does not preclude C ∗ r ( G, σ ) from embedding into C ∗ r ( G × σ T d ). It would beinteresting to know how these two C ∗ -algebras are related.5. Examples of twisted Steinberg algebras
In this section we discuss two important classes of examples of twisted Steinberg alge-bras: twisted group algebras and twisted Kumjian–Pask algebras.5.1.
Twisted discrete group algebras.
Suppose that G is a topological group. (Thatis, G is a group endowed with a topology with respect to which multiplication and inversionare continuous.) Then G is an ample groupoid if and only if G has the discrete topology,in which case any T d -valued 2-cocycle on G is locally constant. One defines a twist overa discrete group G via a split extension by an abelian group A , as in [4, Chapter IV.3].When A = T d , the twist gives rise to a T d -valued 2-cocycle on G , with which one candefine a twisted group algebra over C d . The twisted convolution and involution defined inProposition 3.2 generalise those of classical twisted group algebras over C d , and hence ourtwisted Steinberg algebras generalise these twisted (discrete) group algebras. Interestingquestions still exist about this class of algebras, even for finite groups. See, for example,[20]. Moreover, twisted group C ∗ -algebras (as studied in [21]) have featured prominentlyin the study of C ∗ -algebras associated with groups and group actions; in particular, theyhave proved essential in establishing superrigidity results for certain nilpotent groups (see[8]).5.2. Twisted Kumjian–Pask algebras.
For each finitely-aligned higher-rank graph (or k -graph) Λ, there is both a C ∗ -algebra C ∗ (Λ) called the Cuntz–Krieger algebra (see [23])and a dense subalgebra KP(Λ) called the
Kumjian–Pask algebra (see [7]). Letting G Λ denote the boundary-path groupoid defined in [30], we have C ∗ (Λ) ∼ = C ∗ ( G Λ ) and KP(Λ) ∼ = A ( G Λ ) . Twisted higher-rank graph C ∗ -algebras were introduced and studied in a series of papersby Kumjian, Pask, and Sims [15, 16, 17, 18]. Twisted higher-rank graph C ∗ -algebrasprovide a class of (somewhat) tractable examples that can be used to demonstrate moregeneral C ∗ -algebraic phenomena. See also [1, 11, 27]. We introduce twisted Kumjian–Pask algebras for row-finite higher-rank graphs with no sources using a twisted Steinbergalgebra approach.Let Λ be a row-finite higher-rank graph with no sources and c be a continuous T -valued2-cocycle on Λ, as defined in [17, Definition 3.5]. Then C ∗ (Λ , c ) is the C ∗ -algebra generatedby a universal Cuntz–Krieger (Λ , c )-family, as defined in [17, Definition 5.2]. In [17,Theorem 6.3(iii)], the authors describe how Λ and c give rise to a 2-cocycle σ c : G (2)Λ → T such that C ∗ (Λ , c ) ∼ = C ∗ ( G Λ , σ c ) . By the last two sentences of the proof of [17, Lemma 6.3], the 2-cocycle σ c is normalisedand locally constant. We define KP(Λ , c ) := A ( G Λ , σ c ) , and call this the (complex) twisted Kumjian–Pask algebra associated to the pair (Λ , c ).By Proposition 3.2, KP(Λ , c ) is dense in C ∗ (Λ , c ).In [17, Definition 5.2], Kumjian, Pask, and Sims construct C ∗ (Λ , c ) using a generatorsand relations model involving the same generating partial isometries { t λ : λ ∈ Λ } as C ∗ (Λ), but with the relation t µ t ν = t µν replaced by t µ t ν = c ( µ, ν ) t µν . We expect thatthere is a similar construction of KP(Λ , c ) using these generators and relations, but wedo not pursue this here. A Cuntz–Krieger uniqueness theorem and simplicity of twistedSteinberg algebras of effective groupoids
In this section we extend the Cuntz–Krieger uniqueness theorem and a part of thesimplicity characterisation for Steinberg algebras from [3] to the twisted Steinberg algebrasetting. Throughout this section, we will assume that G is an effective, ample, Hausdorffgroupoid. Theorem 6.1 (Cuntz–Krieger uniqueness theorem) . Let G be an effective, ample, Haus-dorff groupoid, and σ : G (2) → T d be a continuous -cocycle. Suppose that Q is a ring and π : A ( G, σ ) → Q is a ring homomorphism. Then π is injective if and only if π (1 V ) = 0 for every nonempty compact open subset V of G (0) .Proof. It is clear that if π is injective, then π (1 V ) = 0 for every nonempty compactopen subset V of G (0) . Suppose that π is not injective. Then there exists f ∈ A ( G, σ )such that f = 0 and π ( f ) = 0. We aim to find a nonempty compact open subset V of G (0) such that π (1 V ) = 0. Since σ is locally constant, we can use Lemma 3.1(b) towrite f = P D ∈ F a D D , where F is a finite collection of disjoint nonempty compact openbisections of G such that σ ( α − , α ) is constant for all α ∈ D , and a D ∈ C \{ } , for each D ∈ F . Let g := 1 D − f for some D ∈ F . Then g ∈ ker( π ), because π is a homomorphism.Fix α ∈ D , and define c D := σ ( α − , α ) a D = 0. Then g ( s ( α )) = g ( α − α ) = σ ( α − , α ) 1 D − ( α − ) f ( α ) = σ ( α − , α ) a D = c D = 0 . Let g := g | G (0) , and define H := supp( g − g ) ⊆ G \ G (0) . The calculation above impliesthat s ( α ) ∈ supp( g ). Since G is ample and effective, [3, Lemma 3.1] implies that there isa nonempty compact open subset V of supp( g ) ∩ s ( D ) such that V HV = ∅ . Therefore,since supp(1 V ( g − g )1 V ) ⊆ V HV , we have 1 V ( g − g )1 V = 0, and hence1 V g V = 1 V g V = c D V . (6.1)Thus, using that π ( g ) = 0, we deduce from Equation (6.1) that π (1 V ) = c − D π ( c D V ) = c − D π (1 V ) π ( g ) π (1 V ) = 0 , as required. (cid:3) Given a groupoid G , one calls a subset U ⊆ G (0) invariant if, for any γ ∈ G , we have s ( γ ) ∈ U ⇐⇒ r ( γ ) ∈ U. One says that G is minimal if G (0) has no nontrivial open invariant subsets. Equivalently, G is minimal if and only if s ( r − ( x )) = G (0) for every x ∈ G (0) . Theorem 6.2.
Let G be an effective, ample, Hausdorff groupoid and σ : G (2) → T d be acontinuous -cocycle. Then G is minimal if and only if A ( G, σ ) is simple.Proof. Suppose that G is minimal, and let I be a nonzero ideal of A ( G, σ ). Then I isthe kernel of some noninjective ring homomorphism of A ( G, σ ), so Theorem 6.1 impliesthat there is a compact open subset V ⊆ G (0) such that 1 V ∈ I . We claim that theideal generated by 1 V is the whole of A ( G, σ ). Since the twisted convolution product ofcharacteristic functions on the unit space is the same as the untwisted convolution product,the proof follows directly from the arguments used in the proof of [3, Proposition 4.5].For the converse, suppose that G is not minimal. Then there exists a nonempty openinvariant subset U ( G (0) . The set G U := s − ( U ) = { γ ∈ G : s ( γ ) ∈ U } = { γ ∈ G : r ( γ ) ∈ U } WISTED STEINBERG ALGEBRAS 25 is an open subgroupoid of G , so we can view I := A (cid:0) G U , σ | G (2) U (cid:1) as a proper subset of A ( G, σ ). Since U is a nonempty open set and G is ample, we can find a nonempty compactopen bisection B of U , and thus I = { } , because 1 B ∈ I . We claim that I is an ideal of G . Since the vector-space operations are defined pointwise, it is straightforward to showthat I is a subspace. To see that I is an ideal, fix f ∈ I and g ∈ A ( G, σ ). Since U isinvariant, we have supp( f g ) ⊆ supp( f ) supp( g ) ⊆ G U G ⊆ G U , and so f g ∈ I . Similarly, gf ∈ I , and thus I is an ideal. (Note that I is also a ∗ -ideal.) (cid:3) Remark . By [3, Theorem 4.1], the untwisted Steinberg algebra A ( G ) is simple if andonly if G is minimal and effective. Note that Theorem 6.2 does not give necessary andsufficient conditions on G and σ for simplicity of twisted Steinberg algebras. This is ahard problem. We expect, as in the C ∗ -setting of [17, Remark 8.3], that there exist simpletwisted Steinberg algebras for which the groupoid G is not effective.7. Gradings and a graded uniqueness theorem
In this section we describe the graded structure that twisted Steinberg algebras inheritfrom the underlying groupoid, and we prove a graded uniqueness theorem. The argumentsare similar to those used in the untwisted setting (see [6]). Let Γ be a discrete group andsuppose that c : G → Γ is a continuous groupoid homomorphism (or 1 -cocycle ). Then wecall G a graded groupoid , and we define G γ := c − ( γ ) for each γ ∈ Γ. Since c is continuousand Γ is discrete, each G γ is clopen. Since c is a homomorphism, we have G − γ = G γ − and G γ G δ ⊆ G γδ for all γ, δ ∈ Γ. Note that all groupoids are graded with respect to the groupoid homo-morphism into the trivial group.
Proposition 7.1.
Let G be an ample Hausdorff groupoid and σ : G (2) → T d be a con-tinuous -cocycle. Let Γ be a discrete group and c : G → Γ be a continuous groupoidhomomorphism. For each γ ∈ Γ , define the set of homogeneous elements by A ( G, σ ) γ := { f ∈ A ( G, σ ) : supp( f ) ⊆ G γ } . Then A ( G, σ ) is a Γ -graded algebra.Proof. It is clear that A ( G, σ ) γ is a C -submodule of A ( G, σ ), for each γ ∈ Γ. Since A ( G, σ )and A ( G ) agree as vector spaces, [6, Lemma 3.5] implies that every f ∈ A ( G, σ ) can bewritten as a linear combination of homogeneous elements. Thus, to see that A ( G, σ ) = M γ ∈ Γ A ( G, σ ) γ , it suffices to show that any finite collection { f i ∈ A ( G, σ ) γ i : 1 ≤ i ≤ n, and each γ i is distinct from the others } is linearly independent. But this is clear, because supp( f i ) ∩ supp( f j ) = ∅ when i = j .Finally, we have A ( G, σ ) γ A ( G, σ ) δ ⊆ A ( G, σ ) γδ , because supp( f g ) ⊆ supp( f ) supp( g ) and G γ G δ ⊆ G γδ . (cid:3) As in the untwisted case [6, Theorem 5.4], the graded uniqueness theorem follows fromthe Cuntz–Krieger uniqueness theorem. Note that if e is the identity of Γ, then G e is aclopen subgroupoid of G , and so we can identify A ( G, σ ) e with A ( G e , σ ), just as we canidentify A ( G e ) with A ( G ) e . Theorem 7.2 (Graded uniqueness theorem) . Let G be an ample Hausdorff groupoid and σ : G (2) → T d be a continuous -cocycle. Let Γ be a discrete group with identity e , andlet c : G → Γ be a continuous groupoid homomorphism such that the subgroupoid G e is effective. Suppose that Q is a Γ -graded ring and π : A ( G, σ ) → Q is a graded ringhomomorphism. Then π is injective if and only if π (1 K ) = 0 for every nonempty compactopen subset K of G (0) .Proof. It is clear that if π is injective, then π (1 K ) = 0 for every nonempty compact opensubset K of G (0) . Suppose that π is not injective. We claim that there exists f ∈ A ( G e , σ )such that f = 0 and π ( f ) = 0. To see this, fix g ∈ ker( π ) such that g = 0. By the proofof Proposition 7.1, g can be expressed as a finite sum of homogeneous elements; that is, g = P γ ∈ F g γ , where F is a finite subset of Γ, and g γ ∈ A ( G, σ ) γ for each γ ∈ F . Thus X γ ∈ F π ( g γ ) = π (cid:16) X γ ∈ F g γ (cid:17) = π ( g ) = 0 . Since π is graded, we have π ( g γ ) ∈ Q γ for each γ ∈ Γ. Thus each π ( g γ ) = 0, becauseelements of different graded subspaces of Q are linearly independent. Since g = 0, we canchoose γ ∈ F such that g γ = 0. Since g γ is locally constant and G γ is open, there existsa compact open bisection B ⊆ G γ such that g γ ( B ) = { k } , for some k ∈ C d \{ } . Define f := 1 B − g γ ∈ A ( G e , σ ) ∩ ker( π ) . For all α ∈ B , we have f ( s ( α )) = f ( α − α ) = σ ( α − , α ) 1 B − ( α − ) g γ ( α ) = σ ( α, α − ) k = 0 , and hence f = 0. Thus the restriction π e of π to A ( G e , σ ) is not injective.Since G (0) ⊆ G e and we have assumed that the groupoid G e is effective, we can applyTheorem 6.1 to the restricted homomorphism π e to obtain a nonempty compact opensubset K ⊆ G (0) such that π (1 K ) = 0, as required. (cid:3) Acknowledgements.
This research collaboration began as part of the project-orientedworkshop “Women in Operator Algebras” (18w5168) in November 2018, which wasfunded and hosted by the Banff International Research Station. The attendance ofthe first-named author at this workshop was supported by an AustMS WIMSIG CherylE. Praeger Travel Award, and the attendance of the third-named author was supportedby SFB 878 Groups, Geometry & Actions. The research was also funded by the Aus-tralian Research Council grant DP170101821, and by the Deutsche Forschungsgemein-schaft (DFG, German Research Foundation) under Germany’s Excellence Strategy –EXC 2044 – 390685587, Mathematics M¨unster – Dynamics – Geometry – Structure, andunder SFB 878 Groups, Geometry & Actions.
References [1] B. Armstrong and N. Brownlowe,
Product-system models for twisted C ∗ -algebras of topological higher-rank graphs , J. Math. Anal. Appl. (2018), no. 2, 1443–1475.[2] S. Barlak and X. Li, Cartan subalgebras and the UCT problem , Adv. Math. (2017), 748–769.[3] J.H. Brown, L.O. Clark, C. Farthing, and A. Sims,
Simplicity of algebras associated to ´etale groupoids ,Semigroup Forum (2014), no. 2, 433–452.[4] K.S. Brown, Cohomology of Groups , Graduate Texts in Mathematics, vol. 87, Springer, New York,1982.[5] L.O. Clark, R. Exel, and E. Pardo,
A generalized uniqueness theorem and the graded ideal structureof Steinberg algebras , Forum Math. (2018), no. 3, 533–552. WISTED STEINBERG ALGEBRAS 27 [6] L.O. Clark, C. Farthing, A. Sims, and M. Tomforde,
A groupoid generalization of Leavitt pathalgebras , Semigroup Forum (2014), no. 3, 501–517.[7] L.O. Clark and Y.E.P. Pangalela, Kumjian–Pask algebras of finitely aligned higher-rank graphs , J.Algebra (2017), 364–397.[8] C. Eckhardt and S. Raum, C ∗ -superrigidity of -step nilpotent groups , Adv. Math. (2018),175–195.[9] R. Exel, Inverse semigroups and combinatorial C ∗ -algebras , Bull. Braz. Math. Soc. (N.S.) (2008),no. 2, 191–313.[10] , Reconstructing a totally disconnected groupoid from its ample semigroup , Proc. Amer. Math.Soc. (2010), no. 8, 2991–3001.[11] E. Gillaspy, K -theory and homotopies of -cocycles on higher-rank graphs , Pacific J. Math. (2015), no. 2, 407–426.[12] A. Kumjian, On C ∗ -diagonals , Canad. J. Math. (1986), no. 4, 969–1008.[13] A. Kumjian and D. Pask, Higher rank graph C ∗ -algebras , New York J. Math. (2000), 1–20.[14] A. Kumjian, D. Pask, I. Raeburn, and J. Renault, Graphs, groupoids, and Cuntz–Krieger algebras ,J. Funct. Anal. (1997), no. 2, 505–541.[15] A. Kumjian, D. Pask, and A. Sims,
Homology for higher-rank graphs and twisted C ∗ -algebras , J.Funct. Anal. (2012), no. 6, 1539–1574.[16] , On the K -theory of twisted higher-rank-graph C ∗ -algebras , J. Math. Anal. Appl. (2013),no. 1, 104–113.[17] , On twisted higher-rank graph C ∗ -algebras , Trans. Amer. Math. Soc. (2015), no. 7,5177–5216.[18] , Simplicity of twisted C ∗ -algebras of higher-rank graphs and crossed products by quasifreeactions , J. Noncommut. Geom. (2016), no. 2, 515–549.[19] X. Li, Every classifiable simple C ∗ -algebra has a Cartan subalgebra , Invent. Math. (2019), 1–46, DOI10.1007/s00222-019-00914-0.[20] L. Margolis and O. Schnabel, Twisted group ring isomorphism problem , Q. J. Math. (2018), no. 4,1195–1219.[21] J.A. Packer and I. Raeburn, On the structure of twisted group C ∗ -algebras , Trans. Amer. Math. Soc. (1992), no. 2, 685–718.[22] A.L.T. Paterson, Groupoids, inverse semigroups, and their operator algebras , Progress in Mathemat-ics, vol. 170, Birkh¨auser Boston, Inc., Boston, MA, 1999.[23] I. Raeburn, A. Sims, and T. Yeend,
The C ∗ -algebras of finitely aligned higher-rank graphs , J. Funct.Anal. (2004), no. 1, 206–240.[24] J. Renault, A groupoid approach to C ∗ -algebras , Lecture Notes in Math., vol. 793, Springer-Verlag,New York, 1980.[25] , Cartan subalgebras in C ∗ -algebras , Irish Math. Soc. Bulletin (2008), 29–63.[26] A. Sims, Hausdorff ´etale groupoids and their C ∗ -algebras , Operator algebras and dynamics: group-oids, crossed products and Rokhlin dimension, Advanced Courses in Mathematics, CRM Barcelona,Birkh¨auser, to appear, arXiv:1710.10897v2 [math.OA].[27] A. Sims, B. Whitehead, and M.F. Whittaker, Twisted C ∗ -algebras associated to finitely alignedhigher-rank graphs , Documenta Math. (2014), 831–866.[28] B. Steinberg, A groupoid approach to discrete inverse semigroup algebras , Adv. Math. (2010),no. 2, 689–727.[29] J.L. Tu,
La conjecture de Baum-Connes pour les feuilletages moyennables , K -Theory (1999),no. 3, 215–264.[30] T. Yeend, Groupoid models for the C ∗ -algebras of topological higher-rank graphs , J. Operator Theory (2007), no. 1, 95–120. (B. Armstrong and J. Ramagge) School of Mathematics and Statistics, The University ofSydney, NSW 2006, AUSTRALIA
E-mail address : becky.armstrong, [email protected] (L.O. Clark) School of Mathematics and Statistics, Victoria University of Wellington,PO Box 600, Wellington 6140, NEW ZEALAND
E-mail address : [email protected] (K. Courtney) Mathematical Institute, WWU M¨unster, Einsteinstr. 62, 48149 M¨unster,GERMANY
E-mail address : [email protected] (Y.-F. Lin) Mathematical Sciences Research Centre, Queen’s University Belfast, Bel-fast, BT7 1NN, UNITED KINGDOM
E-mail address : [email protected] (K. McCormick) Department of Mathematics, University of Minnesota, Minneapolis, MN,UNITED STATES
E-mail address ::