0% found this document useful (0 votes)
20 views13 pages

2410 Zeous 02

This document presents a homotopy rigidity theorem for simple, separable, nuclear, and Z0-stable C*-algebras, establishing that such algebras are isomorphic if they are trace-preservingly homotopy equivalent. The result does not rely on the Universal Coefficient Theorem (UCT) and serves as a stably projectionless analog of existing homotopy rigidity theorems. The authors also discuss the implications of this theorem in the context of classification results for C*-algebras without assuming the UCT.

Uploaded by

fisex12441
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
20 views13 pages

2410 Zeous 02

This document presents a homotopy rigidity theorem for simple, separable, nuclear, and Z0-stable C*-algebras, establishing that such algebras are isomorphic if they are trace-preservingly homotopy equivalent. The result does not rely on the Universal Coefficient Theorem (UCT) and serves as a stably projectionless analog of existing homotopy rigidity theorems. The authors also discuss the implications of this theorem in the context of classification results for C*-algebras without assuming the UCT.

Uploaded by

fisex12441
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

A HOMOTOPY RIGIDITY THEOREM FOR Z0 -STABLE

C∗ -ALGEBRAS

JORGE CASTILLEJOS, BAUKJE DEBETS, AND GÁBOR SZABÓ


arXiv:2505.04857v1 [math.OA] 7 May 2025

Abstract. We show that two simple, separable, nuclear and Z0 -stable


C∗ -algebras are isomorphic if they are trace-preservingly homotopy equiv-
alent. This result does not assume the UCT and can be viewed as a tra-
cial stably projectionless analog of the homotopy rigidity theorem for
Kirchberg algebras.

Introduction
A milestone in the classification programme of C∗ -algebras states that
the class of simple separable nuclear unital and Z-stable C∗ -algebras that
satisfy the Universal Coefficient Theorem [RS87] (UCT) are classified by an
invariant constructed with K-theory and tracial data [GLN20a, GLN20b,
EGLN24, TWW17, CET+ 21, CGS+ 23]. Therefore, determining when a
C∗ -algebra satisfies these conditions is essential before one can attempt to
apply the classification theorem. Many criteria exist to detect Z-stability in
various contexts, and it is also known that there are many separable simple
unital and nuclear C∗ -algebras that are not Z-stable. However, a major
open problem in the field is determining if nuclear C∗ -algebras satisfy the
UCT. We refer to [RS87, Ska88] for a comprehensive description of the UCT.
The work of Tu shows that groupoid C∗ -algebras of a-T-menable groupoids
(in particular amenable ones) satisfy the UCT ([Tu99]), and this encom-
passes large classes of separable nuclear C∗ -algebras. Nevertheless, all con-
crete known examples of nuclear C∗ -algebras are seen to satisfy the UCT
and there is no apparent candidate for a counterexample (although such
examples do exist outside the nuclear setting; see [Ska88]).
In light of the difficulty of settling the UCT problem, it is natural to seek
classification results where the UCT is not needed. For instance, the clas-
sification theorem of Kirchberg and Phillips [Kir, Phi00, KP00] for simple
nuclear purely infinite C∗ -algebras has as a major methodological advantage
that the classification is initially obtained directly via KK-theory. The UCT
only plays a role when one wants to obtain KK-equivalence from an isomor-
phism of ordinary K-theory. In the classification of stably finite C∗ -algebras,
all the available theories utilise the UCT assumption in several substantial
intermediate steps. This could be relaxed recently in a breakthrough ar-
ticle of Schafhauser [Sch24], in which an isomorphism theorem was proved
for unital simple nuclear Z-stable C∗ -algebras under the assumption that
one starts from an embedding that induces both a KK-equivalence and an
isomorphism of tracial data.
1
2 J. CASTILLEJOS, B. DEBETS, AND G. SZABÓ

In this note, we will focus on classification results without assuming the


UCT for simple and stably projectionless nuclear C∗ -algebras. In this direc-
tion, an important result for this note is the classification of KK-contractible
(i.e. KK-equivalent to {0}) stably projectionless simple separable and Z-
stable C∗ -algebras via their tracial cones and scales [EGLN20, Theorem 7.5].
In particular, this applies to simple separable stably projectionless nuclear
C∗ -algebras with traces that absorb tensorially the Razak–Jacelon algebra
W [Raz02, Jac13]. Utilizing this result as a cornerstone, we establish a
Z0 -stable uniqueness theorem for ∗ -homomorphisms between simple separa-
ble nuclear C∗ -algebras that are trace-preservingly homotopic (see Definition
2.4 and Theorem 2.5). Here Z0 is the known stably projectionless analog
of the Jiang–Su algebra that plays an important role in the classification
of a large class of stably projectionless C∗ -algebras [GL20]. For classifiable
C∗ -algebras with the UCT, the assumption of Z0 -stability is reflected in the
Elliott invariant via the assumption that the pairing map between the traces
and the K0 -group has to vanish.
By combining our aforementioned uniqueness theorem for maps with an
Elliott intertwining argument, we obtain the following rigidity property for
the class of separable, simple, nuclear and Z0 -stable C∗ -algebras.
Theorem A. Let A and B be simple, separable, nuclear and Z0 -stable C∗ -
algebras. If A and B are trace-preservingly homotopy equivalent, then A is
isomorphic to B.
Acknowledgement. JC was supported by UNAM–PAPIIT IA103124. BD
was partially supported by European Research Council Consolidator Grant
614195–RIGIDITY. GS was supported by research project G085020N funded
by the Research Foundation Flanders (FWO), and the European Research
Council under the European Union’s Horizon Europe research and innova-
tion programme (ERC grant AMEN–101124789). Both BD and GS were
partially supported by KU Leuven internal funds projects STG/18/019 and
C14/19/088.

1. Preliminaries
1.1. Notation. We will denote the multiplier algebra of A by M(A) and
its forced unitisation by A† . If A is unital, the unitary group is denoted by
U (A). We will write U (1 + A) for the unitary subgroup (1 + A) ∩ U (A† ).
Observe that if A is unital, we can canonically identify U (A) with U (1 + A).
We will denote the n×n-matrices with complex coefficients by Mn (C). We
denote the standard matrix units by (eij )ni,j=1 . We will also freely identify
Mn (A) with Mn (C) ⊗ A whenever it is convenient for us. The cone of lower
semicontinuous densely defined traces on A is denoted by T + (A).
We will frequently write a ≈ε b as short-hand for ka − bk ≤ ε. For
∗ -homomorphisms ϕ, ψ : A → B, we write ϕ ≈ ψ to say that they are
u
approximately unitarily equivalent, i.e., there is a net of unitaries (uλ ) ⊂
U (1 + A) with lim uλ ϕ(a)u∗λ = ψ(a) for all a ∈ A. If one assumes A to be
λ→∞
separable, then such nets can be replaced by sequences.
Lastly, we shall say that a separable C∗ -algebra A is KK-contractible if
KK(A, A) = 0.
A HOMOTOPY RIGIDITY THEOREM FOR Z0 -STABLE C∗ -ALGEBRAS 3

1.2. Robert’s classification theorem. Given two positive elements a and


b in a C∗ -algebra A, it is said that a is Cuntz-below b, a - b, if for any ε > 0
there is x ∈ A such that a ≈ε x∗ bx. It is said that a is Cuntz equivalent
to b, a ∼ b, if a - b and b - a. The Cuntz semigroup of A is defined as
Cu(A) := (A⊗K)+ /∼ equipped with orthogonal addition and order given by
Cuntz-subequivalence. The equivalence class of a given element a ∈ (A⊗K)+
will be denoted by [a].
The augmented Cuntz semigroup of a unital C∗ -algebra A, denoted by
Cu∼ (A), is defined as the ordered semigroup of formal differences [a]−n[1A ],
with a ∈ Cu(A) and n ∈ N, i.e.,
Cu∼ (A) := {[a] − n[1A ] | [a] ∈ Cu(A), n ∈ N}.
This set carries an order by declaring that [a] − n[1A ] ≤ [b] − m[1A ] holds in
Cu∼ (A) if there is k ∈ N such that [a] + m[1A ] + k[1A ] ≤ [b] + n[1A ] + k[1A ]
in Cu(A).
Now assume A is non-unital and π : A† → C is the canonical quotient
map. The augmented Cuntz semigroup of A is then defined as the subsemi-
group of Cu∼ (A† ) given by
Cu∼ (A) := {[a] − n[1A† ] | [a] ∈ Cu(A† ), Cu(π)([a]) = n}.
We endow Cu∼ (A) with the order coming from Cu∼ (A† ). We refer to
[Rob12, RS21] for more details about this construction. As we shall see
below, the augmented Cuntz semigroup is a powerful tool to classify maps
between certain classes of C∗ -algebras.
When a C∗ -algebra is simple, exact, Z-stable and admits non-trivial
traces, Cu∼ (A) can be calculated using its K-theory and tracial data ([RS21,
Theorem 6.11]). Indeed, if R := R ∪ {∞} and Lsc(T + (A), R) denotes the
lower semicontinuous functions T + (A) → R that are linear and map the
zero trace to 0, one always has a natural isomorphism
Cu∼ (A) ∼
= K0 (A) ⊔ Lsc(T + (A), R). (1.1)
The ordered semigroup structure on the right hand side is given as follows.
The sets K0 (A) and Lsc(T + (A), R) are considered disjoint and each set is
separately endowed with its natural order and addition. For x ∈ K0 (A), we
consider the function x̂ : T + (A) → R ⊂ R given by evaluation on x. For
any f ∈ Lsc(T + (A), R), the addition operation of mixed terms is given via
x + f := x̂ + f ∈ Lsc(T + (A)). The order is defined by declaring that f ≤ x
holds if f ≤ x̂ in Lsc(T + (A), R), and x ≤ f if there is a strictly positive
function h ∈ Lsc(T + (A), R) such that x̂ + h = f . We refer to [RS21, Section
6.3] for more details about the above isomorphism.
A major tool for this paper is a classification result by Robert that ap-
plies to ∗ -homomorphisms where the domain is an inductive limit of 1-
dimensional noncommutative CW complexes (henceforth abbreviated as 1-
NCCW complexes) with vanishing K1 -groups and the codomain has stable
rank one (i.e., invertible elements are dense in its minimal unitisation). The
class of 1-NCCW complexes was introduced by Eilers–Loring–Pedersen in
[ELP98]. These algebras are defined as pullback C∗ -algebras of the form
C([0, 1], F ) ⊕F ⊕F E with the linking morphism C([0, 1], F ) → F ⊕ F given
by ev0 ⊕ ev1 .
4 J. CASTILLEJOS, B. DEBETS, AND G. SZABÓ

Theorem 1.1 ([Rob12, Theorem 1.0.1]). Let A be either a 1-NCCW com-


plex with trivial K1 -group, or a sequential inductive limit of such C∗ -algebras,
or a C∗ -algebra stably isomorphic to one such inductive limit. Let B be a
C∗ -algebra with stable rank one. Then for every Cu-morphism
α : Cu∼ (A) → Cu∼ (B)
such that α([sA ]) ≤ [sB ], where sA ∈ A+ and sB ∈ B+ are strictly positive
elements, there exists a ∗ -homomorphism
ϕ:A→B

such that Cu (ϕ) = α. Moreover, ϕ is unique up to approximate unitary
equivalence.
We shall henceforth refer to the class of C∗ -algebras that satisfy the as-
sumptions of the above theorem in place of A as Robert’s class. For subse-
quent applications of the theorem we note that every C∗ -algebra belonging
to Robert’s class also has stable rank one.
1.3. The Razak–Jacelon algebra W and the C∗ -algebra Z0 . Among
the class of inductive limits of 1-NCCW complexes with vanishing K1 -
groups, there are two important examples with remarkable properties that
we discuss below.
The Razak–Jacelon algebra W is the unique algebraically simple nuclear
stably projectionless Z-stable monotracial and KK-contractible C∗ -algebra
up to isomorphism. It is constructed as an inductive limit of certain 1-
NCCW complexes known as Razak building blocks [Raz02, Jac13]. This
algebra absorbs tensorially the universal UHF algebra and hence it is also
Z-stable. Via classification results ([EGLN20, Corollary 6.7]), it is known
that it is self-absorbing, i.e., W ⊗ W ∼
= W. Furthermore, by the Kirchberg–
Phillips classification theorem, one sees also that W ⊗ O∞ ∼ = O2 ⊗ K. The
C∗ -algebra W can be regarded as the stably finite analogue of O2 .
The Razak–Jacelon algebra belongs to the broader class of stably pro-
jectionless simple nuclear Z-stable KK-contractible C∗ -algebras. This class
was classified by Elliott–Gong–Lin–Niu ([EGLN20, Theorem 7.5]). (Note
that [CE20] guarantees that the assumption of Z-stability agrees with the
assumption of finite nuclear dimension appearing in this reference.) In par-
ticular, this applies to simple stably projectionless nuclear C∗ -algebras that
absorb tensorially the Razak–Jacelon algebra W. An important result for
this note is that ∗ -homomorphisms between KK-contractible C∗ -algebras
are classified by their tracial behaviour.
Theorem 1.2 ([Sza21, Theorem 6.3], [EGLN20]). Let A and B be sim-
ple separable stably projectionless nuclear Z-stable C∗ -algebras such that
KK(A, A) = KK(B, B) = 0. Let ψ, ϕ : A → B be ∗ -homomorphisms. Then
ϕ and ψ are approximately unitarily equivalent if and only if τ ◦ ψ = τ ◦ ϕ
for all τ ∈ T + (B).
On the other hand, the C∗ -algebra Z0 is a stably projectionless C∗ -algebra
that is also an inductive limit of 1-NCCW complexes with vanishing K1 -
groups. It is the unique (up to isomorphism) algebraically simple and mono-
tracial C∗ -algebra in Robert’s class with K0 (Z0 ) = Z. (As it belongs to
A HOMOTOPY RIGIDITY THEOREM FOR Z0 -STABLE C∗ -ALGEBRAS 5

Robert’s class, one automatically has K1 (Z0 ) = 0.) It has interesting and
useful properties like being self-absorbing and Z-stable ([GL20, Definition
8.1, Corollary 13.4]) and it can also be regarded as a stably projectionless
analog of Z. In [GL20, Theorem 15.8], it was shown that separable, simple,
nuclear and Z0 -stable C∗ -algebras satisfying the UCT are classified by the
Elliott invariant, which in this case obeys the condition that the pairing
between traces and the K0 -group has to be trivial.
The C∗ -algebras W and Z0 both belong to Robert’s class. As an appli-
cation, one can produce a useful interplay between these algebras by con-
structing special ∗ -homomorphisms Z0 → W and W → Z0 . To be more
precise, equation (1.1) yields that
Cu∼ (Z0 ) ∼
=Z⊔R and Cu∼ (W) ∼
= {0} ⊔ R. (1.2)
One can then define order preserving maps between the augmented Cuntz
semigroups in the following way: the map W → Z0 is induced by the nat-
ural inclusion Cu∼ (W) ֒→ Cu∼ (Z0 ) and the ∗ -homomorphism Z0 → W is
induced by the map that is equal to the identity on R and sends Z to 0.
It follows from the construction of these maps that they vanish in K0 and
preserve the corresponding tracial state.
Theorem 1.3 (cf. [GL20, Definition 8.12]). There exist unique trace pre-
serving ∗ -homomorphisms ϕZ0 : W → Z0 and ϕW : Z0 → W (up to approx-
imate unitary equivalence). In particular,
K0 (ϕW ) = 0 and K0 (ϕZ0 ) = 0.
There is another useful automorphism of Z0 that we will use in the next
section. This automorphism is obtained from defining a map Λ at the level
of Cu∼ (Z0 ) that sends n to −n on Z and agrees with the identity map
on R. By Robert’s classification theorem, there exists a ∗ -endomorphism
σ̃ : Z0 → Z0 that induces Λ. Since by the same theorem one has that σ̃ 2 is
approximately inner, it follows from the Elliott intertwining argument that
σ̃ is approximately unitarily equivalent to an automorphism; see [Rør01,
Corollary 2.3.4].
Theorem 1.4 (cf. [GL20, Definition 8.13]). There exists a unique trace pre-
serving automorphism σ : Z0 → Z0 (up to approximate unitary equivalence)
such that K0 (σ) = −idK0 (Z0 ) .
We include the statement of the following lemma proved in [GL20], which
is another application of Robert’s classification theorem.
Lemma 1.5 ([GL20, Lemma 8.14]). Consider the ∗ -homomorphisms Υ, Ω :
Z0 → M2 (Z0 ) given by
Υ := idZ0 ⊕ σ and Ω := (ϕZ0 ◦ ϕW ) ⊗ 1M2 . (1.3)
Then Υ is approximately unitarily equivalent to Ω.

2. The Main result


In this section we prove the main result of this note. We begin by in-
troducing some notation that will be used throughout. Given n ∈ N and
6 J. CASTILLEJOS, B. DEBETS, AND G. SZABÓ

j ∈ {1, 2, . . . , 2n + 1}, we consider the ∗ -homomorphism


ιA
j : A → M2n+1 (A) by ιA
j (a) = eii ⊗ a. (2.1)
For j ∈ {1, . . . , 2n}, we consider κAj : M2 (A) → M2n+1 (A) by
  X
κA
j (akℓ )k,ℓ=0,1 = ej+k,j+ℓ ⊗ akℓ . (2.2)
k,ℓ=0,1
We will typically omit the superscript A when it is clear from context which
algebra we are referring to.
We note that by the classification theorem [EGLN20, Theorem 15.8], it
follows that Z0 is isomorphic to Mn (Z0 ) for all n ∈ N. We say that an
isomorphism Φ : Z0 → Mn (Z0 ) is K-positive, if K0 (Φ) = K0 (ι1 ). Note
that due to the existence of the aforementioned automorphism σ on Z0 ,
K-positivity is not automatic in this context.
Lemma 2.1. Let n ∈ N. Consider the ∗ -homomorphism Γn : Z0 →
M2n+1 (Z0 ) given by
n
X n
X
Γn (a) := ι2j+1 (a) + ι2j (σ(a)).
j=0 j=1

Then any K-positive isomorphism Z0 → M2n+1 (Z0 ) is approximately uni-


tarily equivalent to Γn .
Proof. Notice that Γn (a) is a matrix of the form
 
a
 σ(a) 
 
 a 
Γn (a) =  .
 
 σ(a) 
 .. 
 . 
a
Since K0 (ιk ) = K0 (ι0 ) for all k ≤ 2n + 1 and K0 (σ) = −idK0 (Z0 ) , we see
that
 
Xn n
X
K0 (Γn ) = K0  ι2j+1 ◦ idZ0 + ι2j ◦ σ 
j=0 j=1
n
X n
X
= K0 (ι1 ) − K0 (ι1 )
j=0 j=1
= K0 (ι1 ).
On the other hand, the uniqueness of the tracial states τZ0 and tr2n+1 on
Z0 and M2n+1 (C), respectively, yields that the unique trace on M2n+1 (Z0 )
is of the form tr2n+1 ⊗ τZ0 . Hence, using that σ is a trace-preserving auto-
morphism (i.e. τZ0 ◦ σ = τZ0 ), we obtain
(tr2n+1 ⊗ τZ0 ) ◦ Γn = τZ0 .
If we keep in mind (1.2), the above implies that Cu∼ (Γn ) = Cu∼ (Φ) for any
K-positive isomorphism Φ : Z0 → M2n+1 (Z0 ). By Robert’s Theorem 1.1,
Γn is thus approximately unitarily equivalent to Φ. 
A HOMOTOPY RIGIDITY THEOREM FOR Z0 -STABLE C∗ -ALGEBRAS 7

Lemma 2.2. Let A be a σ-unital C∗ -algebra and assume that a unitary


u ∈ U (M(A)) homotopic to 1M(A) . Then there exists a sequence (vn )n∈N ⊂
U (1 + A) such that (Ad(vn ))n∈N converges to Ad(u) in the point-norm topol-
ogy.
Proof. This statement is a consequence of a special case of [GS22, Lemma
4.3], applied to D = 0, the trivial action in place of β, and for the C∗ -algebra
A in place of B. 
We now proceed with the key technical lemma of this note, which is a
non-unital and stably finite version of the reduction argument for homotopic
maps in [Sza21, Lemma 5.10] (this argument originates in [Phi97]). We will
use the following standard notation: given a ∗ -homomorphism Φ : A →
C([0, 1], B) and t ∈ [0, 1], we write Φt (a) := Φ(a)(t). The basic idea of the
proof will be to use the maps ϕW and ϕZ0 (see Theorem 1.3) to move a
given unitary equivalence of Φs ⊗ idW and Φt ⊗ idW from B ⊗ W to B ⊗ Z0
combined with a handy identification of Z0 with M2n+1 (Z0 ).
Lemma 2.3. Let A and B be separable C∗ -algebras. Let Φ : A → C([0, 1], B)
be a ∗ -homomorphism. Suppose that
Φs ⊗ idW ≈u Φt ⊗ idW for all s, t ∈ [0, 1].
Then
Φ0 ⊗ idZ0 ≈u Φ1 ⊗ idZ0 . (2.3)
Proof. Let F ⊂ A ⊗ Z0 be a finite set and ε > 0. We want to find a unitary
in B ⊗ Z0 that (F, ε)-approximately conjugates the first map in (2.3) onto
the second. Without loss of generality, we may assume that there exist finite
sets of contractions FA ⊂ A and FZ0 ⊂ Z0 such that
F = {a ⊗ z | a ∈ FA , z ∈ F} .
By uniform continuity, there exists n ∈ N such that if s, t ∈ [0, 1] are arbi-
trary parameters with |s − t| ≤ n1 then
Φs (a) ≈ε/9 Φt (a), a ∈ FA . (2.4)
By assumption, one has for each j ∈ {1, . . . , n} a unitary vj ∈ U (1 + B ⊗ W)
such that
vj (Φ0 (a) ⊗ ϕW (z)) vj∗ ≈ε/9 Φ j (a) ⊗ ϕW (z)
n

for all a ∈ FA and z ∈ FZ0 . Similarly, there are unitaries uj ∈ U (1 + B ⊗ W)


for j ∈ {0, 1, . . . , n − 1} such that
 
uj Φ j (a) ⊗ ϕW (z) u∗j ≈ε/9 Φ1 (a) ⊗ ϕW (z), (2.5)
n

for all a ∈ FA and z ∈ FZ0 .


We will use the maps ϕW : Z0 → W, ϕZ0 : W → Z0 , σ : Z0 → Z0 and
Ω, Υ : Z0 → M2 (Z0 ) introduced in Section 1.3. By Lemma 1.5, the maps
Ω and Υ are approximately unitarily equivalent. Hence, there is a unitary
u ∈ U (1 + M2 (Z0 )) such that
uΩ(z)u∗ ≈ε/9 Υ(z), z ∈ F Z0 . (2.6)
8 J. CASTILLEJOS, B. DEBETS, AND G. SZABÓ

Consider the unitary


 
0 1M(Z0 )
∈ M2 (M(Z0 )),
1M(Z0 ) 0
which is homotopic to 1M2 (M(Z0 )) . We set Υ′ : Z0 → M2 (Z0 ) by
   
′ 0 1M(Z0 ) σ(a) 0
Υ (z) := Ad ◦ Υ(z) = . (2.7)
1M(Z0 ) 0 0 a
It follows from Lemma 2.2 that Υ ≈u Υ′ . The approximate unitary equiv-
alence between Ω and Υ entails Ω ≈u Υ′ . Then, there exists a unitary
v ∈ U (1 + M2 (Z0 )) such that
vΩ(z)v∗ ≈ε/9 Υ′ (z). (2.8)
We define θ, θ ′ : B ⊗ W → B ⊗ M2 (Z0 ) by
θ := idB ⊗ (Ad(u) ◦ (ϕZ0 ⊗ 1M2 ))
and
θ ′ := idB ⊗ (Ad(v) ◦ (ϕZ0 ⊗ 1M2 )).
In particular, these yield
(2.6)
θ(b ⊗ ϕW (z)) = b ⊗ uΩ(z)u∗ ≈ε/9 b ⊗ Υ(z) (2.9)
and
(2.8)
θ ′ (b ⊗ ϕW (z)) = b ⊗ vΩ(z)v∗ ≈ε/9 b ⊗ Υ′ (z)
for z ∈ FZ0 and any contraction b ∈ B. Therefore
  (2.9)  
θ(uj ) Φ j (a) ⊗ Υ(z) θ(uj )∗ ≈ε/9 θ(uj )θ Φ j (a) ⊗ ϕW (z) θ(uj )∗
n
  n  
= θ uj Φ j (a) ⊗ ϕW (z) u∗j
n

(2.5)
≈ε/9 θ(Φ1 (a) ⊗ ϕW (z))
(2.9)
≈ε/9 Φ1 (a) ⊗ Υ(z), (2.10)
for j ∈ {0, . . . , n − 1}. A similar calculations shows
θ ′ (vj ) Φ0 (a) ⊗ Υ′ (z) θ ′ (vj )∗ ≈ε/3 Φ j (a) ⊗ Υ′ (z)

(2.11)
n

for j ∈ {1, . . . , n}.


With the notation introduced in (2.1) and (2.2), we define unitaries in
M2n+1 (B ⊗ Z0 ) in the following way:
n
X
V := e11 ⊗ 1M(B⊗Z0 ) + κ2j (θ ′ (vj )),
j=1
n
X
U := κ2j−1 (θ(uj−1 )) + e2n+1,2n+1 ⊗ 1M(B⊗Z0 ) .
j=1
A HOMOTOPY RIGIDITY THEOREM FOR Z0 -STABLE C∗ -ALGEBRAS 9

Schematically, these unitaries correspond to block-diagonal matrices of the


form
 
1M(B⊗Z0 )

 θ ′ (v1 ) 


θ (v2 )
V =
 

 . . 
 . 
θ ′ (vn )
and
 
θ(u0 )

 θ(u1 ) 

U =
 .. .

 . 
 θ(un−1 ) 
1M(B⊗Z0 )
If b ∈ B ⊗ Z0 and x0 , x1 , . . . , xn ∈ M2 (B ⊗ Z0 ), then
 
n
X Xn

V ι1 (b) +
 κ2j (xj ) V = ι1 (b) +
 κ2j (θ ′ (vj )xj θ ′ (vj )∗ ) (2.12)
j=1 j=1

and
 
Xn
U κ2j−1 (xj−1 ) + ι2n+1 (b) U ∗
j=1
n
X
= κ2j−1 (θ(uj−1 )xj−1 θ(uj−1 )∗ ) + ι2n+1 (b). (2.13)
j=1

We will now employ the ∗ -homomorphism Γn : Z0 → M2n+1 (Z0 ) given in


Lemma 2.1 as
Xn n
X
Γn (a) := ι2j+1 (a) + ι2j (σ(a)).
j=0 j=1

Observe that we can write Γn using either Υ or Υ′ in the following way


   
Υ(a) a
 ..   Υ′ (a) 
Γn (a) = 
 . =
 
.

(2.14)
. .
 Υ(a)   . 
a Υ′ (a)
Having this observation mind, we then obtain for a ∈ FA and z ∈ FZ0 that
V (Φ0 (a) ⊗ Γn (z))V ∗
 
n
X
Φ0 (a) ⊗ Υ′ (z)  V ∗

= V ι1 (Φ0 (a) ⊗ z) + κ2j
j=1
n
(2.12) X
κ2j θ ′ (vj ) Φ0 (a) ⊗ Υ′ (z) θ ′ (vj∗ )
 
= ι1 (Φ0 (a) ⊗ z) +
j=1
10 J. CASTILLEJOS, B. DEBETS, AND G. SZABÓ

n
(2.11) X  
≈ε/3 ι1 (Φ0 (a) ⊗ z) + κ2j Φ j (a) ⊗ Υ′ (z)
n
j=1
n 
(2.7) X    
= ι1 (Φ0 (a) ⊗ z) + ι2j Φ j (a) ⊗ σ(z) + ι2j+1 Φ j (a) ⊗ z
n n
j=1
n 
(2.4) X    
≈ε/9 ι1 (Φ0 (a) ⊗ z) + ι2j Φ j−1 (a) ⊗ σ(z) + ι2j+1 Φ j (a) ⊗ z
n n
j=1
n 
X    
= ι2j−1 Φ j−1 (a) ⊗ z + ι2j Φ j−1 (a) ⊗ σ(z) + ι2n+1 (Φ1 (a) ⊗ z)
n n
j=1
n
(1.3) X  
= κ2j−1 Φ j−1 (a) ⊗ Υ(z) + ι2n+1 (Φ1 (a) ⊗ z). (2.15)
n
j=1
Hence
U V (Φ0 (a) ⊗ Γn (z))V ∗ U ∗
 
n
(2.15) X  
≈4ε/9 U  κ2j−1 Φ j−1 (a) ⊗ Υ(z) + ι2n+1 (Φ1 (a) ⊗ z) U ∗
n
j=1
n
(2.13) X    

= κ2j−1 θ(uj−1 ) Φ j−1 (a) ⊗ ϕW (z) θ(uj−1 ) + ι2n+1 (Φ1 (a) ⊗ z)
n
j=1
n
(2.10) X
≈ε/3 κ2j−1 (Φ1 (a) ⊗ Υ(z)) + ι2n+1 (Φ1 (a) ⊗ z)
j=1
(2.14)
= Φ1 (a) ⊗ Γn (z). (2.16)
Let γ : Z0 → M2n+1 (Z0 ) be any K-positive isomorphism. By Lemma 2.1,
γ and Γn are approximately unitarily equivalent. So there is a unitary
W ∈ U (1 + M2n+1 (Z0 )) such that
W γ(z)W ∗ ≈ε/9 Γn (z), z ∈ F Z0 . (2.17)
Let us consider the isomorphism η : B ⊗ Z0 → B ⊗ M2n+1 (Z0 ) given by
η := idB ⊗ (Ad(W ) ◦ γ), and set
w := η −1 (U V ) ∈ U (1 + B ⊗ Z0 ). (2.18)
Finally, we observe that
(2.18)
w(Φ0 (a) ⊗ z)w∗ = η −1 (U V η (Φ0 (a) ⊗ z) V ∗ U ∗ )

= η −1 (U V (Φ0 (a) ⊗ W γ(z)W ∗ ) V ∗ U ∗ )


(2.17)
≈ε/9 η −1 (U V (Φ0 (a) ⊗ Γn (z)) V ∗ U ∗ )
(2.16)
≈7ε/9 η −1 (Φ1 (a) ⊗ Γn (z))
(2.17)
≈ε/9 η −1 (Φ1 (a) ⊗ W γ(z)W ∗ )

= η −1 (η(Φ1 (a) ⊗ z))


A HOMOTOPY RIGIDITY THEOREM FOR Z0 -STABLE C∗ -ALGEBRAS 11

= Φ1 (a) ⊗ z.
This shows that Φ1 (a) ⊗ z ≈ε w(Φ0 (a) ⊗ z)w∗ for all a ∈ FA and FZ0 . This
verifies Φ0 ⊗ idZ0 ≈u Φ1 ⊗ idZ0 . 
We now present the notion of trace-preserving homotopy for ∗ -homomorphisms
between certain C∗ -algebras. It will be a key ingredient in the main result
of this paper.
Definition 2.4. Let ϕ, ψ : A → B be ∗ -homomorphisms between C∗ -
algebras with T + (B) 6= ∅. We say that ϕ and ψ are trace-preservingly
homotopic if there is a ∗ -homomorphism Φ : A → C([0, 1], B) with Φ0 = ϕ,
Φ1 = ψ such that
τ (Φt (a)) = τ (Φs (a))
for all a ∈ A, s, t ∈ [0, 1] and τ ∈ T + (B).
We now state a homotopy rigidity result for ∗ -homomorphisms, which can
be viewed as the main result of this paper. The key feature of it, as pointed
out in the introduction, is that it does not assume that any of the underlying
C∗ -algebras has to satisfy the UCT.
Theorem 2.5. Let A and B be separable, simple and nuclear C∗ -algebras
with T + (B) 6= ∅. Suppose ϕ, ψ : A → B are trace-preservingly homotopic
∗ -homomorphisms. Then

ϕ ⊗ idZ0 ≈u ψ ⊗ idZ0 .
Proof. By hypothesis there exists a ∗ -homomorphism Φ : A → C([0, 1], B)
such that Φ0 = ϕ, Φ1 = ψ and
θ ◦ Φt = θ ◦ Φs , θ ∈ T + (B), s, t ∈ [0, 1].
This implies
τ ◦ (Φt ⊗ idW ) = τ ◦ (Φs ⊗ idW ), τ ∈ T + (B ⊗ W), s, t ∈ [0, 1].
On the other hand, it follows that A⊗W and B ⊗W are stably projectionless
and Z-stable ([EGLN20, Corollary 6.7]). Then, by Theorem 1.2, we obtain
that Φt ⊗ idW ≈u Φs ⊗ idW for all s, t ∈ [0, 1]. Finally, by Lemma 2.3,
the ∗ -homomorphisms ϕ ⊗ idZ0 and ψ ⊗ idZ0 are approximately unitarily
equivalent. 
As a byproduct of Theorem 2.5, we obtain a precursor Z0 -stable classi-
fication theorem of sorts for separable, simple, nuclear C∗ -algebras without
requiring the UCT. For this result, we will need to work with the notion
of trace-preserving homotopy equivalence for C∗ -algebras induced from the
notion above.
Definition 2.6. Let A and B be C∗ -algebras with T + (A) 6= ∅ = 6 T + (B).
We say that A and B are trace-preservingly homotopy equivalent if there
exist ∗ -homomorphisms ϕ : A → B and ψ : B → A such that ψ ◦ ϕ and ϕ ◦ ψ
are trace-preservingly homotopic to idA and idB , respectively.
Theorem 2.7. Let A and B be separable, simple, nuclear C∗ -algebras. If A
and B are trace-preservingly homotopy equivalent, then A⊗Z0 is isomorphic
to B ⊗ Z0 .
12 J. CASTILLEJOS, B. DEBETS, AND G. SZABÓ

Proof. By assumption there exist ∗ -homomorphisms ϕ : A → B and ψ :


B → A such that ψ ◦ ϕ and ϕ ◦ ψ are trace-preservingly homotopic to idA
and idB , respectively. By Theorem 2.5 it follows that
(ψ ◦ ϕ) ⊗ idZ0 ≈u idA⊗Z0 and (ϕ ◦ ψ) ⊗ idB ≈u idB⊗Z0 .
A standard application of Elliott’s intertwining argument yields that ϕ⊗idZ0
is approximately unitarily equivalent to an isomorphism. In particular we
obtain A ⊗ Z0 ∼
= B ⊗ Z0 . 
Remark 2.8. We take a moment to compare Theorem 2.7 with the main
result of [Sch24]. Under the assumption that one has a unital embed-
ding A → B between separable unital nuclear Z-stable C∗ -algebras that
induces both a KK-equivalence and a bijection between the tracial state
spaces, it is shown there that A and B are isomorphic. It is not hard to
see (regardless whether A and B are unital or not) that any embedding
ϕ : A → B fitting into a trace-preserving homotopy equivalence must in-
duce both a KK-equivalence and a bijection between T + (A) and T + (B).
Thus one can interpret our main result as a stably projectionless analog of
Schafhauser’s theorem, although admittedly a weaker one because the con-
clusion in Schafhauser’s result is reached by merely assuming the existence
of a nice embedding in one direction. Upon assuming a trace-preserving
homotopy equivalence, however, one immediately assumes the existence of
nice embeddings in both directions, making our assumption conceptually
much stronger in comparison. It is an interesting question if Schafhauser’s
theorem has a more direct analog for non-unital C∗ -algebras.

References
+
[CGS 23] José R. Carrión, James Gabe, Christopher Schafhauser, Aaron Tikuisis, and
Stuart White. Classifying ∗ -homomorphisms I: unital simple nuclear C∗ -
algebras. Preprint arXiv:2307.06480v3.
[CE20] Jorge Castillejos and Samuel Evington. Nuclear dimension of simple stably
projectionless C∗ -algebras. Anal. PDE, 13(7):2205–2240, 2020.
[CET+ 21] Jorge Castillejos, Samuel Evington, Aaron Tikuisis, Stuart White, and Wilhelm
Winter. Nuclear dimension of simple C∗ -algebras. Invent. Math., 224(1):245–
290, 2021.
[EGLN20] George A. Elliott, Guihua Gong, Huaxin Lin, and Zhuang Niu. The classi-
fication of simple separable KK-contractible C∗ -algebras with finite nuclear
dimension. J. Geom. Phys., 158:103861, 51, 2020.
[EGLN24] George A. Elliott, Guihua Gong, Huaxin Lin, and Zhuang Niu. On the classi-
fication of simple amenable C∗ -algebras with finite decomposition rank, II. J.
Noncommut. Geom., June 2024.
[ELP98] Søren Eilers, Terry A. Loring, and Gert K. Pedersen. Stability of anticommu-
tation relations: an application of noncommutative CW complexes. J. reine
angew. Math., 499:101–143, 1998.
[GL20] Guihua Gong and Huaxin Lin. On classification of non-unital amenable simple
C∗ -algebras, II. J. Geom. Phys., 158:103865, 102 pp., 2020.
[GLN20a] Guihua Gong, Huaxin Lin, and Zhuang Niu. A classification of finite simple
amenable Z-stable C∗ -algebras, I: C∗ -algebras with generalized tracial rank
one. C. R. Math. Acad. Sci. Soc. R. Can., 42(3):63–450, 2020.
[GLN20b] Guihua Gong, Huaxin Lin, and Zhuang Niu. A classification of finite simple
amenable Z-stable C∗ -algebras, II: C∗ -algebras with rational generalized tracial
rank one. C. R. Math. Acad. Sci. Soc. R. Can., 42(4):451–539, 2020.
A HOMOTOPY RIGIDITY THEOREM FOR Z0 -STABLE C∗ -ALGEBRAS 13

[GS22] James Gabe and Gábor Szabó. The stable uniqueness theorem for equivariant
Kasparov theory. Amer. J. Math., to appear, 2022. arXiv:2202.09809v4.
[Jac13] Bhishan Jacelon. A simple, monotracial, stably projectionless C∗ -algebra. J.
Lond. Math. Soc. (2), 87(2):365–383, 2013.
[Kir] Eberhard Kirchberg: The Classification of Purely In-
finite C∗ -Algebras Using Kasparov’s Theory. URL
https://www.uni-muenster.de/imperia/md/content/MathematicsMuenster/ekneu1.pdf.
Preprint.
[KP00] Eberhard Kirchberg and N. Christopher Phillips. Embedding of exact C∗ -
algebras in the Cuntz algebra O2 . J. reine angew. Math., 525:17–53, 2000.
[Phi97] N. Christopher Phillips. Approximate unitary equivalence of homomorphisms
from odd Cuntz algebras. In Operator algebras and their applications (Water-
loo, ON, 1994/1995), volume 13 of Fields Inst. Commun., pp. 243–255.
[Phi00] N. Christopher Phillips. A classification theorem for nuclear purely infinite
simple C∗ -algebras. Doc. Math., 5:49–114, 2000.
[Raz02] Shaloub Razak. On the classification of simple stably projectionless C∗ -
algebras. Canad. J. Math., 54(1):138–224, 2002.
[Rob12] Leonel Robert. Classification of inductive limits of 1-dimensional NCCW com-
plexes. Adv. Math., 231(5):2802–2836, 2012.
[RS87] Jonathan Rosenberg and Claude Schochet. The Künneth theorem and the uni-
versal coefficient theorem for Kasparov’s generalized K-functor. Duke Math.
J., 55(2):431–474, 1987.
[RS21] Leonel Robert and Luis Santiago. A revised augmented Cuntz semigroup. Math.
Scand., 127(1):131–160, 2021.
[Rør01] Mikael Rørdam. Classification of Nuclear C∗ -Algebras. Encyclopaedia of Math-
ematical Sciences, Springer, 2001.
[Sch24] Christopher Schafhauser. KK-rigidity of simple nuclear C∗ -algebras. Preprint
arXiv:2408.02745v2.
[Ska88] Georges Skandalis. Une notion de nucléarité en K-théorie (d’après J. Cuntz).
K-Theory, 1(6):549–573, 1988.
[Sza21] Gábor Szabó. The classification of Rokhlin flows on C∗ -algebras. Comm. Math.
Phys., 382(3):2015–2070, 2021.
[Tu99] Jean-Louis Tu. La conjecture de Baum-Connes pour les feuilletages
moyennables. K-Theory, 17(3):215–264, 1999.
[TWW17] Aaron Tikuisis, Stuart White, and Wilhelm Winter. Quasidiagonality of nu-
clear C∗ -algebras. Ann. of Math. (2), 185(1):229–284, 2017.

Jorge Castillejos, Unidad Cuernavaca del Instituto de Matematicas, UNAM,


Av. Universidad s/n, 62210 Cuernavaca, Morelos, México
Email address: [email protected]

Gábor Szabó, Department of Mathematics, KU Leuven, Celestijnenlaan 200B,


box 2400, 3001 Leuven, Belgium
Email address: [email protected]

You might also like