Enough vector bundles on orbispaces

John Pardon∗
arXiv:1906.05816v1 [math.AT] 13 Jun 2019

13 June 2019

Abstract
We show that every coarsely finite-dimensional orbispace with isotropy groups of
bounded order has enough (finite-dimensional) vector bundles. It follows that the
K-theory of finite-dimensional vector bundles on compact orbispaces is well behaved.
Global presentation results for smooth orbifolds and derived smooth orbifolds also
follow.
The goal of this paper is to construct nontrivial vector bundles on orbispaces (by ‘orbis-
pace’ we mean locally equivalent to the quotient of a space by a finite group action).
Theorem 1. Let X be a separated orbispace, with isotropy groups of bounded order, whose
coarse space |X| is coarsely finite-dimensional (every open cover has a locally finite refine-
ment with finite-dimensional nerve). Then there exists a complex vector bundle V of rank
n > 0 over X, whose fiber over x ∈ X is isomorphic to a direct sum of copies of the regular
representation of Gx . We may take n = n(d, m) if |X| is coarsely d-dimensional (every open
cover has a locally finite refinement with nerve of dimension ≤ d) and has isotropy groups
of order ≤ m.
This result confirms a conjecture of André Henriques [7]. It was proven by Lück–Oliver
[8, Corollary 2.7] under the additional assumption that the isotropy groups Gx of points
x ∈ X inject into the fundamental group π1 (X, x).
Theorem 1 has a number of known consequences, which we mention but do not explain in
detail here. For example, it implies that the K-theory of finite-dimensional vector bundles on
compact orbispaces defines a cohomology theory, see [8, §3] and [7, §6.3]. It also implies that
every smooth orbifold of dimension ≤ d with isotropy groups of order ≤ m is the quotient of
a smooth manifold by a smooth action of the compact Lie group U(n) for n = n(d, m) < ∞
(refining a result of Henriques–Metzler [6]). Finally, Theorem 1 implies that every (quasi-
smooth) derived smooth orbifold with tangent and obstruction spaces of dimension ≤ d and
isotropy groups of order ≤ m is the derived zero set of a smooth section of a vector bundle
of rank ≤ n = n(d, m) over a smooth orbifold of dimension ≤ n.
The essence of Theorem 1 is already carried by the case of orbispaces arising from sim-
plicial complexes of groups. Therefore the main part of this paper consists of treating this
special case, which we state as Theorem 2.
∗
This research was conducted during the period the author served as a Clay Research Fellow and was
partially supported by a Packard Fellowship and by the National Science Foundation under the Alan T.
Waterman Award, Grant No. 1747553.

1
Theorem 2. Let X be (the separated orbispace arising from) a d-dimensional simplicial
complex of groups of order ≤ m. There exists a complex vector bundle V of rank n =
n(d, m) > 0 over X, whose fiber over x ∈ X is isomorphic to a direct sum of copies of the
regular representation of Gx .

Proof. A simplicial complex of groups (see also Haefliger [5], Corson [3], or Bridson–Haefliger
[2]) is a pair (Z, G) consisting of a simplicial complex Z together with the following data:
• For every simplex σ ⊆ Z, a group Gσ .
• For every pair of simplices σ ⊆ τ , an injective group homomorphism Gτ ֒→ Gσ .
• For every triple of simplices ρ ⊆ σ ⊆ τ , an element of Gρ conjugating the inclusion
Gτ ֒→ Gρ to the composition of inclusions Gτ ֒→ Gσ ֒→ Gρ .
• For every quadruple of simplices π ⊆ ρ ⊆ σ ⊆ τ , the resulting product of elements
of Gπ conjugating Gτ ֒→ Gπ to Gτ ֒→ Gρ ֒→ Gπ to Gτ ֒→ Gσ ֒→ Gρ ֒→ Gπ to
Gτ ֒→ Gσ ֒→ Gπ and back to Gτ ֒→ Gπ must be the identity element of Gπ .
A simplicial complex of finite groups naturally gives rise to a separated orbispace with finite
stabilizers X, obtained by gluing together σ × BGσ for σ ⊆ Z via the data given above,
where BG denotes the stack ∗/G for G finite. We note that injectivity of the homomorphisms
Gτ ֒→ Gσ is required for this gluing to makes sense.
We take an inductive (i.e. obstruction theoretic) approach to the construction of the
desired vector bundle V → X. Our orbispace X is built by attaching cells of the form
(D k , ∂D k ) × BG for k ≥ 0 and finite groups G. Note that the fact that the attaching maps
are injective on isotropy groups plays well with the desired fibers of V : for an inclusion of
groups f : H ֒→ G, the pullback of the regular representation of G is isomorphic to the
direct sum of |G| / |H| copies of the regular representation of H. It thus would suffice, for
example, to show that every complex vector bundle over ∂D k × BG extends to D k × BG.
A complex vector
L bundle V over Y × BG splits canonically as a direct sum of isotypic
pieces V = ρ∈Ĝ Vρ associated to the complex irreducible representations ρ of G, where
Vρ = Wρ ⊗ ρ (canonically) for complex vector bundles Wρ over Y . These Wρ over ∂D k are
classified by elements of πk−1 (BU(nρ )), which form obstructions to the desired extension of
Wρ from ∂D k to D k . To overcome this, we will construct not only the vector bundle V → X
but rather V along with some extra data to guarantee that these obstructions vanish.
Recall that BU(n) → BU induces an isomorphism on homotopy groups in degrees ≤ 2n
and that the homotopy groups of BU are given by π2i (BU) = Z and π2i+1 (BU) = 0 (due
to Bott). Recall also that ith Chern class of a generator of π2i (BU) equals (i − 1)! times a
generator of H 2i (S 2i ) (also due to Bott). In particular, the Chern class detects all nontrivial
elements of πi (BU), so the obstructions encountered just above will vanish if we can ensure
that the total Chern class of V is trivial (or even just rationally trivial).
We will approach the Chern classes of V via Chern–Weil theory. Given a unitary con-
nection θ on a complex vector bundle V over a smooth manifold X, its curvature Ω(V, θ) is
a 2-form valued in End(V ). The Chern character form

ch(V, θ) := tr exp(iΩ/2π) ∈ Ωeven (X; R)

is closed, and its class in cohomology ch(V ) ∈ H even (X; R) is the Chern character of V ,
independent of θ. This theory applies equally well when X is a smooth orbifold, however it

2
does not immediately capture all the information we need. For example, if X = Y × BG,
so a vector bundle over X is the same thing as a vector bundle over Y with a fiberwise
G action, the Chern character ch(V ) ∈ H even (X; R) = H even (Y ; R) ignores the G action,
whereas we need to understand the Chern characters ch(Wρ ) ∈ H even (X; R) for each of the
isotypic pieces Vρ = Wρ ⊗ ρ of V .
We thus consider what we will call the refined Chern character of V → X, which is a
cohomology class on the inertia stack

IX := Eq X →


→ X = X ×X×X X
of X (such classes have been studied in the past by Adem–Ruan [1]). Recall that the inertia
stack of a smooth orbifold is again a smooth orbifold, being given in local coordinates by
I(M/G) = {g ∈ G, x ∈ M | gx = x}/G. The refined Chern character chI (V ) ∈ H even (IX; C)
is represented by the closed form

chI (V, θ) := tr g exp(iΩ/2π) ∈ Ωeven (IX; C)

 

whose cohomology class is independent of θ. The refined Chern character (form) is com-
patible with pullback in the sense that for a smooth map of orbifolds f : X → Y , we have
f ∗ chI (V ) = chI (f ∗ V ) and f ∗ chI (V, θ) = chI (f ∗ V, f ∗ θ).
To better understand the refined Chern character, we discuss some examples. If X = BG,
a vector bundle over X is simply a representation of G, the inertia stack IX = G/G is
the set of conjugacy classes of G, and the refined Chern character chI (V ) : G/G → C
is the character g 7→ tr(g|V L ) of V regarded as a representation of G. More generally, if
X = Y × BG and V = ρ Wρ ⊗ ρ, then IX = Y × G/G and the refined Chern character
chI (V ) : G/G → H even (Y ; C) is given by chI (V )(g) = ρ tr(g|ρ) ch(Wρ ). Since the trace
P

functions g 7→ tr(g|ρ) for ρ ∈ Ĝ form a basis for the space of maps G/G → C, we see
that for X = Y × BG, the refined Chern character determines (and is determined by) the
Chern characters of each of the associated bundles Wρ . For arbitrary X, the degree zero
component of the refined Chern character chI0 (V ) ∈ H 0 (IX; C) simply records the characters
of the fibers of V .
To make sense of Chern–Weil theory on X which is a finite simplicial complex of finite
groups, we fix a family of (germs of) smooth retractions τ → σ for every pair of simplices
σ ⊆ τ such that the maps τ → σ → ρ and τ → ρ agree for ρ ⊆ σ ⊆ τ (such a family
of smooth retractions may be constructed by induction). We require everything (functions,
differential forms, bundles, connections, etc.) defined on a simplex τ to be pulled back under
τ → σ in a neighborhood of every σ ⊆ τ (such neighborhoods are not fixed, rather they
must simply exist). Note that for bundles, this requirement actually consists of the data
of a compatible family of isomorphisms with the pullback bundles (one could equivalently
consider only vector bundles built out of transition functions which satisfy the given pullback
conditions).
Let us now try again to construct inductively our desired vector bundle V , or rather V
together with a trivialization of its refined Chern character. Namely, we try to construct a
triple (V, θ, γ), where θ is a unitary connection on V and γ ∈ Ωodd (IX; C) satisfies

chI (V, θ) = nχ1 + dγ

3
where nχ1 ∈ H 0 (IX; C) ⊆ Ω0 (IX; C) denotes “n times the characteristic function of the
identity” as a function Gx /Gx → C (i.e. the character of the direct sum of n |Gx |−1 copies of
the regular representation of Gx ) over x ∈ X. As before, we just need to show that such a
triple (V, θ, γ) defined over ∂D k ×BG k
Q admits an extension to D ×BG. As we saw earlier, the
obstruction to extending V lies in ρ∈Ĝ πk−1 (BU) (we assume that n is sufficiently large in
terms of k to write BU instead of BU(nρ )), and the refined Chern character maps this group
injectively into H even (∂D k × G/G). Hence this obstruction vanishes due to the trivialization
γ of the refined Chern character form of (V, θ). We may thus extend V to D k × BG, the
connection θ extends by a partition of unity argument, and so we are left with the problem
of extending γ.
The obstruction to extending γ lies in H even (D k × G/G, ∂D k × G/G; C), which vanishes
for k odd and equals the space of functions G/G → C for k even. Concretely, this obstruction
is given by Z Z
I
ch (V, θ) − γ.
Dk ∂D k
To resolve this obstruction, we make use of the freedom we had in choosingQhow to extend
V from ∂D k × BG to D k × BG. The space of such extensions is a torsor for ρ πk (BU), and
acting via an element of this group has the effect (for k even) of adding a linear combination
of characters of representations of G to the obstruction class. We may also consider the
operation of replacing the entire triple (V, θ, γ) with the direct sum of a > 0 copies of
(V, θ, γ), which has the effect of multiplying the obstruction class by a. Hence it would
suffice to know that the obstruction class is, as a function G/G → C, a rational linear
combination of characters of irreducible representations of G, with denominators bounded
in terms of k.
Since there can be infinitely many k-cells (D k , ∂D k )×BG in X, we cannot actually do this
direct sum replacement for each cell individually, rather we must do a single such replacement
which is sufficient for extending over all k-cells simultaneously. The obstruction to this
extension problem now lies in H even (IX≤k , IX≤k−1 ; C), where X≤i ⊆ X denotes the i-skeleton
of X. This obstruction class is a lift of the refined Chern character chI (V ) ∈ H even (IX≤k )
determined by the choice of γ, and by the long exact sequence of the pair (IX≤k , IX≤k−1),
we can make the obstruction class any lift we like by choosing the appropriate γ. Hence to
ensure that the obstruction class is rational with denominators bounded in terms of k, it
suffices to show that the refined Chern character is rational with denominators bounded in
terms of the degree.
To make sense out of the above discussion of rationality and denominators, we should
introduce a Z-structure on the cohomology groups H ∗ (IX). To define the Z-structure on
H ∗ (IX), we view it as the sheaf cohomology group H ∗ (X; I) where I is the pushforward of
the constant sheaf C under the projection IX → X. The stalk Ix of this pushforward sheaf
I over a point x ∈ X is the space of functions Gx /Gx → C, which has a natural Z-structure
given by integral linear combinations of characters of complex irreducible representations of
Gx . These fiberwise Z-structures fit together to define a Z-structure on the sheaf I on X,
which thus defines the cohomology groups H ∗ (X; I) = H ∗ (IX) over the integers.
It remains to show that the refined Chern character chI (V ) ∈ H even (IX) is rational with
denominators bounded in terms of the degree. Our first step is to reduce to the case of
line bundles by proving the splitting principle. Namely, we show that for every V → X,

4
there is a map Y → X such that H ∗ (IX) → H ∗ (IY ) is split-injective (integrally) and
the pullback f ∗ V splits as a direct sum of line bundles. The usual candidate for Y is the
space of decompositions of fibers of V into ordered orthogonal one-dimensional subspaces.
To see that this works in our context, we consider what is the fiber of IY → IX over a
given point (x, g) ∈ IX. The fiber is the space of g-invariant ordered decompositions of Vx
into one-dimensional subspaces. Since the group generated by g is abelian, its irreducible
representations are one-dimensional, and hence there are plenty of such decompositions which
are g-invariant, namely when each one-dimensional subspace is contained in some g-isotypic
piece of Vx . Thus IY → IX is an iterated projective space bundle, from which it follows that
H ∗ (IX) → H ∗ (IY ) is split-injective (a splitting for the pullback to a projective space bundle
is given by the composition of multiplication by a power of the fiberwise hyperplane class,
which is a global integral cohomology class as it is the first Chern class of the tautological
bundle, and pushforward). The pullback of V to Y now splits as a direct sum of line bundles.
Since the refined Chern character is additive under direct sum of vector bundles, it suffices
to show that the refined Chern character of a line bundle is rational with denominators
bounded in terms of the degree.
There is a classifying orbispace BU(n) for principal U(n)-bundles P → X over orbispaces
X. We will only need the case n = 1 below, but for now there is no reason not to keep the
discussion general. This universal bundle
EU(n) → BU(n)

satisfies the following defining property: EU(n) is a space with a U(n) action with finite
stabilizers such that for every finite subgroup G ⊆ U(n), the fixed set EU(n)G is con-
tractible. Note that if two finite subgroups G, G′ ⊆ U(n) are conjugated by g ∈ U(n), then
′
gEU(n)G = EU(n)G , so this is only countably many conditions on EU(n) (as there are
only countably many conjugacy classes of finite subgroups of U(n)). We may construct the
space EU(n) in countably many steps by starting with E−1 U(n) = ∅, defining Ei U(n) for
i ≥ 0 from Ei−1 U(n) by adding countably many cells to kill the countably many countable
groups πi Ei−1 U(n)G for representatives
S G ⊆ U(n) of all conjugacy classes of finite subgroups,
and then defining EU(n) = i Ei U(n) as the ascending union. An obstruction theory ar-
gument shows that every orbispace P with a U(n) action with finite stabilizers admits a
U(n)-equivariant map to EU(n) (note that since EU(n) is a space, maps from an orbispace
P to EU(n) are the same as maps from its coarse space |P |).
To show that the refined Chern character of a line bundle is rational with denominators
bounded in terms of the degree, it is enough to study the universal space EU(1) → BU(1).
The classical universal space EU(1) → BU(1) is C∞ \ 0 → CP ∞ . The classifying space
EU(1) is the direct limit
 ·1 ·2 ·3 
EU(1) = lim EU(1)0! − → EU(1)1! −→ EU(1)2! − → ···
−→
·n ·n
where EU(1) − → EU(1) indicates the map induced by the nth power map U(1) − → U(1),
and the subscript EU(1)k indicates equipping EU(1) with k times the usual action of U(1).
To see that this direct limit is indeed EU(1), we note that the fixed set of the k-torsion
subgroup of U(1) is all of EU(1)n! (and thus contractible) once k divides n!. The universal
base BU(1) is thus the corresponding direct limit of BU(1)n! := EU(1)n! /U(1) which is CP ∞

5
equipped with a purely ineffective orbifold structure with isotropy group Z/n!Z. The space
IBU(1) is thus given by
IBU(1) = U(1)δtors × BU(1)
where δ indicatesQtaking the discrete topology.
Q Hence the universal refined Chern character
even ∞
is an element of U (1)tors H (CP ; C) = U (1)tors C[H], and it is given by


ζ · exp(H) ζ∈U (1)tors
.

Now the Z-structure on any finite approximation H i (IBU(1)k ; C) = ζ k =1 H i (CP ∞ ; C) is

Q

given by the Z-linear span of (ζ a )ζ k =1 · H i (CP ∞ ; Z) for a ∈ Z. Thus the universal refined
Chern character of line bundles is indeed rational with denominators bounded in terms of
the degree, as desired.
Remark 3. One may view the above proof of Theorem 2 in global (in the sense of Schwede
[9]) homotopy theoretic terms as follows. Vector bundles are classified by maps to BU(n),
and since BU(n) is not contractible, the extension problem for vector bundles has nontrivial
obstructions. Vector bundles with rationally trivialized Chern character are classified by
maps to the total space of a fibration over BU(n) whose fiber classifies odd-dimensional
rational cohomology classes on the inertia stack. This total space is rationally contractible,
which is reflected in our observation that the obstructions to this new extension problem are
torsion. The definition of this fibration over BU(n) depends on the additivity of the Chern
character.
As an epilogue, we review some basic facts about stacks (on the category of topological
spaces), we give a precise definition of what we mean by a ‘separated orbispace’, and we
show how Theorem 2 implies Theorem 1.
Let Top denote the category of topological spaces and continuous maps, and let Grpd
denote the 2-category of (essentially) small groupoids. A stack is a functor F : Topop → Grpd
which satisfies descent, i.e. such that for every topological space U and every open cover
{Ui → U}i , the natural functor

F (Ui ∩ Uj ) →
hY i
F (Ui ) →
Y Y
F (U) → Eq → →
→ F (U i ∩ Uj ∩ U k )
i i,j i,j,k

is an equivalence. Stacks form a 2-category, with morphisms given by natural transformations

of functors. The 2-category of stacks is complete, meaning all (small) limits exist; further-
more these limits may be calculated pointwise in the sense that (limα Fα )(U) = limα (Fα (U)).
Note that, as we are working in a 2-categorical context, all functors are 2-functors, all di-
agrams are 2-diagrams, all limits are 2-limits, etc. (though we will usually omit the prefix
‘2-’).
The Yoneda lemma implies that the Yoneda functor X 7→ Hom(−, X) embeds the cat-
egory of topological spaces fully faithfully into the 2-category of stacks, and moreover that
the natural map from F (X) to the groupoid of maps of stacks Hom(−, X) → F (−) is an
equivalence. The category of topological spaces is complete, and the Yoneda embedding is
continuous (commutes with limits). Hence we will make no distinction between a topological

6
space X and the associated stack Hom(−, X) of maps to X, nor between objects of F (X)
and maps Hom(−, X) → F (−) (which we will simply write as X → F ).
Every stack X has a coarse space |X| (a topological space) which is initial in the category
of maps from X to topological spaces. Concretely, the points of |X| are the isomorphism
classes of maps ∗ → X, and a subset U ⊆ |X| is open iff for every map Y → X from a
topological space Y , the inverse image of U is an open subset of Y .
A stack is called representable iff it is in the essential image of the Yoneda embedding
(i.e. it is isomorphic to a topological space). A morphism of stacks F → G is called repre-
sentable iff for every map X → G from a topological space X, the fiber product F ×G X is
representable.
For any property P of morphisms of topological spaces which is preserved under pullback,
a representable morphism of stacks F → G is said to have property P iff the pullback
F ×G X → X has P for every map X → G from a topological space X. The following are
examples of properties P of morphisms f : X → Y which are preserved under pullback:
• f is injective.
• f is surjective.
• f is a closed inclusion.
• f is open.
• f is étale, meaning that for every x ∈ X there exists an open neighborhood x ∈ U ⊆ X
such that f |U : U → Y is an open inclusion.
• f is separated, meaning that for every distinct pair x1 , x2 ∈ X with f (x1 ) = f (x2 ),
there exist open neighborhoods xi ∈ Ui ⊆ X which are disjoint U1 ∩ U2 = ∅.
• f is proper, meaning that for every y ∈ Y and every collection of open sets {Ui ⊆ X}i
covering f −1 (y), there exists a finite subcollection which covers f −1 (V ) for some open
neighborhood y ∈ V ⊆ Y .
• f admits local sections, meaning that there is an open cover {Ui ⊆ Y }i such that every
restriction f |f −1 (Ui ) : f −1 (Ui ) → Ui admits a section.
• f is a finite covering space, meaning that there is an open cover {Ui ⊆ Y }i such that
every restriction f |f −1 (Ui ) : f −1 (Ui ) → Ui is isomorphic to Ui⊔ni → Ui for some integer
ni ≥ 0.
Each of these properties P is also closed under composition, and thus also under fiber
products, meaning that for maps X → X ′ and Y → Y ′ over Z, if both X → X ′ and Y → Y ′
have P then X ×Z Y → X ′ ×Z Y ′ also has P (indeed, X ×Z Y → X ×Z Y ′ is a pullback of
Y → Y ′ ).
To check that a given map of spaces satisfies one of the properties P above, it is often
helpful to make use of the fact that these P are all local on the target, meaning that for every
open cover {Ui ⊆ Y }i , if X ×Y Ui → Ui has P for every i, then so does X → Y . This leads
to the following generalization for maps of stacks: if F → G is a representable morphism
of stacks and G′ → G is a representable morphism of stacks admitting local sections, then
F → G has P iff F ×G G′ → G′ has P. In fact, in this statement we need not assume
that G′ → G is representable, just that it admit local sections in the generalized sense that
for every map X → G from a topological space X, there exists an open cover {Ui ⊆ X}i
such that each G′ ×G Ui → Ui admits a section. We thus say that “P descends along maps
admitting local sections”. The same descent property holds for representability itself:

7
Lemma 4 (Representability descends under maps admitting local sections). Let F → G be
a map of stacks, and let G′ → G be a map of stacks admitting local sections. If F ×G G′ → G′
is representable, then so is F → G.
Proof. By replacing F → G and G′ → G with their pullbacks under U → G for a topological
space U, we may assume without loss of generality that G is representable. Since G′ → G
′
admits local sections, we may replace G → G with the composition i Ui → G′ → G where
F
{Ui ⊆ G}i is an open cover. Now each F ×G Ui is representable by assumption, and gluing
these spaces together on their common overlaps F ×G (Ui ∩ Uj ) gives a topological space
representing F .
The class of so called topological stacks (those which admit a presentation via a topological
groupoid) are somewhat better behaved than general stacks. A topological groupoid M → →O
consists of a pair of topological spaces O (‘objects’) and M (‘morphisms’), two maps M → →O
(‘source’ and ‘target’), a map O → M (‘identity’), an involution M → M (‘inverse’), and
a map M ×O M → M (‘composition’) satisfying the axioms of a groupoid. A topological
groupoid M → → O presents a stack [M → → O] defined as follows. An object of [M → → O](X)
is an open cover {Ui ⊆ X}i together with maps Ui → O and Ui ∩ Uj → M satisfying a
compatibility condition, and an isomorphism in [M → → O](X) consists of maps Ui ∩ Ui′′ → M
satisfying a compatibility condition. There is a natural map O → [M → → O] (take the trivial
open cover {X ⊆ X}) and the fiber product O ×[M → →O] O is naturally identified with M. The
→
morphism O → [M → O] admits local sections (by definition), so since O ×[M → →O] O → O
is representable, it follows by descent that O → [M → → O] is representable. Conversely,
a representable map U → X admitting local sections from a topological space U to a
stack X determines a topological groupoid U ×X U → → U presenting X. Indeed, the fiber
product U ×X U is representable (since U and U → X are), it admits two maps to U
(the two projections), an involution (exchanging the two factors), and a composition map
(U ×X U) ×U (U ×X U) = U ×X U ×X U → U ×X U (forgetting the middle factor), and
one can check using the stack property that the natural map [U ×X U → → U] → X is an
equivalence. A stack X for which there exists a representable map U → X admitting local
sections from a space U is called topological, and such U → X is called an atlas for X.
For a topological stack X with atlas U → X, for any property P which descends along
maps admitting local sections, the map U → X has P iff both maps U ×X U → U have P,
and the diagonal X → X × X has P iff the map U ×X U → U × U has P. In particular,
for any topological stack X, the diagonal X → X × X is representable, and thus every
map Z → X from a topological space Z is representable. More generally, for any map of
topological stacks X → Y , the relative diagonal X → X ×Y X is representable (by descent
from its pullback V ×X V → V ×Y V for an atlas V → X). If X → Y is representable and Y
is topological, then so is X (if U → Y is an atlas for Y , then its pullback V = U ×Y X → X
is an atlas for X), and the relative diagonal X → X ×Y X has P iff the relative diagonal
V → V ×U V of atlases has P (the latter is the pullback of the former under the map
V ×U V → X ×Y X, which is representable and admits local sections since it is a pullback
of U → Y ).
For a topological space V , a topological group G, and a continuous group action G y V ,
we may consider the action groupoid G × V → → V with maps (g, x) 7→ x and (g, x) 7→ gx.
The stack associated to this groupoid is denoted V /G and is called the stack quotient of

8
the action G y V . If G is discrete, the two maps G × V → V are étale, so V → V /G is
étale. If G is compact and V is Hausdorff, the map G × V → V × V is proper (factor as
G × V → G × V × V which is a closed inclusion since V is Hausdorff and G × V × V → V × V
which is proper since G is compact), so V /G has proper diagonal. If G is Hausdorff, then
the map G × V → V × V is separated, so V /G has separated diagonal.
A groupoid presentation of a topological stack X also gives a description of its coarse
space |X| as follows. For an atlas U → X, consider the equivalence relation ∼X on U given
by the image of U ×X U → U × U. There is a map X → U/∼X (which is tautological
once we regard X as [U ×X U → → U]), inducing a map |X| → U/∼X , which is a bijection,
essentially by definition. Since an open substack of X pulls back to an open subset of U
∼
invariant under ∼X , it follows that |X| − → U/∼X is open and is thus a homeomorphism.
In particular, it follows that the coarse space |V /G| of the stack quotient V /G is the usual
topological quotient of V by G.
An action G y V is called locally trivial iff V admits a cover by G-invariant open sets
{G × Zi ⊆ V }i where G y G × Zi acts by left multiplication on G and trivially on Zi . If
G y V is locally trivial, then the natural map from the stack quotient to the topological
quotient V /G → |V /G| is an equivalence. Indeed, this assertion is local on |V /G|, so it
suffices to consider the case of G y G × Z, where it holds by inspection.

Definition 5. A separated orbispace is a stack X for which there exists a representable étale
surjection U → X from a topological space U (an ‘étale atlas’), and the diagonal X → X ×X
is separated and proper.

Proposition 6. A stack X is a separated orbispace iff |X| is Hausdorff and there exists an
open cover {Vi /Gi ⊆ X}i where Gi are finite discrete groups acting on Hausdorff spaces Vi .

Proof. Let X be a separated orbispace, and let us show that there is an open cover {Vi /Gi ⊆
X}i . Fix an étale atlas U → X, and let u ∈ U. The automorphism group Gu := {u}×X {u} ⊆
U ×X U is finite and discrete since X → X × X is separated and proper. Since the two
projections U ×X U → → U are étale, for every g ∈ G there exists an open neighborhood
g ∈ Ug ⊆ U ×X U such that each projection restricted to Ug is an open inclusion (this gives
another proof that G has the discrete topology). Since G is finite and U ×F X U → U ×U
is separated, we may take these Ug to be disjoint. Now the complement of g Ug is closed
and disjoint from G, so it projects to a closed (by properness) subset of U × U disjoint
from
F (u, u). Hence there is an open neighborhood u ∈ V ⊆ U such that V ×X V ⊆
g Ug . Thus V ×X V is a disjoint union of pieces indexed by G, and each piece maps
homeomorphically to V under each projection. By further shrinking V , we may assume
that the map V ×X V → G respects composition (this is possible since composition is
continuous). It follows that V ×X V → → V is an action groupoid G × V → → V for an action
G y V . Since V = {1} × V ֒→ G × V = V ×X V → V × V expresses the diagonal of V as
a composition of proper maps, we conclude that V is Hausdorff. Now the map of groupoids
(G × V → → V ) = (V ×X V → → V ) → (U ×X U → → U) induces a map of stacks V /G → X. To
+
see that this is an open inclusion, let V ⊆ U denote the orbit of V under the morphisms
U ×X U → → U. Since the projections U ×X U → → U are étale, it follows that V + ⊆ U is
open, and hence [V + ×X V + → → V + ] → [U ×X U → → U] is an open inclusion of stacks; denote

9
by Z ⊆ X this open substack, so V → Z is an étale atlas. Thus the topological groupoid
V ×X V = V ×Z V → → V presents Z, so V /G → X is an open inclusion as desired.
Let us now show that for any separated orbispace X, its coarse space |X| is Hausdorff.
We saw earlier that |X| is the quotient of U by the image of U ×X U → U × U (which is an
equivalence relation). This equivalence relation is closed since U ×X U → U × U is proper,
so |X| is Hausdorff provided the quotient map U → |X| is open. Now openness of U → |X|
does not depend on which atlas U → X we take: there are étale maps U ← U ×X U ′ → U ′
over |X|, which means that U → |X| is open iff U ′ → |X| is open. Moreover, openness of
U → |X| can be checked locally on |X|, so we may assume without loss of generality that
X = V /G. Hence it is enough to note that the quotient map V → |V /G| (induced by the
canonical étale atlas V → V /G) is open. Thus |X| is Hausdorff.
Finally, let us show that if |X| is Hausdorff and there is an open cover {Vi /Gi ⊆FX}i , then
X is a separated orbispace. The maps Vi → Vi /Gi are representable étale, so U := i Vi → X
is an étale atlas. To show that the diagonal X → X × X is separated and proper, it is
equivalent to show that U ×X U → U × U is separated and proper. This reduces to showing
that V ×X V ′ → V × V ′ is separated proper for any pair V /G ֒→ X ←֓ V ′ /G′ . Since |X| is
Hausdorff, the map V ×|X| V ′ → V × V ′ a closed inclusion (as it is a pullback of the diagonal
of |X|), so it follows that V ×X V ′ → V ×V ′ is separated and proper iff V ×X V ′ → V ×|X| V ′ is
separated and proper. Now for the purposes of studying the latter map V ×X V ′ → V ×|X| V ′ ,
we may as well shrink V , V ′ , and X so that V /G = X = V ′ /G′ . Now the diagonal of
V /G = V ′ /G′ = X is separated and proper, hence so is V ×X V ′ → V × V ′ .
Corollary 7. Every separated orbispace X has an étale atlas U → X with U Hausdorff;
equivalently, every étale atlas U → X has U locally Hausdorff.
Proof. By Proposition 6 there is an open cover {Vi /Gi ⊆ X}i with Vi Hausdorff, so U :=
′
F
i Vi → X is an étale atlas with U Hausdorff. Now for any two étale atlases U, U → X,
consideration of the surjective étale maps U ← U ×X U ′ → U ′ shows that U is locally
Hausdorff iff U ′ is locally Hausdorff. Given any étale atlas U → X with U locally Hausdorff
and any open cover {Ui ⊆ U}i with Ui Hausdorff, the disjoint union U ′ := i Ui → X is an
F
étale atlas with U ′ Hausdorff.
Corollary 8. For any étale atlas U → X on a separated orbispace, there exists an open
cover {Vi /Gi ⊆ X}i as in Proposition 6 such that each map Vi → X factors through an open
inclusion Vi → U.
Proof. Let U → X be given and fix any open cover {Vi /Gi ⊆ X}i as in Proposition 6. Since
U ×X Vi → Vi is étale, it admits local sections, and hence by replacing each Vi with an open
cover of itself, we may assume that each projection U ×X Vi → Vi admits a section (which
is thus an open inclusion). Now the resulting maps Vi → U are étale by the factorization
Vi → U ×X Vi → U, so by again replacing each Vi with an open cover, we may assume they
are open inclusions.
Alternatively, we could note that the open cover {Vi /Gi ⊆ X}i produced by the proof of
Proposition 6 is in fact of the desired form.
Corollary 9. A map of separated orbispaces X → Y is representable iff it is injective on
isotropy groups. In particular, a separated orbispace X is a space iff it has trivial isotropy.

10
Proof. Any representable morphism of stacks is injective on isotropy groups (just test against
points). Thus we are left with showing that a map of separated orbispaces X → Y which is
injective on isotropy groups is representable.
Since representability descends under maps admitting local sections, it suffices to show
that the fiber product X ′ = X ×Y Y ′ is representable for some étale atlas Y ′ → Y . Note
that X ′ has trivial isotropy since Y ′ has trivial isotropy and X ′ → Y ′ is injective on isotropy
groups (being a pullback of X → Y ).
We claim that X ′ is a separated orbispace, provided we take Y ′ to be Hausdorff (which
we can by Corollary 7). The pullback of an étale atlas U → X is an étale atlas U ′ → X ′
(since X ′ → X is representable, being a pullback of Y ′ → Y ). The diagonal of X ′ is the
composition X ′ → X ′ ×X X ′ → X ′ × X ′ . The second map X ′ ×X X ′ → X ′ × X ′ is separated
and proper as it is a pullback of X → X × X. To analyze the first map X ′ → X ′ ×X X ′ ,
note that it pulls back to U ′ → U ′ ×U U ′ under the map U ′ ×U U ′ → X ′ ×X X ′ which
admits local sections (being a pullback of U → X). Since Y → Y × Y is separated, its
pullback Y ′ ×Y Y ′ → Y ′ × Y ′ is also separated, which implies each projection Y ′ ×Y Y ′ → →Y′
is separated since Y ′ is Hausdorff, which implies Y ′ → Y is separated. Hence its pullback
U ′ → U is separated, so U ′ → U ′ ×U U ′ is a closed inclusion, hence separated and proper.
Thus X ′ → X ′ ×X X ′ is separated and proper, and we conclude that X ′ is a separated
orbispace.
We are thus reduced to showing that a separated orbispace Z with trivial isotropy is a
space. By Proposition 6, we know that Z is given locally by V /G for V Hausdorff and G finite
discrete acting freely (since Z has trivial isotropy). Free actions G y V with V Hausdorff
and G finite are locally trivial, so we conclude that the map Z → |Z| is an equivalence.
Alternatively, we could note that the chart V /G near a given x ∈ X constructed in the proof
of Proposition 6 in fact satisfies G = Gx by definition.
A sieve on a topological space X is a subset S ⊆ 2X consisting of open S sets such that
′ ′
U ⊆ U ∈ S implies U ∈ S. A covering sieve on X is a sieve S such that U ∈S U = X. An
open cover {Ui ⊆ X}i is said to generate the covering sieve on X consisting of those open
sets which are contained in some Ui .
A connection sieve on a map of spaces f : X → Y is a covering sieve S on X such that
(1) for U ∈ S, the composition U → X → Y is an open inclusion, and (2) for U, V ∈ S with
f (U) = f (V ), either U = V or U ∩ V = ∅. Note that for sieves satisfying condition (1),
condition (2) is equivalent to condition (2′ ) for U, V ∈ S either U ∩ V = ∅ or f (U ∩ V ) =
f (U) ∩ f (V ). If S ′ ⊆ S is an inclusion of covering sieves and S is a connection sieve on
X → Y , then so is S ′ . In particular, if S and S ′ are connection sieves are on X → Y then so
is S ∩ S ′ . To check that an open cover {Ui ⊆ X}i generates a connection sieve, it is enough
to check axioms (1) and (2′ ) for the open sets Ui .
Let us call a map X → Y strongly étale iff it admits a connection sieve. Open inclusions
are strongly étale, and strongly étale maps are separated and étale (however the converse is
false). Being strongly étale is preserved under pullback (take the connection sieve generated
by the pullback of the original connection sieve), and the class of strongly étale maps is closed
under composition (a connection sieve SX/Z for a composition X → Y → Z is given by those
elements of a fixed connection sieve SX/Y for F X → YF whose image lies in a fixed connection
sieve SY /Z for Y → Z). A disjoint union i Xi → i Yi of strongly étale maps Xi → Yi is

11
strongly étale (take disjoint union of connection sieves), and the projection A × Y → Y is
strongly étale for any discrete space A. Being strongly étale is not local on the target (there
are finite covering spaces with non-Hausdorff target which are not strongly étale).
Recall that a topological space is called paracompact iff every open cover admits a locally
finite refinement [4]. If X is paracompact and Hausdorff, then there exists a partition of
unity subordinate to any given locallyPfinite open cover {Ui ⊆ X}i , namely functions fi :
X → R≥0 with supp fi ⊆ Ui such that i fi ≡ 1 (recall that the support supp f of a function
f : X → R≥0 is by definition the complement of the largest open set over which f vanishes
identically).
Lemma 10. If Y is paracompact Hausdorff, then a map X → Y is strongly étale iff there
exists an open cover {Ui ⊆ Y }i such that each map X ×Y Ui → Ui is strongly étale.
Proof. Fix an open cover {Ui ⊆ Y }i and connection sieves Si on X ×Y Ui → Ui . Since Y
is paracompact, we may assume that our open cover {Ui ⊆ Y }i is locally finite. Using a
partition of unity subordinate to this open cover, we may find another open coverT{VI ⊆ Y }I
indexed by non-empty finite subsets I of the original index set, such that VI ⊆ i∈I Ui and
VI ∩ VJ = ∅ unless I ⊆ J or J ⊆ I (explicitly, we may take VI to be the locus where
mini∈I fi > max / fi ). We may now define a connection sieve on X → Y as the union over
Ti∈I
X×Y VI
I of 2 ∩ i∈I Si .
Proposition 11. Every separated orbispace X whose coarse space |X| is paracompact has
an étale atlas U → X for which theFprojections U ×X U → → U are strongly étale. In fact,
there exists such U of the form U = i Vi for an open cover {Vi /Gi ⊆ X}i as in Proposition
6.
Proof. Fix an open cover {Vi /Gi ֒→ X}i as in Proposition 6. Since |X| is paracompact, we
may shrink the spaces Vi (Gi -equivariantly) so as to ensure that the associated open cover
{|Vi /Gi | ⊆ |X|}i of coarse spaces is locally finite. Choose a partition of unity {fi : |X| →
R≥0 }i subordinate to the open cover {|Vi /Gi | ⊆ |X|}iF . Let Vi0 ⊆ Vi denote the open subset
where fi > 0. We will show that the étale atlas U := i Vi0 has the desired property.
It suffices to show that for any pair of open inclusions V /G ֒→ X ←֓ W/H and pair of
functions fV , fW : |X| → R≥0 supported inside |V /G| and |W/H|, respectively, the projection
W 0 ×X V 0 → V 0 is strongly étale. We begin by considering the map W ×X V → V , which is
a finite covering space over its open image W/H ×X V ⊆ V (by pullback from W → W/H,
which is a finite covering space by descent from its pullback H × W → W ). Thus every point
of W/H ×X V has a neighborhood over which W ×X V → V is strongly étale. Even better,
since V is Hausdorff and G is finite, each G-orbit inside W/H ×X V has a neighborhood over
which W ×X V → V is strongly étale. In other words, each point of |W/H| ∩ |V /G| ⊆ |X|
has a neighborhood over which W ×X V → V is strongly étale.
Now since |X| is paracompact, there exists a locally finite open cover of |X| by |X| \
(supp fV ∪supp fW ) together with open subsets {Ai ⊆ |W/H|∩|V /G|}i over which W ×X V →
V is strongly étale. Fix a partition of unity g : |X| → R≥0 supported
P inside |X| \ (supp fV ∪
supp fW ) and {gi : |X| → R≥0 } supported inside Ai , that is g + i gi ≡ 1. Now the patching
procedure for connection sieves from the proof of Lemma 10 shows that W ×X V → V is
strongly étale over the complement of supp g. In particular, it follows by restriction that
W 0 ×X V 0 → V 0 is strongly étale.

12
 A simplicial complex is a pair X = (V, S) consisting of a set V (“vertices”) and a set
S ⊆ 2V \ {∅} (“simplices”) of finite subsets of V such that S contains all singletons and
∅ 6= A ⊆ B ∈ S implies A ∈ S. The star st(X, σ) ⊆ X of a simplex σ in a simplicial
complex X is the subcomplex consisting of all simplices τ ⊆ X with σ ∪ τ ∈ S(X).
A map of simplicial complexes X → Y is a map of vertex sets V (X) → V (Y ) which
maps simplices to simplices (the image of an element of S(X) is an element of S(Y )). A
map of simplicial complexes is called injective iff the map on vertex sets (hence also the
map on simplices) is injective. A map of simplicial complexes f : X → Y is called étale
(resp. locally injective) iff the induced maps on stars st(X, σ) → st(Y, f (σ)) are isomorphisms
(resp. injective). We will call a map of simplicial complexes X → Y sufficiently étale iff every
simplex σ ⊆ Y (equivalently, every vertex) is the image of a simplex τ ⊆ X at which X → Y
is étale (this is a useful weakening of the condition of being surjective and étale, which in
the context of simplicial complexes is too strong).
V (X)
The
P geometric realization kXk of a simplicial complex X is the set of tuples t ∈ R≥0
with v tv = 1 such that {v : tv > 0} ∈ S(X), topologized by declaring that the realization
of the complete simplex on k +1 vertices has the usual topology and that a realization kXk is
given the strongest topology for which (the realization of) every map from a complete simplex
to X is continuous. The geometric realization of an étale map of simplicial complexes is an
étale map of spaces.
A locally injective simplicial complex groupoid M → → O consists of simplicial complexes O
and M together with structure maps satisfying the axioms of a groupoid, where both maps
M → → O are locally injective. Local injectivity of the two maps M → → O implies that the
∼
natural map kM ×O Mk − → kMk ×kOk kMk is a homeomorphism, and thus the geometric
realization kMk → → kOk is a topological groupoid. If ∂O ⊆ O denotes the subcomplex
consisting of those simplices σ ⊆ O for which it is not the case that the first projection
M → O is étale at every simplex τ ⊆ M mapped to σ under the second projection, then the
natural map kOk \ k∂Ok → [kMk → → kOk] is étale. A locally injective simplicial complex
groupoid M → → O is called sufficiently étale iff this map is surjective (equivalently, every
vertex of O is M-isomorphic to one not in ∂O).
The abstract simplex category Simp has objects finite totally ordered sets and has mor-
phisms weakly order preserving maps; every object of Simp is isomorphic to [n] := {0 < · · · <
n} for a unique integer n ≥ 0. A simplicial object in a category C is a functor Simpop → C,
and the category of simplicial objects in C is denoted sC. If C is complete (resp. cocomplete)
then so is sC, and limits (resp. colimits) are calculated pointwise.
Here we will consider only simplicial sets (objects of the category sSet) and simplicial
groupoids (objects of the category sGrpd). A simplicial set (or groupoid) will be denoted X• ,
where Xn is its set (or groupoid) of n-simplices. We denote by ∆n• ∈ sSet the standard n-
simplex, given by [m] 7→ Hom([m], [n]). The Yoneda lemma implies that Xn = Hom(∆n• , X• ).
A map of simplicial sets X• → Y• is called injective iff it is so levelwise (i.e. every
Xn → Yn is injective). A map X• → Y• is called étale (resp. locally injective) iff for every
∼
map [n] → [m] in Simp, the induced map Xm − → Ym ×Yn Xn is a bijection (resp. injective)
(it is equivalent to impose this condition only for n = 0). A map X• → Y• is called étale
∼
(resp. locally injective) at a given n-simplex σ of X• iff Xm − → Ym ×Yn {σ} is a bijection
(resp. injective) (this condition at a given σ implies the same at any preimage of σ under

13
any structure map of X• ). A map X• → Y• is called sufficiently étale iff the n-simplices of
X• at which the map is étale surject onto the n-simplices of Y• (it is equivalent to impose
this condition only for n = 0). These notions generalize to maps of simplicial groupoids by
replacing ‘injectivity’ and ‘surjectivity’ for maps of sets with ‘full faithfulness’ and ‘essential
surjectivity’ for functors of groupoids. These properties are all preserved under pullback and
closed under composition.
The geometric realization kX• k of a simplicial set X• is the colimit of ∆n over all maps
∆n• → X• . Geometric realization is cocontinuous, and the natural map klimα (X• )α k →
limα k(X• )α k is bijective, however it need not be a homeomorphism even for finite limits.
The map kX• ×Y• Z• k → kX• k ×kY• k kZ• k is a homeomorphism if at least one of the maps
X• → Y• and Z• → Y• is locally injective. Thus a locally injective simplicial set groupoid
M• → → O• determines a topological groupoid kM• k → → kO• k.
Let us now introduce the geometric realization kX• k of any simplicial groupoid X• which
is étale, meaning that it admits a locally injective sufficiently étale map U• → X• from a
simplicial set U• . For any simplicial groupoid X• and any locally injective map U• → X• , the
pair of simplicial sets U• ×X• U• → → U• forms a locally injective simplicial set groupoid, whose
geometric realization kU• ×X• U• k → → kU• k thus defines a topological stack. The geometric
realization of X• is defined as this topological stack [kU• ×X• U• k → → kU• k] associated to any
locally injective sufficiently étale map U• → X• .
Lemma 12. The geometric realization of an étale simplicial groupoid X• is well defined.
Proof. Let U• , U•′ → X• be locally injective and sufficiently étale. Let U•′′ := U• ×X• U•′ → X• ,
and consider the map of simplicial set groupoids (U•′′ ×X• U•′′ → → U•′′ ) → (U• ×X• U• → → U• ).
′′
Since U• → U• is locally injective and sufficiently étale, it follows that this map induces an
isomorphism of topological stacks, and the same applies to U•′ in place of U• .
Lemma 13. The geometric realization kX• k of an étale simplicial groupoid X• with finite
isotropy is a separated orbispace.
Proof. All geometric realizations are Hausdorff, so kU• ×X• U• k → kU• k × kU• k is separated,
hence kX• k has separated diagonal. Since X• has finite isotropy, the map U• ×X• U• → U• ×U•
has finite fibers, which combined with local injectivity of U• ×X• U• → → U• implies that
kU• ×X• U• k → kU• k × kU• k is proper, hence kX• k has proper diagonal.
To construct an étale atlas for kX• k, let ∂U• ⊆ U• denote the simplicial subset consisting
of those simplices of U• at which U• → X• is not étale. Then kU• k \ k∂U• k → kX• k is étale.
To see that it is surjective, it suffices to show that kU• ×X• U• k \ kU• ×X• ∂U• k → kU• k is
surjective (note that surjectivity descends under surjective maps such as kU• k → kX• k), and
this follows since U• → X• is sufficiently étale.
A simplicial complex X gives rise to a simplicial set b• X (its barycentric subdivision)
whose n-simplices are chains of simplices σ0 ⊆ · · · ⊆ σn ⊆ X (in other words, b• X is the
nerve of S(X)). Barycentric subdivision preserves injectivity, local injectivity, étale, and
sufficiently étale. There is a natural identification of geometric realizations kXk = kb• Xk.
Moreover, for a locally injective sufficiently étale simplicial complex groupoid M → → O,
→ →
there is a natural identification [kMk → kOk] = k[b• M → b• O]k (the simplicial groupoid
[b• M →→ b• O] is étale since M →
→ O is sufficiently étale).

14
 A simplicial complex of groups (Z, G) also admits a barycentric subdivision b• (Z, G)
which is a simplicial groupoid. In fact, we will define b• (Z, G) for any simplicial complex Z
equipped with a functor G : S(Z)op → Grpd from the face poset to groupoids (a simplicial
complex of groups (Z, G) determines such a functor which σ to BGσ ). The groupoid of n-
simplices in the barycentric subdivision b• (Z, G) is now defined as the groupoid of functors
from the category 0 → · · · → n to the category whose objects are pairs σ ∈ S(Z) and o ∈ Gσ
and whose morphisms are inclusions σ1 ⊆ σ2 covered by maps o1 → o2 |σ1 . The barycentric
subdivision of a simplicial complex of groups is étale, as can be seen as follows. For any
σ ⊆ Z and o ∈ Gσ , we consider the functor S(st(Z, σ))op → Grpd given by τ 7→ Gσ∪τ ×Gσ {o}.
Applying the nerve construction from just above to this functor, we obtain a simplicial set
mapping to b• (Z, G), which is the required locally injective map which is étale over (σ, o) ∈
b0 (Z, G). An essentially equivalent discussion (albeit without barycentrically subdividing)
appears in [2, 12.24–12.25]. Since b• (Z, G) is étale, it has a geometric realization kb• (Z, G)k
which we also write as k(Z, G)k. When G are finite groups (or, more generally, groupoids
with finite isotropy), then the geometric realization k(Z, G)k is a separated orbispace by
Lemma 13, and this is what we have been calling the separated orbispace presented by the
simplicial complex of finite groups (Z, G).

Lemma 14. Let M → → O be a locally injective simplicial complex groupoid with the following
properties:
• The vertices of every simplex of O are pairwise non-isomorphic via M.
• If simplices σ and σ ′ of O have vertex sets which are isomorphic via M, then σ and σ ′
are themselves isomorphic via M.
Then there is a simplicial complex of groups giving rise to the same simplicial groupoid as
M→ → O.
Proof. The hypotheses imply that there is a simplicial complex Z whose vertices are the
isomorphism classes in the vertex groupoid V (M) → → O(M), and whose simplices are the
M-isomorphism classes of simplices of O. Now M → → O defines a functor G : S(Z)op →
Grpd, and there is a natural isomorphism between b• (Z, G) and the simplicial groupoid
[b• M →
→ b• O]. By definition, all the groupoids Gσ have a single isomorphism class, and all
the functors Gτ → Gσ are faithful (this follows from local injectivity of M →
→ O). Choosing
(independently) a base object of each Gσ shows (Z, G) comes from a simplicial complex of
groups.
The nerve N(X, {Ui }i ) of a collection of open sets {Ui ⊆ X}i is the simplicial complex
whose vertices V are the indices
T i with Ui 6= ∅, and in which a collection I of indices spans
a simplex (i.e. I ∈ S) iff i∈I Ui 6= ∅. A partition of unity {fi : X → R≥0 }i subordinate
to a locally finite open cover {Ui ⊆ X}i defines a map from X to the geometric realization
kN(X, {Ui }i )k of the nerve of the open cover.

Proposition 15. For any separated orbispace X whose coarse space |X| is paracompact,
there exists a simplicial complex of groups (Z, G) (where the groups Gz for vertices z ∈ Z
are isotropy groups of points of X) and a map X → k(Z, G)k which is injective on isotropy
groups. If |X| is coarsely finite-dimensional (resp. coarsely d-dimensional), then Z may be
taken to be finite-dimensional (resp. d-dimensional).

15
Proof. We begin with an open F cover {Vi /Gi ⊆ X}i with the properties guaranteed by Propo-
sition 11, and we set U := i Vi . Since |X| is paracompact, by Gi -equivariantly shrinking
the spaces Vi , we may assume that the associated open cover {|Vi /Gi | ⊆ |X|}i of coarse
spaces is locally finite. We fix a covering sieve S on U ×X U which is invariant under the
‘exchange’ (i.e. ‘inverse’) involution of U ×X U and is a connection sieve for both projections
U ×X U → → U. We also fix a partition of unity {fi : |X| → R≥0 }i subordinate to the open
cover {|Vi /Gi | ⊆ |X|}i F
Denote by o(x) ⊆ U and oi (x) ⊆ Vi the fibers over x ∈ |X|, F so o(x) = i oi (x); similarly
define m(x) ⊆ U ×X U and mij (x) ⊆ Vi ×X Vj with m(x) = i,j mij (x). These sets are finite
since the Gi are finite and {|Vi /Gi | ⊆ |X|}i is locally finite. Furthermore, they have the
discrete topology, since U is Hausdorff and U ×X U is Hausdorff (since U is Hausdorff and
X → X × X is separated).
The Hausdorff property implies that the inclusions oi (x) ⊆ Vi and mij (x) ⊆ Vi ×X Vj
admit retractions defined in some open neighborhood. Now the inverse images of small open
neighborhoods x ∈ |Zx | ⊆ |X| (i.e. open substacks Zx ⊆ X) form a basis of neighborhoods
of oi (x) and m Fij (x), so for sufficiently small
F Zx , these inverse images are naturally disjoint
unions Ux = o∈o(x) Uo and Ux ×X Ux = m∈m(x) Um . Note that this applies only inside Vi
and Vi ×X Vj for which oi (x) and mij (x) are non-empty: the full inverse image of Zx inside
U may intersect other Vi nontrivially. By shrinking Zx further, we may ensure that the
retraction (Ux ×X Ux → → Ux ) → (m(x) → → o(x)) is a map of groupoids. For later purposes,
let us also take Zx small enough so that:
• If x ∈ |Vi /Gi | then |Zx | ⊆ |Vi /Gi |.
• If x ∈/ supp fi then |Zx | ∩ supp fi = ∅.
• If fi (x) > 0 then fi > 0 over all of |Zx |.
Each of these conditions can be ensured on its own, and since the open cover {|Vi /Gi | ⊆ |X|}i
is locally finite, we can ensure all at once.
We also shrink Zx so as to ensure that each Um ∈ S (our chosen connection sieve), which
has the following implication: given o, o′ ∈ U with Uo ∩ Uo′ 6= ∅, the relation Um ∩ Um′ 6= ∅
is a partial bijection between lifts m, m′ ∈ U ×X U of o and o′ . Furthermore, the domain of
this bijection is as large as possible: for Uo ∩ Uo′ 6= ∅ (so o ∈ oi (x) and o′ ∈ oi (x′ ) for some
i) with x, x′ ∈ |Vj /Gj |, we get a full bijection between the inverse images of o and o′ inside
mij (x) and mij (x′ ) (this follows since the projection Vi ×X Vj → Vi is a finite covering space
of degree |Gj | over Uo ∪ Uo′ ).
We now consider the nerves N(U, {Uo }o∈U ) and N(U ×X U, {Um }m∈U ×X UT). Note that for
simplices in these nerves, namely subsets O ⊆ U or M ⊆ U ×X U with o∈O Uo 6= ∅ or
T
m∈M Um 6= ∅, the maps O → |X| or M → |X| are injective. The natural maps on index
sets U ×X U → → U determine maps of nerves

N(U ×X U, {Um }m∈U ×X U ) →

→ N(U, {Uo }o∈U ).

These maps are locally injective; indeed, local injectivity means that for every o, o′ ∈ U with
Uo ∩ Uo′ 6= ∅ and every m ∈ U ×X U projecting to o, there is at most one lift m′ ∈ U ×X U of
o′ with Um ∩ Um′ 6= ∅, and this is a direct consequence of our assumption that every Um ∈ S.

16
Let us now argue that there is a natural composition map

N(U ×X U, {Um }m∈U ×X U ) ×N (U,{Uo }o∈U ) N(U ×X U, {Um }m∈U ×X U )

−→ N(U ×X U, {Um }m∈U ×X U ).
T
More precisely, we claim that for non-empty finite subsets M ⊆ U × X U with m∈M Um 6= ∅
′
T
and M ⊆ U ×X U with m∈M ′ Um 6= ∅ projecting to O ⊆ U (under the first and second
projections, respectively), the subset M ′′ ⊆ U ×X U defined by applyingTthe composition
map (U ×X U) ×U (U ×X U) → U ×X U to M and M ′ also satisfies m∈M ′′ Um 6= ∅.
This
T claim ∼ follows
T from ∼ theTproperty that every Um ∈ S (indeed, this property implies that
m∈M mU −
→ U
o∈O o ←− m∈M ′ Um ). We have thus defined a locally injective simplicial
complex groupoid

M := N(U ×X U, {Um }m∈U ×X U ) →

→ N(U, {Uo }o∈U ) =: O.
We now show that this locally injective simplicial complex groupoid is sufficiently étale.
Every isomorphism class of vertex (equivalently, every x ∈ |X| with |Zx | = 6 ∅) has a repre-
sentative o ∈ Vi ⊆ U with fi (o) = fi (x) > 0. To show that these representatives are étale,
let m ∈ U ×X U be a morphism with source o and target o′ , and let Uo′ ∩ Uo′′ 6= ∅. We
must show that there is a bijection between morphisms o → o′ and o → o′′ . The morphisms
in question all have source inside Vi , so we really can consider just Vi × U for the present
purpose. Now the bulleted conditions on |Zx | from above imply that since Uo′ ∩ Uo′′ 6= ∅ and
fi (o′ ) > 0, we have Uo′ ∪ Uo′′ ⊆ |Vi /Gi | ⊆ |X|. Thus over Uo′ ∩ Uo′′ the map Vi ×X U → U
is a finite covering space of degree |Gi |. Thus all points have the same number of lifts, so it
follows that the connection sieve property gives us a bijection between lifts.
We have already seen above that our sufficiently étale locally injective simplicial complex
groupoid satisfies the first hypothesis of Lemma 14, and the second hypothesis follows from
the partial bijection property derived above from the connection sieve. Thus by Lemma 14,
there is a simplicial complex of groups (Z, G) giving rise to the same simplicial groupoid
b• (Z, G) = [b• M →→ b• O] and thus (since M → → O is sufficiently étale) to the same geometric
→
realization k(Z, G)k = [kMk → kOk]. The groups associated Gz associated to vertices
z ∈ Z are by definition isotropy groups of the vertex groupoid V (M) → → V (O), which are by
definition isotropy groups of points of X.
To conclude, it remains to define a map X → k(Z, G)k which is injective on isotropy
groups. To define this map, we shrink the Zx so that the open cover {|Zx | ⊆ |X|}x∈|X| is
locally finite, and choose a partition of unity {gx : |X| → R≥0 }x∈|X| subordinate to the open
cover {|Zx | ⊆ |X|}x∈|X| . These maps gx lift to maps go : U → R≥0 supported inside Uo and
gm : U ×X U → R≥0 supported inside Um . The collection of these lifts defines a map of
topological groupoids

(U 0 ×X U 0 →
→ U ) → kN(U ×X U, {Um }m∈U ×X U )k →
0


→ kN(U, {Uo }o∈U )k
where U 0 is the disjoint union of the open loci Vi0 ⊆ Vi where fi > 0 (theFmaps go and
gm do not define a map over all of U ×X U → → U due to the fact that Ux = o∈o(x) Uo and
F
Ux ×X Ux = m∈m(x) Um may not be the full inverse images of Zx inside U and U ×X U). It
remains to check that this map is injective on isotropy groups; in other words, for o, o′ ∈ U 0

17
we must show that the map {o}×X {o′ } → kN(U ×X U, {Um }m∈U ×X U )k is injective. Distinct
elements of {o} ×X {o′ } are, in particular, distinct lifts of o, which therefore cannot lie in
any common Um since Um → U is injective, so we see that the map is indeed injective on
isotropy groups.
A complex vector bundle over a stack X is a representable map V → X together with
maps V ×X V → V and C × V → V (both over X) such that for every map U → X from
a topological space U, there exists an open cover {Ui ⊆ U}i and integers ni ≥ 0 such that
V ×X Ui → Ui is isomorphic to Cni × Ui → Ui equipped with its fiberwise addition and
scaling maps.
Proof of Theorem 1. Apply Proposition 15 to find a simplicial complex of finite groups (Z, G)
and a map X → k(Z, G)k which is injective on isotropy groups. Now Theorem 2 applies
to k(Z, G)k to give a vector bundle V → k(Z, G)k, and the pullback of this bundle to X
satisfies the desired property since X → k(Z, G)k is injective on isotropy groups.

References
[1] Alejandro Adem and Yongbin Ruan, Twisted orbifold K-theory, Comm. Math. Phys. 237
(2003), no. 3, 533–556. MR 1993337

[2] Martin R. Bridson and André Haefliger, Metric spaces of non-positive curvature,
Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathe-
matical Sciences], vol. 319, Springer-Verlag, Berlin, 1999. MR 1744486

[3] Jon Michael Corson, Complexes of groups, Proc. London Math. Soc. (3) 65 (1992), no. 1,
199–224. MR 1162493

[4] Jean Dieudonné, Une généralisation des espaces compacts, J. Math. Pures Appl. (9) 23
(1944), 65–76. MR 0013297

[5] André Haefliger, Complexes of groups and orbihedra, Group theory from a geometrical
viewpoint (Trieste, 1990), World Sci. Publ., River Edge, NJ, 1991, pp. 504–540. MR
1170375

[6] Andre Henriques and David S. Metzler, Presentations of noneffective orbifolds, Trans.
Amer. Math. Soc. 356 (2004), no. 6, 2481–2499. MR 2048526

[7] Andre Gil Henriques, Orbispaces, ProQuest LLC, Ann Arbor, MI, 2005, Thesis (Ph.D.)–
Massachusetts Institute of Technology. MR 2717316

[8] Wolfgang Lück and Bob Oliver, The completion theorem in K-theory for proper actions
of a discrete group, Topology 40 (2001), no. 3, 585–616. MR 1838997

[9] Stefan Schwede, Global homotopy theory, New Mathematical Monographs, vol. 34, Cam-
bridge University Press, Cambridge, 2018. MR 3838307

18