arXiv:1601.04123v1 [math.

AT] 16 Jan 2016

Intermediate Thom Spectra, Hopf-Galois

Extensions and a New Construction of M U
Jonathan Beardsley
January 19, 2016

Introduction

Hopf-Galois extensions of rings, introduced in [12], naturally generalize the

idea of Galois extensions of fields, but with the Galois group replaced by a
Hopf-algebra or bialgebra that coacts on the extension. The base ring A is
then obtained as the primitives of the H-coaction on B. Geometrically this
can be thought of as saying that Spec(B) is a G = Spec(H)-module and
that Spec(A) is obtained from Spec(B) by taking G-orbits. We refer to [18]
for an excellent survey on the topic.
The theory of Galois and Hopf-Galois extensions for E -ring spectra
was introduced by Rognes in [21], and for E1 -rings in the thesis of Roth
[22]. Their definitions are extended in a natural way in this note to define
Hopf-Galois extensions of En -ring spectra. There is not a general Galois
correspondence between sub-Hopf-algebras and intermediate extensions for
Hopf-Galois extensions (of either discrete rings or ring spectra). However,
as we will see in the following note, it is occasionally possible to define
pieces of such a correspondence. It is of course necessary to determine at
the very least what sub- and quotient- Hopf-algebras should even be for
spectral bialgebras. While we dont give a general characterization of such
structures, we contend that a morphism of connected n-fold loop spaces
H G such that H is the fiber of a morphism of n-fold loop spaces G Q
presents H as a sub-bialgebra of G and Q as a quotient bialgebra of G in
the quasicategory of Kan complexes. This structure then induces sub- and
quotient bialgebras in spectra S[H] S[G] and S[G] S[Q]. In many
cases, as indicated in both [21] and [22], Thom spectra give examples of
Hopf-Galois extensions of ring spectra whose associated bialgebras are the
suspension spectra of topological monoids. Thus it stands to reason that a

fibration H G Q might lead to an intermediate Hopf-Galois extension

of ring spectra. This is the content of Theorem 1:
Theorem. Suppose i : H G and f : G BGL1 (S) are morphisms of
En -monoidal Kan complexes for n > 1, with G and H reduced and simply
f

connected. If the composition G BGL1 (S) BGL1 (HZ) is nullhomotopic then there is a triangle of Hopf Galois extensions of En1 -monoidal
ring spectra, where the associated bialgebras are written over their respective extensions:
S[G]

S[H] ##

// M f
::

S[G/H]

M (f i)

As fibrations of n-fold loop spaces are relatively easy to come by (for

instance apply the loops functor n times to any fibration of spaces), it is
perhaps not surprising that one immediately has a number of interesting
examples as described in the table in Section 5, which we reproduce here:
Fibration

Hopf-Galois Extension

Bialgebra

M SU M U

S[CP ]

BString BSpin K(Z, 4)

M String M Spin

S[K(Z, 4)]

BU BSO Spin

M U M SO

S[Spin]

BString BU[6, ) B 3 Spin

M String M U[6, )

S[B 3 Spin]

BSp BSO B(SO/Sp)

M Sp M SO

S[B(SO/Sp)]

SU(n) SU(n + 1) S 2n+1

X(n) X(n + 1)

S[S 2n+1 ]

BSO BO Z/2

M SO M O

S[Z/2]

2 S 3 h3i 2 S 3 S 1

HZ
2 HZ/2

S[S 1 ]

BSU BU

CP

Several of the examples in the above table appear to be intimately linked

to real Bott periodicity and the classical computations of [7]. It is not known
to this author whether or not all of the above can be shown to be Hopf-Galois
extensions using classical K
unneth spectral sequence computations.
In the final section of this note we describe the example that motivated
all of this theory. There we show that the complex cobordism spectrum can
be constructed by canonically attaching E1 -cells to the sphere spectrum. In
particular, the filtration of BU SU by the spaces SU(n) yields the
2

filtration of M U by the X(n) spectra described by Ravenel [20]. Moreover,

this filtration seems to bear some resemblance to Lazards construction of
the Lazard ring by iterative generation of the universal formal n-bud [13].
Hopefully this characterization of M U will lead to a better understanding
of the appearance of formal group structure in chromatic homotopy theory.
One important distinction to make between this work and earlier work on
Hopf-Galois extensions of ring spectra is that the work of Rognes and Roth
relies on the theory of model categories introduced by Quillen [19], whereas
[4] uses the theory of quasicategories introduced originally by Boardman and
Vogt [5] and elaborated upon by Joyal [10] and Lurie [15] [14]. A great deal
of work has been done by the latter author to ensure that Quillens theory
of model categories maps faithfully to the theory of quasicategories and that
both frameworks are useful for doing homotopy theory.

Background

We use the notation of [14] and [15] without change. We make heavy use
of the theory of -operads therein, and refer to [14] for a full description
of that theory or [4] for a brief review. For the extent of this note all Kan
complexes will have base points and all maps between them will be basepoint
preserving. We will denote the quasicategory of such objects by T . We will
denote the stabilization of T , the quasicategory of spectra, by S. It has a
symmetric monoidal structure with unit object the sphere spectrum S. In
certain cases we will need our connected Kan complexes to be reduced, i.e.
have a unique 0-simplex, but when this occurs we will say so.
Definition 1. Let i : H G be a morphism of En -monoidal Kan complexes. Define the quotient of G by H, denoted G/H, to be the relative
tensor product G H in the sense of 4.4.2 of [14], where the H-module
structure on G is determined by i. Recall that as the cone point of an operadic colimit diagram G/H admits a universal morphism q : G G/H.
Note that i is not made explicit in the notation, but will always be clear
from context.
Lemma 1. If i : H G is a morphism of En -monoidal Kan complexes
then G/H is an En1 -monoidal Kan complex and the quotient morphism
q : G G/H is a morphism of En1 -monoidal Kan complexes.
Proof. To construct G/H as an En1 -monoidal Kan complex, we first notice
that the relative bar construction of Section 4.4 of [14] is constructed from
the data of modules over an E1 -algebra and that AlgEn (T ) AlgE1 (AlgEn1 (T )).
3

In other words the operadic bar construction of Construction 4.4.2.7 of [14]

is relative to the operad O E
n1 and Bar (G, H, ) is a simplicial object
of AlgEn1 (T ). As the forgetful functor AlgEn1 (T ) T is En1 -monoidal
it preserves the relevant structure in G/H = G H and the map it receives
from G G.
Recall from [4] that if H is a cocommutative bialgebra in S (with an
En -algebra structure for some n), the quasicategory of H-comodules admits
a symmetric monoidal structure. Moreover, since H is a bialgebra and
admits a unit S H, there is a trivial comodule functor S ComodH
which takes a spectrum A to the comodule with comodule structure map
A AS AH. We will say that an H-comodule is a trivial H-comodule
if it is in the essential image of this functor.
Remark 1. Let H be a cocommutative En -bialgebra in S, and let B be an
Em -algebra in CoM odH . Then the comodule structure map c : B B H
gives B H the structure of a an Ek -B-algebra, for k = min(n, m). On the
other hand the trivial coaction B B S B H induces a different B 1
algebra structure on B H which corresponds to the map B B H B H
B H.
Definition 2. Let A be an En -ring spectrum and B be an Em -algebra in
LM odA for m < n. Let H be a cocommutative En -bialgebra in S. Then
we say that H coacts on B over A if B is an H-comodule and the following
condition is satisfied: the coaction map c : B B H is a morphism of
A-algebras where B H is equipped with the A-algebra structure induced
by the composition of the unit map A B with the second B-algebra
structure described in the Remark 1.
The following is an example of when this sort of coaction occurs:
Proposition 1. Let R be an En1 -ring spectrum and let f : X BGL1 (R)
be a morphism of En -monoidal Kan complexes. Then R[X] coacts on M f
over R.
Proof. The maps f : X BGL1 (R) and f 1 : X X X BGL1 (R)
are objects of TBGL1 (R) . The coaction map of R[X] on M f is given by
Thomifying the diagonal : X X X (thought of as a morphism over
BGL1 (R)). The trivial coaction is given by Thomifying the inclusion map
i1 : X X X. Its clear that the diagram X X X commutes
(again, over BGL1 (R)). This argument should be compared to the argument
in Theorem 1.2 of [16].
4

Remark 2. In Proposition 2 below, we use the notation and terminology of

[4]. In particular, the cotensor product of an H-comodule B with the trivial
H-comodule S, denoted BH S, is given as the limit of the cosimplicial object
BS

//

//
// B H S

// B H H S . . .
//

where the coface maps are given by the H-coaction on B, the diagonal map
of H, and the unit map of H. This object can be constructed using Luries
two sided bar construction in [14]. We will denote this cosimplicial spectrum
by Cobar (B, H, S).
Proposition 2. Let H be a cocommutative En -bialgebra in S and B and A
En -algebras in CoM odH . Assume also that B is an A-algebra and H-coacts
on B over A. Then the cotensor product BH S has an A-algebra structure
and receives a universal morphism A BH S.
Proof. Since H-coacts on B over A, the cosimplicial construction whose
limit is BH S lifts to a diagram of A-algebras and as such inherits a unit
morphism A BH S.
The following definition, for E -ring spectra in the symmetric monoidal
simplicial model category of S-modules, is due to Rognes [21]. It was later
generalized by Roth to E1 -ring spectra, again using the category of Smodules [22]. Our definition below generalizes both of these definitions and
only differs in that it is phrased in the language of quasicategories. The motivating example that should be kept in mind is the unit morphism S M U
from the sphere spectrum to the complex cobordism spectrum, as described
in [21].
Definition 3 (Hopf-Galois Extensions). Let H be a cocommutative En bialgebra in S and A an Em -algebra in CoM odH that is a trivial H-comodule.
Let B be an Em -algebra in CoM odH , : A B a morphism of Em -algebras
and assume that H coacts on B over A. If:
1c

B 1

1. the composite morphism B A B B A B H B H is an

equivalence of En -ring spectra, and
2. the cotensor product BH S exists and the canonical A-algebra map
A BH S is an equivalence,
then we say that the map : A B is an H-Hopf-Galois extension of
Em -ring spectra.

The Main Theorem

Theorem 1. Suppose i : H G and f : G BGL1 (S) are morphisms of

En -monoidal Kan complexes for n > 1, with G and H reduced and simply
f

connected. If the composition G BGL1 (S) BGL1 (HZ) is nullhomotopic then there is a triangle of Hopf Galois extensions of En1 -monoidal
ring spectra, where the associated bialgebras are written over their respective
extensions:
S

S[G]

S[H] ##

// M f
::

S[G/H]

M (f i)

We will prove this theorem using two propositions (which in turn will
rely on a number of lemmas). It should be clear from the results of [2] [1]
and [3] that S M f and S M (f i) are both Hopf-Galois extensions
with bialgebras S[G] and S[H] respectively. Thus it remains to show that
the morphism M (f i) M f is a Hopf-Galois extension with bialgebra
S[G/H]. We will show that M f can be produced as a Thom spectrum over
M (f i), which will immediately yield the coaction of S[G/H] on M f over
M (f i) as well as the torsor equivalence M f M (f i) M f M f S[G/H]
(as the Thom diagonal and Thom isomorphism respectively). Then we will
show that M f S[G/H]S M (f i).
Proposition 3. There is a morphism of En1 -monoidal Kan complexes
G/H BGL1 (M (f i) such that the colimit of the composite morphism
G/H BGL1 (M (f i) LM odM (f i) is equivalent to M f .
Proof. Let G/H denote an En1 -monoidal quotient Kan complex of G by the
H-action on G induced by i following Definition 1. By Lemma 2, the fiber
of the universal morphism G G/H is an En1 -monoidal Kan complex
i
which is equivalent to H G. Hence by Lemma 3 the En1 -monoidal left
f

Kan extension of G BGL1 (S) S along q : G G/H takes the unique

0-simplex of G/H to M (f i). By Proposition 5, this Kan extension factors
as a morphism of En1 -monoidal Kan complexes through BGL1 (M (f i)).
Taking the Thom spectrum of the induced morphism G/H BGL1 (M (f
i)) produces M (f i)/((G/H)) as a Thom spectrum over M (f i). By
Lemma 4 and Corollary 3.1.4.2 of [14] we have that the left operadic Kan
extension along G G/H followed by the left operadic Kan extension along
G/H is equivalent to the left operadic Kan extension along G (i.e.
6

Kan extensions compose). Thus the iterated Kan extension which produces
M (f i) = S/H first then quotients it by the action of G/H is equivalent to
the one-step Kan extension S/G M f . Hence M f is produced as a Thom
spectrum over M (f i). This fact alone gives us a coaction M f M f G/H
and the torsor condition
M f M (f i) M f M f S[G/H].

Proposition 4. The S[G/H] cofixed point spectrum of M f , computed by

the cotensor product M f S[G/H] S, where S has the trivial S[G/H]-coaction,
is equivalent to M (f i).
Proof. For the following proof we work within T/BGL1 (S) equipped with the
Day convolution symmetric monoidal structure. We will sometimes denote
an object of TBGL1 (S) by (X, f ) instead of f : X BGL1 (S). We start with
the morphism of En -monoidal Kan complexes f : G BGL1 (S). Note that
the quotient map G G/H induces a strictly cocommutative coaction on
f : G BGL1 (S) in TBGL1 (S) by the coalgebra : G/H BGL1 (S) (i.e.
the trivial map from G/H to BGL1 (S)). Moreover, since G/H is a Kan
complex, and thus a unital coalgebra in T , : G/H BGL1 (S) is a unital
coalgebra in TBGL1 (S) , hence there is a coaction of : G/H BGL1 (S) on
the point BGL1 (S). We may take the cosimplicial cobar construction in
TBGL1 (S) , Cobar ((G, f ), (G/H, ), (, )). By tensoring there is a morphism
of E -monoidal Kan complexes BGL1 (S) BGL1 (HZ). Thus by composition we extend our cobar construction to a cosimplicial object in TBGL1 (HZ) .
Recall, however, that f : G BGL1 (S) BGL1 (HZ) is null, thus there is
an equivalence of cosimplicial objects over BGL1 (HZ) between the one with
f

bottom map G BGL1 (S) BGL1 (HZ) and the one with bottom map

G BGL1 (HZ). As the Thom spectrum functor BGL1 (HZ) LM odHZ

is symmetric monoidal, we have an equivalence of cosimplicial HZ-modules
HZ Cobar (M f, S[G/H], S) and HZ Cobar (S[G], S[G/H], S). Thus the
Bousfield-Kan spectral sequences for each of these objects are isomorphic.
The Bousfield-Kan spectral sequence for the latter object is precisely the
Eilenberg-Moore spectral sequence computing the integral homology of H,
and converges strongly since G/H is simply connected (see 4.1 of [6]). Hence
the unit map of the former, HZ M (f i) HZ (M f S[G/H]S) is equivalent to the unit map of the latter HZ S[H] HZ (S[G]S[G/H] S), which
is the homology equivalence realizing H as the fiber of the map G G/H.

Since M (f i) is a connective spectrum, and the limit of a connective cosimplicial spectrum remains connective, we have the equivalence.
Remark 3. Note that the above proposition would fail in the case that M f
was not HZ-oriented. However, the result would still hold after completing
at 2, since the composite morphism X BGL1 (S) BGL1 (HZ/2) is
always nullhomotopic for any map X BGL1 (S).
Corollary 1. Let f : X BGL1 (S) be a morphism of En -monoidal Kan
complexes such that X BGL1 (S) BGL1 (HZ) is nullhomotopic. Then
the induced morphism of En -ring spectra S M f is a Hopf-Galois extension
with associated bialgebra S[X].
Proof. Apply Theorem 1 to the fibration X X.

The Lemmas

Lemma 2. The fiber of the morphism G G/H in the category of En1 monoidal Kan complexes is equivalent to H as an En1 -algebra.
Proof. From Proposition 3.2.2.1 of [14] we recall that the fiber of a morphism of En1 -algebra objects is computed in the underlying category of
Kan complexes. From Corollary 8.3 of [17] we recall that the fiber of the
map G G/H is indeed equivalent to H (i.e. G G/H is a principal
H-fibration). Thus the fiber of the map of En1 -monoidal Kan complexes
G G/H is equivalent as an En1 -monoidal Kan complex to H.
Lemma 3. The En1 -monoidal left Kan extension of G BGL1 (S) S
along G G/H is computed by taking the colimit of the composition
f ib(G G/H) G BGL1 (S) S.
Proof. Following the notation given in Definition 3.1.2.2 and the construction in Remark 3.1.3.15 of [14], we have a correspondence of -operads is
given by
a
M (G 1 )
G/H Fin 1 .
G {1}

In other words, there is a family of -operads indexed by 1 which looks

like G (the -operad associated to G as an En -monoidal Kan complex) at
one end and G/H at the other end. Formula () of Definition 3.1.2.2 of [14]

states that the value of the desired Kan extension at a 0-simplex G/H
is given by the colimit diagram:

((M
act )/ M G ) (M )/ M T

where the morphism (M )/ M takes the cone point to . In other

words, the value of the Kan extension at is computed by taking the colimit
over the diagram in M of objects (and active morphisms) living over .
As the simplicial set M is nothing more than the mapping cylinder of
the morphism of En -monoidal Kan complexes G G/H , we have the
result.
Lemma 4. There is a 2 -family of -operads induced by the morphisms of
En1 -monoidal Kan complexes G G/H and G/H , denoted M
2 Fin , and the induced projection M 2 is a flat categorical fibration.
Proof. The composition is given by a 2-simplex in the quasicategory of
En1 -monoidal quasicategories, hence by a morphism of simplicial sets in

2
Hom(2 , Hom(E
n1 , qCat)) Hom( En1 , qCat). By the quasicategorical Grothendieck construction of [15], we obtain a coCartesian fibration of
1

simplicial sets p : M 2 E
n1 such that p (0) G , p (1) G/H
and p1 (2) , where G , G/H and are the -operads witnessing
the En1 -monoidal structure on G, G/H and . The projection map induces
a family of -operads M 2 . This projection is a flat fibration as it
satisfies the requirements of Example B.3.4 of [14], i.e. there are coCartesian
lifts of every edge in 2 2 2 Fin .
The following proposition is relatively important to the main theorem
of this note, so we will explain the intuition behind it first. We have two
constructions that we wish to show are equivalent:
1. the Kan extension of G BGL1 (S) S along G G/H, which
looks like:
G
q



G/H

// BGL1 (S) c 44// S

2. and the composition : G/H BGL1 (M (f i)) LM odM (f i) S

where the first map comes from the universal property of G/H and
9

the trivial H-action on BGL1 (M (f i)):

f

// BGL1 (M (f i)) c // LM odM (f i)u//11 S

33



G/H.

// BGL1 (S)

The former, the Kan extension, is computable by Lemma 3 above, and

thus it can be identified as picking out a G/H-action on M f S/H
relative to the G-action on S. The latter has the desired property of
yielding a colimit whose target is a Thom spectrum over M (f i), and as
such supports a Thom isomorphism and Thom diagonal. Hence by showing
that the two functors are equivalent we are able to see that M f is a relative
Thom spectrum over M (f i).
Proposition 5. The Kan extension of G BGL1 (S) S along G G/H
is equivalent as an En1 -monoidal morphism to a morphism that factors as
a morphism of En1 -monoidal Kan complexes through BGL1 (M (f i)).
Proof. We prove the proposition by constructing a composition G G/H
BGL1 (M (f i)) and proving that it is equivalent (after including into
LM odM (f i) and forgetting down to S) to the Kan extension. Since M (f i)
is a colimit of H G BGL1 (S) S, M (f i) must be M (f i)oriented. Thus the composition H G BGL1 (S) BGL1 (M (f i)
is null homotopic. Hence the induced action of H on BGL1 (M (f i))
is trivial, and by Lemma 5 there is a factorization of En1 -monoidal Kan
complexes G G/H BGL1 (M (f i)). Let denote the composition
G/H BGL1 (M (f i)) LM odM (f i) S and let denote the Kan
extension G/H S described in the proof of Theorem 1. By Corollary
3.1.3.4 of [14] we know that , as a left Kan extension, is produced as a left
adjoint. In other words, there is an adjunction
Lanq : AlgG (S) AlgG/H (S) :
where the right adjoint is given by composition with the quotient map
q : G G/H and the left adjoint Lanq is given by taking the left operadic
Kan extension along q (here the operad in question is E
n1 , but for simplicity
we leave it out of the notation). Hence we have an equivalence of mapping
spaces:
AlgG (S)(c f, q) AlgG/H (S)(, )
10

where c : BGL1 (S) S LM odS is the canonical inclusion (note that on

the left hand side we are implicitly using the fact that q c f by the
definition of a Kan extension). By the natural equivalence of Lemma 5, we
obtain an equivalence of mapping spaces
AlgG/H (S)(, ) AlgG (S)(c f, u c t f )
where u c t is the composition
c

BGL1 (S) BGL1 (M (f i)) LM odM (f i) S,

t being given by tensoring with M (f i) and u being the forgetful functor.
In case it is not clear that u c t preserves En1 -algebras, notice that
it is the application of the monad associated to M (f i) as an En -ring
spectrum, so it is at least En1 lax monoidal (by Corollary 7.3.2.7 of [14]).
Now we have two morphisms of En1 -monoidal quasicategories: c f and
uc tf , but the latter is precisely the former composed with the application
of extension/restriction of scalars adjunction associated to M (i f ). Hence
the application of the unit of this monad, which is a natural transformation
idS u t, induces a natural transformation (of En1 -monoidal functors)
c f u t f . Passing back along the adjunction Lanq , we obtain a
morphism of En1 -algebras . Finally, noticing that this morphism is
an equivalence on objects, we have that it is a natural equivalence. Thus
and are equivalent. Noticing that factors through BGL1 (M (f i) we
have proven the proposition.
Lemma 5. Given a morphism of En1 -monoidal Kan complexes G
such that the H-module structure on X induced by the composition
G X is the trivial H-module structure, there is a factorization
G/H X and a natural equivalence of functors between G G/H
and G X.

X
i
H
G
X

Proof. The quotient G/H is constructed as the relative tensor product

G H (see Definition 1). By using the En1 -H-algebra structure of G
and (since G and H both receive En -monoidal morphisms from H), we
can construct G/H as a bar construction in En1 -H-algebras (as in Theorem 4.4.2.8 of [14]). Since the H-action on X is trivial and is induced by
the composition H G X, the morphism G X then defines a morphism from the simplicial object defining G/H into the constant simplicial
object on X. Or, in other words, there is a coherently H-bilinear morphism

11

G H X essentially coming from the commutative diagram

GH

// G




// X

where the upper horizontal map and the left-hand vertical map are the Haction on G and the unit of H, respectively, and the other maps are the
given map G X. Thus by the universal property of the colimit (again
being taken within En1 -monoidal Kan complexes) the needed factorization
is obtained.

Examples

A large number of morphisms of En -monoidal Kan complexes fit into the

framework described in the introduction and Theorem 1. We assume that a
morphism of connected n-fold loop spaces is always modeled by a morphism
of reduced En -monoidal Kan complexes. The following Lemma allows us to
positively identify such examples.
Proposition 6. Let F E be a morphism of connected En -monoidal Kan
complexes for n 1 and let X be the En1 -monoidal fiber of the induced
En1 -morphism BF BE. Then F E X satisfies the conditions of
Theorem 1. In particular X is equivalent to E/F as an En1 -monoidal Kan
complex.
Proof. The universal property of E/F (cf. Lemma 5) induces a morphism
of En1 -monoidal Kan complexes E/F X, which induces an equivalence
on underlying Kan complexes (see also Proposition 8.8 of [17]).
Below we have a table giving a number of applications of Theorem 1:

12

Fibration

Hopf-Galois Extension

Bialgebra

BSU BU CP

M SU M U

S[CP ]

BString BSpin K(Z, 4)

M String M Spin

S[K(Z, 4)]

BU BSO Spin

M U M SO

S[Spin]

M String M U[6, )

S[B 3 Spin]

M Sp M SO

S[B(SO/Sp)]

X(n) X(n + 1)

S[S 2n+1 ]

M SO M O

S[Z/2]

HZ
2

S[S 1 ]

BString BU[6, )

B 3 Spin

BSp BSO B(SO/Sp)

SU(n) SU(n + 1)

S 2n+1

BSO BO Z/2
2 S 3 h3i

2 S 3

S1

HZ/2

Remark 4. A few entries from the above table may require some explanation:
1. The spectra X(n) were defined by Ravenel in [20] and play an essential
role in Devinatz, Hopkins and Smiths proof of Ravenels Nilpotence
Conjecture [8] [9].
2. The fibration BString BU[6, ) B 3 Spin is perhaps not well
known to many readers and can be found in [11] along with many
other interesting fibrations.
3. Notice that the final two Hopf-Galois extensions above are only HopfGalois extensions over the 2-complete sphere but Theorem 1 can be
applied with minor changes in light of Remark 3.
Remark 5. Perhaps some of the most useful consequences of the above
types of identifications are the torsor equivalences of Hopf-Galois extensions.
The above table yields the following equivalences, some of which are easily
verified by classical computations and some of which are not:
1. M U M SU M U M U S[CP ]
2. M Spin M String M Spin M Spin S[K(Z, 4)]
3. M SO M U M SO M SO S[Spin]
4. M SO M Sp M SO M SO S[B(SO/Sp)]
5. X(n + 1) X(n) X(n + 1) X(n + 1) S[S 2n+1 ]
6. M U[6, ) M String M U[6, ) M U[6, ) S[B 3 Spin]
13

7. M SO M O M SO M SO S2 [Z/2]
1
8. HZ/2 HZ2 HZ/2 HZ/2 S
2 [S ]

Remark 6. In the case that the fibration of interest is a fibration of E monoidal Kan complexes, Theorem 1 can be expressed in a slightly different
way. Note that given a fibration H G G/H, with associated HopfGalois extension M H M G, Theorem 1 allows us to represent M G as
M H/(G/H). In the case that all the spectra involved are E -rings, this
quotient can be presented as an actual pushout in the quasicategory of E rings. Thus, for instance, using the same numbering as above, we obtain
the following equivalences:
1. M U M SU S 1 S
2. M Spin M String K(Z,3) S
3. M SO M U SO/U S
4. M SO M Sp SO/Sp S
6. M U[6, ) M String BBSpin S

5.1

A New Construction of MU

In the case that G/H is equivalent to S n for some finite n, Theorem 4.10
of [3] and Theorem 1 above allow us to give more concrete constructions of
Hopf-Galois extensions as versal En1 -algebras (see e.g. [23] and [3] for more
on this). In fact, this type of characterization follows just from Proposition
5, as the cofixed points condition of Hopf-Galois extensions is unnecessary.
One example that may be of particular interest to stable homotopy theorists is a construction of M U as a canonical sequence of Hopf-Galois extensions obtained by repeatedly attaching an E1 -cell. This construction should
be compared to Lazards construction of the Lazard ring in [13]. AntolinCamarena and Barthel also have ongoing work applying these concepts to
the p-complete Hopf-Galois extension HZ
p HZ/p.
Corollary 2. Let X(n) be the Thom spectrum associated to the morphism
of E2 -monoidal Kan complexes SU(n) BU BGL1 (S). Then X(n +
a

1) is a versal E1 -algebra over X(n) of characteristic n where {n }N is a

canonical sequence of elements in 2n1 (X(n)).

14

Proof. Given the fibration SU(n) SU(n + 1) S 2n+1 , and by an

application of Theorem 1 and Lemma 6 above we can identify X(n + 1) as
the E1 -monoidal Thom spectrum given by the E1 -monoidal left Kan extension S 2n+1 BGL1 (X(n)). The map of E1 -monoidal Kan complexes
n M apE1 (S 2n+1 , BGL1 (X(n))), by application of standard adjunctions, induces a map of Kan complexes
n M apT (S 2n1 , GL1 (X(n))).
Note that
n must have image contained in a connected component u
0 (GL1 (X(n)) Z/2 which induces a translation u : X(n) X(n).
The composition u
n : S 2n1 X(n) lifts to a morphism of speca

tra n : S2n1 X(n). An application of Theorem 4.10 of [3] gives that

a

X(n + 1) is the versal E1 -algebra of characteristic n on X(n).

Remark 7. The content of [3] allows us to consider X(n + 1) as the E1 a

spectrum obtained by attaching an E1 -cell to X(n) along the map n dea

scribed above. Note that 1 , the Hopf element in 1 (S).

References
[1]

Matthew Ando, Andrew J. Blumberg, and David Gepner.

Parametrized spectra, multiplicative Thom spectra, and the twisted
Umkehr map. 2011. url: http://arxiv.org/abs/1112.2203.

[2]

Matthew Ando et al. An -categorical approach to R-line bundles,

R-module Thom spectra, and twisted R-homology.
In: J. Topol. 7.3 (2014), pp. 869893. issn: 1753-8416.

[3]

Omar Antoln-Camarena and Tobias Barthel.

A simple universal property of Thom ring spectra. 2014.
url: http://arxiv.org/abs/1411.7988.

[4]

Jonathan Beardsley. Some Higher Coalgebra. 2015.

url: http://arxiv.org/abs/1508.00861.

[5]

Michael Boardman and Rainer Vogt.

Homotopy Invariant Algebraic Structures on Topological Spaces.
Vol. 347. Lecture Notes in Mathematics. Springer-Verlag, 1973.

[6]

A. K. Bousfield.
On the homology spectral sequence of a cosimplicial space.
In: Amer. J. Math. 109.2 (1987), pp. 361394.

15

[7]

Henri Cartan. Demonstrations homologique des theor`emes de

periodicite de Bott, II. Homologie et cohomologie des groupes
classiques et de leurs espaces homog`enes. In: Seminaire Henri
Cartan-Moore, tome 12 17.2 (1959-1960), pp. 132.

[8]

E. S. Devinatz, M. J. Hopkins, and Jeff Smith.

Nilpotence and stable homotopy theory I.
In: Annals of Mathematics. 2nd ser. 128.2 (1988), pp. 207241.

[9]

Michael J. Hopkins and Jeffrey H. Smith.

Nilpotence and stable homotopy theory II.
In: Ann. of Math. (2) 148.1 (1998), pp. 149.

[10]

[11]

Andre Joyal. The theory of quasi-categories and its applications. Lecture Notes. 2008. url:
http://mat.uab.cat/~kock/crm/hocat/advanced-course/Quadern45-2.pdf.
Nitu Kitchloo, Gerd Laures, and W. Stephen Wilson.
The Morava K-theory of spaces related to BO.
In: Adv. Math. 189.1 (2004), pp. 192236.

[12]

H. F. Kreimer and M. Takeuchi.

Hopf algebras and Galois extensions of an algebra.
In: Indiana Univ. Math. J. 30.5 (1981), pp. 675692.

[13]

Michel Lazard. Commutative formal groups.

Lecture Notes in Mathematics, Vol. 443.
Springer-Verlag, Berlin-New York, 1975.

[14]

Jacob Lurie. Higher Algebra. 2014.

url: www.math.harvard.edu/~lurie/papers/higheralgebra.pdf.

[15]

Jacob Lurie. Higher topos theory. Vol. 170.

Annals of Mathematics Studies.
Princeton, NJ: Princeton University Press, 2009.

[16]

Mark Mahowald. Ring spectra which are Thom complexes.

In: Duke Math. J. 46.3 (1979), pp. 549559.

[17]

J. Peter May. Classifying Spaces and Fibrations.

American Mathematical Society: Memoirs of the American
Mathematical Society v. 155, no. 2.
American Mathematical Society, 1975.

[18]

Susan Montgomery. Hopf Galois theory: a survey.

In: New topological contexts for Galois theory and algebraic geometry
(BIRS 2008). Vol. 16. Geom. Topol. Monogr.
Geom. Topol. Publ., Coventry, 2009, pp. 367400.
16

[19]

Daniel G. Quillen. Homotopical algebra.

Lecture Notes in Mathematics, No. 43.
Springer-Verlag, Berlin-New York, 1967.

[20]

Doug Ravenel.
Complex Cobordism and the Homotopy Groups of Spheres.
Academic Press, 1986.

[21]

John Rognes. Galois extensions of structured ring spectra. Stably

dualizable groups. In: Mem. Amer. Math. Soc. 192.898 (2008).

[22]

Fridolin Roth.
Galois and Hopf-Galois Theory for Associative S-Algebras.
PhD thesis. Universit
at Hamburg, 2009.

[23]

Markus Szymik. Commutative S-algebras of prime characteristics

and applications to unoriented bordism.
In: Algebr. Geom. Topol. 14.6 (2014), pp. 37173743.

17