(September 11, 2010)

Solenoids
Paul Garrett [email protected] http://www.math.umn.edu/˜garrett/

Circles are simple geometric objects, although the theory of functions on them, Fourier series, is complicated
enough to have upset people in the 19th century, and to have precipitated the creation of set theory by
Cantor circa 1880. But we postpone the function theory to another time. Here, we consider circles and
simple mappings among them.
Surprisingly, automorphism groups of families of circles connected by simple maps bring to light structures
and objects invisible when looking at a single circle rather than the aggregate.

Q
In particular, we discover p-adic numbers p inside automorphism groups of families of circles, and even
A
the adeles . These appearances are more important than ad hoc definitions as completions with respect to
metrics, recalled later. That is, p-adic numbers and the adeles appear inevitably in the study of modestly
complicated structures, and are parts of automorphism groups.
This discussion of automorphisms of families of circles is a warm-up to the more complicated situation of
automorphisms of families of higher-dimensional objects [1] acted upon by non-abelian groups.
In all cases, an underlying theme is that when a group G acts [2] transitively [3] on a set X, then X is
in bijection with G/Gx , where Gx is the isotropy subgroup [4] in G of a chosen base point x in X, by
gGx → gx. The point is that such sets X are really quotients of the group G. Topological and other
structures also correspond, under mild hypotheses. Isomorphisms X ≈ G/Gx are informative and useful, as
seen later.

• The 2-solenoid
• Automorphisms of solenoids
• A cleaner viewpoint
• Automorphisms of solenoids, again
• Appendix: uniqueness of projective limits
• Appendix: topology of X ≈ G/Gx

1. The 2-solenoid
As reported by MacLane in his autobiography, around 1942 Eilenberg talked to MacLane (in Michigan)
about families of circles related to each other by repeated windings, for example, double coverings, and
about trying to understand the limiting object. We make this precise and repeat some of the discussion. The
point is that a surprisingly complicated physical object is made from families of circles.

[1] The next example in mind is modular curves, which are two-dimensional. Their definition is considerably more

complicated than that of circles, and requires more preparation.

[2] Of course, the spirit of the notion of action of G on X is that G moves around elements of the set X. But a little

more precision is needed. Recall that an action of G on a set X is a map G × X → X such that 1G · x = x for all
x ∈ X, and (gh)x = g(hx) for g, h ∈ G and x ∈ X.
[3] Recall that a group G acts transitively on a set X if, for all x, y ∈ X, there is g in G such that gx = y.

[4] Recall that the isotropy subgroup G of a point x in a set X on which G acts is the subgroup of G fixing x, that
x
is, Gx is the subgroup of g ∈ G such that gx = x.

1
 Paul Garrett: Solenoids (September 11, 2010)

As a model for the circle S 1 we take S 1 = R/Z. [5] Eilenberg (and MacLane) considered a family of circles
and maps
R Z −→ R/Z −→ R/Z
×2 ×2 ×2
. . . −→ /

where each circle mapped to the next by doubling itself onto the smaller circle. [6] More precisely, this is
literally multiplication by 2 on the quotients / , namelyRZ
x + Z → 2x + Z

R
for x ∈ . Since 2 ⊂ Z Z this is well-defined. That is, each circle is a double cover of the circle to its
immediate right in the sequence. This sequence of circles with doubling maps is the 2-solenoid. [7] We
might ask what is the limiting object

R Z −→ R/Z −→ R/Z
×2 ×2 ×2
??? . . . −→ /

Part of the issue is to say what we might mean by this question.

A different but topologically equivalent model is a little more convenient for our discussion. Consider the
sequence
R Z −→ R/4Z −→ R/2Z −→ R/Z
ϕ43 ϕ32 ϕ21 ϕ10
. . . −→ /8

where each map ϕn,n−1 : R/2n Z → R/2n−1 Z is induced from the identity map on R in the diagram

R id / R
mod 2n mod 2n−1

 
R/2nZ _ _ ϕ_ _ _ _/ R/2n−1Z
n,n−1

That is, this is the map

ϕn,n−1 : x + 2n Z → x + 2n−1Z
This second model has the advantage that the maps ϕn,n−1 are locally distance-preserving on the circles. In
the first model each map stretches the circle by a factor of 2. In the second model as we move to the left in
the sequence of circles the circles get larger. Also, in the second model there is the single copy of lying R
over (or uniformizing) all the circles.
Again we would like to ask what is the limit of these circles? Note that this use of limit is ambiguous, and
we cannot be sure a priori that there is any potential sense to be made of this. Presumably we expect the
limit to be a topological space.

[5] One could also take the unit circle in C. The map x → e2πix mapping R → C factors through the quotient R/Z,
showing that these two things are the same.
[6] The integer 2 could be replaced with other integers, and, for that matter, the sequence 2, 2, 2, . . . could be

replaced with other sequences of integers. Qualitatively, these other choices give similar things, though the details
are significantly different.
[7] The name is by analogy with the wiring in electrical motors and inductance circuits, where things are made by

repeated winding. Since the limiting object here is allegedly created by repeated unwinding, it might be more apt to
call it an anti-solenoid.

2
 Paul Garrett: Solenoids (September 11, 2010)

[1.0.1] Remark: A plausible, naive guess would be that since

T
2n Z = {0}, that perhaps the limit is
R/{0} ≈ R. It is not.
n
[8]

A precise version of the question is the following. Consider

ϕn+1,n ϕn,n−1
... / Xn / Xn−1ϕn−1,n−2/ . . . ϕ21
/ X1 ϕ10
/ X0

with topological spaces Xi and continuous transition maps ϕi,i−1 . The (projective) limit X of the Xn ,
written
X = lim Xi (dangerously suppressing reference to transition maps ϕi,i−1 )
i

is a topological space X and maps ϕn : X → Xn compatible with the transition maps ϕn,n−1 : Xn → Xn−1
in the sense that
ϕn−1 = ϕn,n−1 ◦ ϕn
and such that, for any other space Z with maps fn : Z → Xn compatible with the maps ϕn,n−1 (that is,
fn−1 = ϕn,n−1 ◦ fn ), there is a unique f : Z → X through which all the maps fn factor. That is, in pictures,
first, all the (curvy) triangles commute in

ϕ0

ϕ1
# $
X ... ϕ21
/ X1 ϕ10
/ X0

and, for all families of maps fi : Z → Xi such that all triangles commute in

... ϕ21
/ X1 / X0
ϕ10
E y<
y
yy
yyy
yy
f
. . . f1 0

yyy
yy
yy
yy
Z

there is a unique map f : Z → X such that all triangles commute in

ϕ0

ϕ1
# $
XX2 ... ϕ21
/ X1 / X0
ϕ10
E y<
2 yyy
2
f0 yy
yy
2 . . . f1 y
f 2 y
yy
2 yyy
2 yy
y
Z

[1.0.2] Remark: Note that the definition of the limit definitely does depend on the transition maps among
the objects of which we take the limit, not just on the objects.

As usual with mapping-property definitions:

[8] Even though the limit of spaces R/nZ fails to be R in several ways, this flawed notion can be used as a heuristic
to see how Fourier inversion on R could be inferred from inversion for Fourier series.

3
 Paul Garrett: Solenoids (September 11, 2010)

[1.0.3] Theorem: If a (projective) limit exists it is unique up to unique isomorphism. (Proof in appendix:
it works for the usual abstract reasons.)
A little more concretely, we can prove existence of limits from existence of products (at least for topological
spaces):

[1.0.4] Proposition: A limit X (and maps ϕi : X → Xi ) of a family

ϕn+1,n ϕn,n−1
... / Xn / Xn−1ϕn−1,n−2/ . . . ϕ21
/ X1 ϕ10
/ X0

is a subset X (with the subset topology) [9] of the product [10] Y = i Xi (with projections pi : Y → Xi ) on
Q
which the projections are compatible with the transition maps ϕi,i−1 , that is,

X = {x ∈ Y : ϕi,i−1 (pi (x)) = pi−1 (x) for all i}

with maps ϕi obtained by restriction of the projection maps pi from the whole product to X, namely

ϕi = pi |X : X → Xi

Proof: First, let j : X → Y be the inclusion of X into Y , and let ϕi : X → Xi be the restriction of the
projection pi : Y → Xi to the subset X of the product Y . That is,

ϕi = pi ◦ j

Then, before thinking about any other space Z and other maps, we do have a diagram

ϕ0

ϕ1
p0

p1
# $
X
j
/Y ... X1 X0

with commuting (curvy) solid triangles. While the maps from Y do not respect the transition maps
ϕi,i−1 : Xi → Xi−1 , by the very definition of the subset X of Y , the restrictions ϕi = pi ◦ j of the
projections pi to X do respect the transition maps. Thus, the solid triangles commute in the diagram

ϕ0

ϕ1
p0

p1
# $
X
j
/Y ... ϕ21
/ X1 ϕ10
/ X0

but not necessarily any triangle involving dotted arrows.

[9] The subset topology on a subset X of a topological space Y can be characterized as the topology on X such that

the inclusion map j : X → Y is continous, and such that every continuous map f : Z → Y from another space Z
such that f (Z) ⊂ X factors through the inclusion. That is, there a continuous F : Z → X such that f = i ◦ F . This
does not prove existence. Also, one can show that the subspace topology is the coarsest topology on X such that the
inclusion X → Y is continuous. Finally, the construction of this topology, which proves existence, is that a set U in
X is open if and only if there exists an open V in Y such that U = X ∩ V .
[10] By now we know that the usual product can be characterized intrinsically, and that intrinsic characterization is

all we use in this proposition.

4
 Paul Garrett: Solenoids (September 11, 2010)

Now consider another space Z. By the mapping properties of the product, for any collection of maps
fi : Z → Xi (not only those meeting the compatibility condition ϕi,i−1 ◦ fi = fi−1 ) there is a unique
F : Z → Y through which all the projections pi : Y → Xi factor. That is, we have a unique F such that the
triangles commute in the diagram

ϕ0

ϕ1
p0

p1
# $
j
/Y ... X
X 2X E 1 < X0
2 yyy
y
2
f0 yy
yy
2 . . . f1
F 2 yy
2 yyy
y
2 yyy
y
Z

Note that we cannot include the transition maps in the diagram since the projections pi : Y → Xi do not
respect them. But, since the maps fi are compatible with the maps ϕi,i−1 , we could suspect that the image
F (Z) ⊂ Y is a smaller subset of the product Y . Indeed, for z ∈ Z, using the compatibility

pi−1 (F (z)) = fi−1 (z) = ϕi,i−1 (fi (z)) = ϕi,i−1 (pi (F (z)))

we see that F (Z) ⊂ X, as claimed. That is, F factors through the inclusion map j : X → Y , and the
composites pi ◦ F factor through j : X → Y , giving a picture with commuting solid or dashed (but not
dotted) triangles
ϕ0

ϕ1
p0

p1
# $
X aC
j
/Y ... ϕ21
/ X1 / X0ϕ10
X2 E
y <
C 2 yy
C y
C 2
f0 yy
yy
C 2 . . . f1 y
f CF 2
y
C 2 yy
C yyy
C2 yyy
Z

(Again, the projections from Y do not respect the transition maps.) That is, with the compatibility
conditions, the maps from Z do factor through the subset X of the product. ///

This general argument gives some surprising qualitative information about projective limits:

[1.0.5] Corollary: The projective limit of a family Xi of compact [11] Hausdorff spaces is compact.
[1.0.6] Remark: In particular, the (projective) limit of circles is compact, since circles (with their usual
topologies) are compact. In particular, this limit cannot be the non-compact . R
[11] We need a better definition of compact than the metric-space definition that every sequence contains a convergent

subsequence. Instead, we need the definition that both applies to general topological spaces and is more useful. That
is, first in words, a set E inside a topological space is compact if every open cover admits a finite subcover. That is,
S S
for E ⊂ i∈I Ui with opens Ui , there is a finite subset Io of I such that still E ⊂ i∈Io Ui . It is not obvious that
this definition is superior to the sequence definition.

5
 Paul Garrett: Solenoids (September 11, 2010)

Proof: (of corollary) The essence of the argument is that products of compact spaces are compact,
Q by
Tychonoff, and closed subspaces of compact Hausdorff [12] spaces are compact. Thus, the product Y = i Xi
is compact. The compatibility conditions ϕi,i−1 (pi (x)) = pi−1 (x) are closed conditions in the sense that
{x ∈ Y : ϕi,i−1 (pi (x)) = pi−1 (x)} = closed set in Y
since the maps pi and ϕi,i−1 are continuous and since the spaces Xi are assumed Hausdorff. [13] [14] The
intersection of an arbitrary family of closed sets is closed, [15] so the (projective limit) X of points x meeting
this condition for all i, is closed. And in a compact space Y , closed subsets are compact. [16] ///

[1.0.7] Remark: By paraphrasing the assertion of the proposition, we now have a concrete (if not perfectly
useful) model of the limit X in a diagram
ϕ0
ϕ1

ϕ2
# $ %
X ... ϕ32
/ X2 ϕ21
/ X1 ϕ10
/ X0

with spaces indexed by non-negative integers, namely, the collection of all sequences xo , x1 , x2 , . . . such that
the transition maps ϕn,n−1 map them to each other, that is,
ϕn,n−1 (xn ) = xn−1
for all indices n. This follows from the usual model of the product as Cartesian product, which for countable
products can be written as the collection of all sequences xo , x1 , x2 , . . . with xi ∈ Xi . We may choose to
write a compatible family of elements as
. . . → x3 → x2 → x1 → x0
[12] Recall that a topological space is Hausdorff if any two points have disjoint neighborhoods. It is useful to know

that Z is Hausdorff if and only if the diagonal Z ∆ = {(z, z) ∈ Z × Z : z ∈ Z} is closed in Z × Z. Indeed, for Z
Hausdorff, points x 6= y in Z have disjoint neighborhoods U and V . Then U × V is open in the product topology in
Z × Z, contains x × y, and since U ∩ V = φ the set U × V does not meet the diagonal {(z, z) : z ∈ Z} in Z × Z.
Thus, the diagonal is the complement of the union of all such opens U × V , so is closed. The converse reverses the
argument: for closed diagonal, given x 6= y in Z, there is an open U × V containing x × y and not meeting the
diagonal, since the product topology has sets U × V as a basis. Since U × V does not meet the diagonal, U and V
are disjoint neighborhoods of x, y in Z.
[13] As a critical auxiliary point, we should note that for any topological space X the diagonal imbedding δ : X →

X × X by δ(x) = (x, x) is a homeomorphism (topological isomorphism) to the image, with the subspace topology.
Certainly δ is a set bijection. For a neighborhood U of x in X, the open U × U in X × X meets δ(X) at δ(U ). On
the other hand, given opens U, V in X, the basis open U × V in X × X meets δ(X) in δ(U ∩ V ), and, indeed, U ∩ V
is open. Thus, the images by δ of opens are open, and vice-versa.
[14] Characterizing Hausdorff-ness by the closed-ness of the diagonal is useful to show that for continuous maps

f : X → Z and g : X → Z with Z Hausdorff, the set {x ∈ X : f (x) = g(x)} is closed, as follows. The map
(f × g)(x, y) = f (x) × g(y) from X × X to Z × Z is continuous, that is, inverse images of opens are open. Then
inverse images of closed sets are closed, and the inverse image of Z ∆ under f × g is closed. The intersection of the
diagonal with the inverse image of Z ∆ by f × g is {(x, x) : f (x) = g(x)}. Closed-ness in X × X gives closedness
in the diagonal (with the subspace topology), and we just noted that the diagonal is homeomorphic (topologically
isomorphic) to X.
[15] That an arbitrary intersection of closed sets is closed is equivalent to the defining property that an arbitrary

union of open sets is open, since a set is closed if and only if its complement is open.
[16] This important fact is easy to prove: let E be a closed subset of a compact space Y , and let {U : i ∈ I} be an
i
open cover of E. Let U = Y − E. Then {Ui : i ∈ I} ∪ {U } is a cover of the entire space Y . By the compactness of Y ,
there is a finite subcover U1 , . . . , Un , U . (If E 6= Y the subcover must use U .) Then U1 , . . . , Un is a finite cover of E.

6
 Paul Garrett: Solenoids (September 11, 2010)

This description of the limit as a set of sequences is deficient in several regards (for example, it does not tell
us a topology), but it is occasionally useful, certainly as a heuristic.

2. Automorphisms of solenoids
Even without trying to imagine what meaning to attach to a solenoid X or other limit object, we can
directly make sense of automorphisms of X by looking at automorphisms [17] of the diagram. Then, with a
large-enough group G of automorphisms to act transitively [18] on X, we can write X as a quotient
X ≈ G/Gx = {Gx -cosets in G} = {gGx : g ∈ G}
of G, where Gx is the isotropy subgroup [19] (in G) of a point x in X.
One virtue of identifying automorphisms X ≈ G/Gx is that this identification might be done piece-by-piece,
identifying subgroups of the whole group, then assembling them at the end. And it is important to note
that this is an isomorphism of G-spaces, meaning (topological) spaces A, B on which G acts continuously.
As expected, a map of G-spaces is a set map ψ : A → B such that
ψ(g · a) = g · ψ(a)
for a ∈ A, g ∈ G, where on the left the action is of G on A, and on the right it is the action on B.
True, the solenoid is itself a group already, being a projective limit of groups, so this approach might seem
silly. However, we can present the solenoid as a quotient of more familiar (and simpler) objects. In any case,
since we’ll consider an abelian group G of automorphisms of the solenoid, any group quotient G/Gx is again
a group, [20] and, incidentally, Gx and the quotient are independent of x.
Thus, even without thinking of projective limits, one kind [21] of automorphism f of the 2-solenoid is a
R Z R Z
collection of maps fn : /2n → /2n such that all squares commute in the diagram

. . . ϕ / R/8Z
43 ϕ32
/ R/4Z ϕ21
/ R/2Z ϕ10
/ R/Z
f3 f2 f1 f0
   
... ϕ43
/ R Z
/8
ϕ32
/ R Z /4
ϕ21
/ R Z
/2
ϕ10
/ RZ
/

Without being too extravagant [22] we want to think of some obvious families of maps fn . Since all our
R
circles are quotients of in a compatible fashion, we can certainly create a simple sort of family of maps fn
R
by letting r ∈ act, by
fn (xn + 2n ) = xn + r + 2nZ Z
[17] Note that this discussion is different from the argument that objects defined by mapping properties have no

endomorphisms that leave the other objects unmoved. Here we are moving the objects in the diagram.
[18] Again, for a group to act transitively means that G moves any point of X to any other point, that is, for x, y ∈ X

there is g ∈ G such that gx = y.

[19] Again, with a group G acting on a set X, the isotropy subgroup G of an element x ∈ X is the subgroup not
x
moving x, that is, Gx = {g ∈ G : gx = x}. It is straightforward to see that this is a subgroup of G, not merely a
subset.
[20] We will attend to topological details slightly later.

[21] It is not a priori clear that any useful collection of maps would necessarily send each R/2n Z to itself, but this
will suffice for now.
[22] But sometimes extravagance can have a simplicity that is hard to achieve otherwise.

7
 Paul Garrett: Solenoids (September 11, 2010)

with the same real number r for every index. [23]

[2.0.1] Remark: We are neglecting continuity, but will return to this point when we recapitulate this
discussion of automorphisms in a form that is better suited to discussion of the topology. This copy of R
does act continuously. One may verify that it is not transitive, and that the isotropy groups in of points R
in the solenoid are trivial. Thus, overlooking the failure of the action to be transitive, one might naively
R R
imagine that the limit is a copy of . True, the orbit · x of any given point is dense [24] in the solenoid,
but it is not closed. [25]
Another relatively simple family of maps is created by taking a sequence of integers yn and maps

Z
fn (xn + 2n ) = xn + yn + 2n Z
and requiring that the sequence yn be chosen so that the squares in the diagram commute. That is, we must
have
Z Z
(xn + yn + 2n ) + 2n−1 = xn−1 + yn−1 + 2n−1 Z
Since already
Z
(xn + 2n ) + 2n−1 Z = xn−1 + 2n−1Z
it is necessary and sufficient that

Z
(yn + 2n ) + 2n−1 Z = yn−1 + 2n−1Z
That is, the compatible sequence of integers yn gives an element in another projective limit, the 2-adic
Z
integers [26] 2 .
... mod 8/
Z/8Z mod 4/
Z/4Z mod 2/
Z/2Z mod 1/
Z/ Z
Each of the limit objects is finite, so certainly compact. Thus, this projective limit is compact, whatever
other features it may have.

Still without worrying about the topology, we claim

[2.0.2] Proposition: The product group R × Z2 acts transitively on the 2-solenoid. The point → 0 → 0 → 0
in the solenoid has isotropy group which is the diagonally imbedded copy of the integers

Z∆ = {(`, −`) ∈ (Z × Z) ⊂ R × Z2 : `∈ } Z

Proof: Given a compatible family

Z Z
. . . → x3 + 8 → x2 + 4 → x1 + 2 → x0 + Z Z
[23] And the x ∈
i R/2i Z are chosen compatibly in the first place, that is, such that (xi +2i Z)+2i−1 Z = xi−1 +2i−1 Z,
for all indices i.
[24] Recall that a subset E of a topological space X is dense if every non-empty open set in X has non-empty

intersection with E.
[25] This highly-wound copy of R may be the thing that earned the name solenoid.
[26] Replacing 2 by another prime p throughout gives a p-solenoid and p-adic integers Z
p . This approach is not the
Z
most conventional way to present the p-adic integers, but does illustrate the role that p plays in situations that are
Z
not obviously number-theoretic. We will review a more conventional description of p later, for comparison.

8
 Paul Garrett: Solenoids (September 11, 2010)

Z R Z
of elements xn + 2n ∈ /2n , act by r ∈ R RZ
as above such that x0 + r = 0 ∈ / . Since the xn ’s are
compatible, it must be that (r +x1 ) mod 1 = (x0 +r) = 0, (x2 +r) mod 2 = (x1 +r), (x3 +r) mod 4 = x2 +r,
Z
and so on. That is, every xn + r ∈ , and the sequence yn = xn + r gives a compatible family
Z Z
. . . → y3 + 8 → y2 + 4 → y1 + 2 → y0 + Z Z
which gives an element in Z2. That is, the further action by −yn on the solenoid will send every element to
(xn + r) − yn = (xn + r) − (xn + r) = 0
This proves the transitivity.
To determine the isotropy group of a point, suppose that r is a real number and the yn is an integer modulo
2n , such that the 0-element
...0 → 0 → 0 → 0
is mapped to itself. That is, require that
0 + r + yn ∈ 0 + 2n Z
Z
for all n. First, this implies that r ∈ . Then yn , which is only determined modulo 2n anyway, is completely
determined modulo 2n by
yn + 2n = −r + 2nZ Z
n
That is, yn = −r mod 2 . And these conditions are visibly sufficient, as well, to fix the 0. Thus, the isotropy
group truly is the diagonal copy of . Z ///

[2.0.3] Corollary: (Still not worrying about the topology) the 2-solenoid is isomorphic to the quotient
( R × Z2)/Z∆
Proof: Notably ignoring the topology, whenever a group G acts transitively on a set X containing a chosen
element x, there is a bijection
X ←→ G/Gx = {gGx : g ∈ G}
by
gx ←→ gGx
A map from G to X by g → gx is a surjection, since G is transitive. This map factors through G/Gx and is
injective, since gx = hx if and only if h−1 gx = x, if and only if h−1 g ∈ Gx , if and only if gGx = hGx .
///

[2.0.4] Remark: We need a somewhat better set-up to keep track of the topologies.

3. A cleaner viewpoint
Having run through an informative heuristic about the structure of the solenoid as a quotient G/Gx , we can
redo things more elegantly, and not lose sight of the topological features of the situation.
First, in any projective limit, families of maps [27] fn : Xn → Xn such that all squares commute in

... ϕ32
/ X2 ϕ21
/ X1 ϕ10
/ X0
O O O
f2 f1 f0

... ϕ32
/ X2 ϕ21
/ X1 ϕ10
/ X0

[27] Again, maps here are continuous maps, but the arguments do not use this explicitly.

9
 Paul Garrett: Solenoids (September 11, 2010)

do give rise to a map f : X → X of the projective limit X = limn Xn to itself, as follows. Again, from the
definition of the projective limit X of the Xn , to give a map F : Z → X is to give a compatible family of
maps Fn : Z → Xn , meaning that all triangles commute in

... ϕ21
/ X1 / X0
ϕ10
E y<
y
yy
yyy
F 0 y
. . . F1 y
yyy
yy
yy
yy
Z

This induces a unique map F : Z → X, making a commutative diagram

p0

p1
# $
XX2 ... ϕ21
/ X1 / X0
ϕ10
E y<
2 yyy
2 yy
2 . . . F1 F0 yyy
F 2 yy
2 yyy
2 yyy
y
Z

In particular, for a compatible family of maps fn : Xn → Xn we can take Z = X and Fn = fn ◦ pn , giving a

commutative
p0

p1
# $
... ϕ21
/ X1 ϕ10 / X0
X <y O 8p O
y p
y p pp
... F1 y F0 p
y p f1 f1
y p
y pp
y p
py p . . . ϕ21 / ϕ10
/ X0
X < X1 :
p1

p0

which then yields a unique F : X → X in a commutative diagram

p0

p1
# $
... / X1 ϕ10 / X0
ϕ21
XO 8
y< O pp
O
y p
y
... F1 y F0 p
pp
F y pp f1 f1
y
y pp
y p
py p . . . ϕ21 / ϕ10
/ X0
X < X1 :
p1

p0

That is, automorphisms of diagrams (in the sense of the previous section) do give automorphisms of the
projective limit objects attached to the diagrams.

10
 Paul Garrett: Solenoids (September 11, 2010)

We also observe that we can identify points in the projective limit as compatible sequences

. . . → x3 → x2 → x1 → x0

with xn ∈ Xn (and the compatibility ϕn,n−1 (xn ) = xn−1 ) without using the Cartesian product model of the
product and identifying the projective limit inside that Cartesian product. To do so, recall the trick that for
any set Y and for {s} a set with a single element, we have a natural bijection

µY : Y = {elements of Y } ←→ {maps {s} → Y }

by
µY : y → f with f (s) = y
These maps µY are natural in the precise sense that for a set map f : Y → Z, we have a commutative
diagram
Y
µY
/ {maps {s} → Y }

f f ◦−
 
Z
µZ
/ {maps {s} → Z}

where f ◦ − is post composition with f , that is ϕ → f ◦ ϕ. And since maps to a projective limit X = lim Xn
are given exactly by compatible family of maps to the Xn , maps of S = {s} to X are given by compatible
families of maps to the Xn as in
p0

p1
# $
X `A ... ϕ21
/ X1 ϕ10 /6 X0
|= nnn
A
A ||| nnnnn
|| nn
A |n|nnnn
{s}
That is, elements of X are given by compatible families of elements of the Xn , as claimed. This will be
useful in proving transitivity of a group action.
A topological group is a group G which has a topology such that multiplication g × h → gh and inversion
g → g −1 are continuous maps G × G → G and G → G, and G is locally compact [28] and Hausdorff. [29]
Further, it is often necessary or wise to require that a topological group have a countable basis. [30] An
action of a topological group G on a topological space X is continuous if the map

G × X → X by g × x → gx is continuous

Now we see how to get an action of a topological group G on a projective limit.

[3.0.1] Claim: Let an : G × Xn → Xn be continuous group actions of a topological group G on topological

spaces Xn , and suppose that these actions are compatible in the sense that squares commute in the diagram

... ϕ32
/ X2 ϕ21
/ X1 ϕ10
/ X0
O O O
a2 a1 a0

. . .idG ×ϕ32/ G × X2idG ×ϕ21/ G × X1idG ×ϕ10/ G × X0

[28] A topological space is locally compact if there is a basis of (open) subsets each having compact closure.

[29] A topological space is Hausdorff if any two distinct points have neighborhoods disjoint from each other.

[30] A topological space X has a countable basis if, as suggested by the terminology, it has a basis that is countable.

11
 Paul Garrett: Solenoids (September 11, 2010)

Then there is a unique continuous group action a : G × X → X on the projective limit X such that we have
a commutative diagram [31]
p0

p1
p2
% &
XO ... ϕ32
/ X2 ϕ21
/ X1 ϕ10
/' X0
O O O


a  a2 a1 a0


G×X . . .idG ×ϕ32/ G × X2idG ×ϕ21/ G × X1idG ×ϕ10/ G × X0
9 7 6
idG ×p2
idG ×p1

idG ×p0

Proof: Composing the maps idG × pn : G × X → G × Xn with the action map an : G × Xn → Xn gives
a compatible family of maps G × X → Xn . By definition of the projective limit X, we have a unique map
G × X → X making the diagram commute, as claimed.
But we should really check the associativity property (gh)x = g(hx) required of a group action, with g, h ∈ G
and x ∈ X, not to mention the condition eG x = x. (Unsurprisingly, it turns out fine.) We need to rewrite the
associativity in terms of maps. In a diagram, the associativity of the action on Xn asserts the commutativity
of the triangle
Xn


@ ^===

==
(g,h,x)→g(h(x))




== (g,h,x)→(gh)(x)

==


==


==


==

idG ×idG ×idXn
G × G × Xn G × G × Xn
That is, associativity is equivalent to the equality of two maps G × G × Xn → Xn . Thus, by the uniqueness
of the induced map on the projective limit X, we obtain the same limit maps G × G × X → X. This gives
the associativity on the projective limit from the known associativities. We did not bother to prove that the
identity acts trivially. ///

In a similar vein, thinking of our glib presumption that the projective limit Z2 of the groups Z/2nZ was a
group, not to mention a topological group, we should verify these things.

[3.0.2] Claim: Projective limits of topological groups, with all but finitely many compact, are topological
groups. [32] Further, countable projective limits [33] of countably-based topological groups have countable
bases.

[3.0.3] Remark: The proof has several parts, which show somewhat more than the claim asserts. For
example, it becomes clear that arbitrary projective limits of groups exist. Arbitrary projective limits of
[31] In this diagram, there is no claim that G × X is the projective limit of the objects on the bottom row, only that

the maps to the top row exist as indicated.

[32] That is, such projective limits of topological groups exist, as topological groups.

[33] All our diagrams have implicitly used only countably-many objects in the family from which the projective limit

is formed. Nevertheless, this countability is not mandated in a more general notion of projective limit, so should be
explicitly noted when it matters.

12
 Paul Garrett: Solenoids (September 11, 2010)

Hausdorff spaces are Hausdorff. Projective limits of families of locally compact Hausdorff spaces Xi , with
all but finitely many Xi compact, are locally compact. And countable limits of countably-based topological
spaces are countably-based.

Proof: Let
p0

p1
# $
G ... ϕ21
/ G1 ϕ10
/ G0

be a projective limit of topological groups, where each transition map ϕi,i−1 is a continuous group
homomorphism, and the pi are continuous maps from the projective limit object G. All that we truly
know about G and the pi at the outset is that G is a topological space and that the pi are continuous. We
must prove that G is a group, in fact a topological group, and that the pi are group homomorphisms.
First, we need to find the very definition of the alleged group operation G × G → G on the limit object,
much as we defined the group action on a limiting object above. Of course, this must be some sort of limit
of the multiplication maps µn : Gn × Gn → Gn by µn : g × h → gh. At the same time, to make a map
G × G → G is to make a compatible family of maps fn : G × G → Gn . Indeed, let

fn = µn ◦ (pn × pn ) : G × G → Gn

That is, there is a unique µ : G × G → G induced by the fn making a commutative diagram

p0

p1
'
/7 G1
ϕ21 ϕ10
/)
h h4 GO 0
GO ...
o o O h
h
oo h h
o h h h
f1 o f0 h h
µ
o oo h h h h µ1 µ0
o h
o h h h
oh oh h ϕ21 ×ϕ21
G×G ... / G1 × G1 ϕ10 ×ϕ10 / G0 × G0
7 5
p1 ×p1
p0 ×p0

The associativity a(bc) = (ab)c of the alleged [34] group operation comes (much as in the discussion of group
actions on limits), first from the commutativity of the diagrams

Gn fM
qq8 MMM
qq qqq
(a,b,c)→a(bc) MMM
(a,b,c)→(ab)c
MMM
qq M
qq idGn ×idGn ×idGn
Gn × GO n × Gn Gn × GO n × Gn
pn ×pn ×pn pn ×pn ×pn
idG ×idG ×idG
G×G×G G×G×G

which proves that the two different maps G × G × G → Gn are the same, and, second, the uniqueness of the

[34] In fact, it is slightly dangerous to use this notation, since it makes it too easy to lose track of what we truly know,

versus what must be shown. The associativity we want to prove is properly written as µ(a, µ(b, c)) = µ(µ(a, b), c)).

13
 Paul Garrett: Solenoids (September 11, 2010)

dotted induced map in

p0

p1
ϕ21
/6( G1 ϕ10
/*
g g3 GO 0
GO ...
m m O g g
m g g
m m m g g g g
m g g
m m m g g g g g a(bc)=(ab)c a(bc)=(ab)c
m g g
m m g g g
m mg g g
G×G×G . . . ϕ21 ×ϕ21 ×ϕ21 / G1 × G1 × G1 ϕ10 ×ϕ10 ×ϕ10 / G0 × G0 × G0
6 4
p1 ×p1 ×p1
p0 ×p0 ×p0

The identity element e in the limit is specified as a sort of limit of the identities en in Gn , specifically, as the
image f (s) of the induced map f in the diagram
p0

p1
# $
G `A ... ϕ21
/ G1 ϕ10 /6 G0
=| nn
A s→e1 || nnn
A || nnnn
| n
A |n|nnnn
f s→e0

{s}

Existence of an inversion map (and its property) is a further exercise in this technique, which we leave to
the reader. Thus, the map µ : G × G → G does have the properties of a group operation on G.

To show that the projections pn : G → Gn are group homomorphisms, we note that to say that f : A → B
is a group homomorphism for groups A, B is to require the commutativity of the square

AO
f
/B
O
µA µB
f ×f
A×A / B×B

where µA and µB are the multiplication maps belonging to A, B, respectively. In the case at hand, we would
want the commutativity of
GO
pn
/ Gn
O
µ µn
pn ×pn
G×G / Gn × Gn

Happily, the commutativity of these squares is part of the commutativity of the diagram defining the
multiplication µ : G × G → G. That is, the fact that the projections are group homomorphisms is a
by-product of the construction of the multiplication on G.
The Hausdorff-ness of the limit will follow from the earlier observation that a limit limi Xi is a subspace of
the corresponding product Πi Xi . An arbitrary product of Hausdorff spaces is Hausdorff. [35] And arbitrary
[35] That products of Hausdorff spaces are Hausdorff has a natural proof, as follows. Given x 6= y in the product of

spaces Xi , there is at least one index j such that the projection pj of the product to Xj distinguishes x and y, that
is, such that pj x 6= pj y. (This assertion itself can be proven as an exercise using the mapping property definition of
product, as suggested in an earlier handout.) Since Xj is Hausdorff, there are disjoint neighborhoods Uj and Vj of
pj x and pj y. Perhaps using the explicit construction of products as cartesian products, let U = Uj × Πi6=j Xj and
V = Vj × Πi6=j Xj . These are disjoint neighborhoods of x and y.

14
 Paul Garrett: Solenoids (September 11, 2010)

subspaces of Hausdorff spaces, given the subspace topology, are Hausdorff. [36] Thus, limits of Hausdorff
spaces are Hausdorff.
Similarly, the local compactness of limits of locally compact topological spaces Xi , with all but finitely many
compact, will follow from the analogous assertion for products. First, observe that a basic open Πi Ui in a
product has closure the product of the closures Ui of the factors Ui . [37] When all the Ui are compact, the
product is compact, by Tychonoff’s theorem. Since only finitely-many Xi are not in fact compact themselves,
but in any case are still locally compact, inside such factors the product topology allows us to take compact-
closure neighborhoods of any point. Thus, every point in the product has a neighborhood (in fact, a basic
neighborhood) with compact closure.

Finally, to discuss countable-based-ness, it suffices to prove that countable products of countably-based

topological spaces Xi are countably-based, since limits are subspaces of products. A basis for a product
topology consists of products Πi Ui where for each i the set Ui is in a countable basis for Xi , and for all
but finitely-many indices i the set Ui is just Xi . To count these possibilities, note first that there are only
countably-many finite subsets of a countable set. Next, for each finite subset {i1 , . . . , in } of the countable
indexing set, there are only countably-many choices of

Ui1 , . . . , Uin (with Uij a basis element in Xij )

The sum of countably many countables is countable. ///

[3.0.4] Remark: The box topology behaves worse than the genuine product topology with regard to
preservation of countable-based-ness: since there are uncountably many not-necessarily-finite subsets of a
countable index set, a product of (infinitely) countably-many countably-based spaces, with the box topology,
will not be countably-based.
Having verified that things work as hoped, especially that topological aspects and group-theoretic aspects
are captured, we return to the solenoid.

4. Automorphisms of solenoids, again

Having cleaned up our viewpoint, we can give an economical and rigorous treatment of the automorphisms
found earlier of the solenoid
p0
p1
$ &
X ... mod 2/
R/2Z mod 1/
R/Z
Our aim is to prove that we have a transitive (continuous) group action of R × Z2, with an isotropy group
Z Z
a diagonal copy ∆ of the integers , and, thus, that

2-solenoid ≈ ( R × Z2)/Z∆
as G-spaces. [38]
[36] That subspaces Y of Hausdorff spaces X are Hausdorff is straightforward: given x 6= y in Y , let U, V be disjoint

neighborhoods of x, y in X. Then U ∩ Y and V ∩ Y are disjoint neighborhoods of x, y in Y .

[37] The product of the closures U of the opens U is a closed set (from the definition of product topology) and contains
i i
the product of the Ui . On the other hand, let x be in the product of the closures. Then every basic neighborhood
Πi Vi of x (with Vi open in Xi ) has the property that Vi meets Ui , since pi x is in the closure of Ui . That is, Πi Vi
meets Πi Ui , so x is in the closure of the product. This proves the equality.
[38] Again, the notion of G-space is the reasonable one, of topological spaces acted upon continuously by a topological

group G. A map of G-spaces ψ : A → B is a continuous map of topological spaces which respects the action of G, in
the sense that ψ(g · a) = g · ψ(a).

15
 Paul Garrett: Solenoids (September 11, 2010)

First, we have an induced continuous group action R

× X → X (the dotted arrow below) induced by
the compatible family of (dashed arrow) maps R R Z
× X → /2n created by composition of the actions
R R Z R Z
× /2n → /2n with the projection × X → × /2n , in R R R Z
p0

p1
( )
XO ... ϕ21

o7
/
h h
R/2O Z
h
/4 /
O
ϕ10
RZ
o o h h h
o h h
oo h h
o h h h (r,x1 )→r+x1
h
oo h h h
(r,x1 )→r+x0
o
o h h h
ohoh h
R ×X ... idR ×ϕ21
/ × /2
6
R R Z
idR ×ϕ10
/ × /
4
R RZ
idR ×p1
idR ×p0

The diagrammatic form of the action of the projective limit (countably-based topological) group

mod 1

mod 2
$ &
Z2 ... mod /2
Z/21Z mod 1/
Z/20Z
on the solenoid X is nearly identical, with the minor complication that the action of Z2 on R/2nZ is via the
Z Z R Z
image group /2n action on /2n , by definition.
Next, we want to prove transitivity of the joint action R Z
× 2 on the solenoid. Specify a point x on the
solenoid by a compatible family of maps (and the induced map f to X)

p0

p1
$ &
X `@ ... mod 2 /
R< /2Z
kk5
mod 1/
R/Z
@ y yy kkkkk
@ y k
@ yy kkkkf0
f
ykykkkfk1
{s}

R RZ
The action of is transitive on the rightmost circle / , so is transitive on maps f0 from {s} to that circle.
Thus, given a point f on the solenoid (given by a family {fn } of maps from {s}), we adjust it by so that R
RZ
f0 (s) = 0 in / .

Then the compatibility condition on the images fn (s) requires that, given f0 (s) = 0, all fn (s) are inside
Z Z R Z
/2n ⊂ /2n . That is, the family of maps fn gives a compatible family

^= ... mod 2 /
<
/2Z Zmod 1/
lll
5 / ZZ
= yy ll
= yy lll
= yyylfl1llllf0
ylll
{s}

which is exactly our definition of Z2 . Thus, visibly this Z2 maps all these points to 0. This proves the
transitivity.
Thus, certainly as sets,
2-solenoid ≈ ( R × Z2)/Z∆
16
 Paul Garrett: Solenoids (September 11, 2010)

The surprising result proved in the appendix will imply that this is a topological isomorphism, if we are sure
that R Z
× 2 has a countable basis. It is standard [39] that R
has a countable basis. It is less standard,
Z
but still standard in light of our earlier discussion, that 2 has a countable basis, since it is a countable
projective limit of countably-based spaces. [40]

5. Appendix: uniqueness of projective limits

As an exercise in proving the uniqueness-up-to-unique-isomorphism (assuming existence) of things specified
by universal mapping properties, we carry out the proof of uniqueness of projective limits. Part of the point
of the exercise is reiteration of the inessentialness of the details of the situation. In particular, as above, a
mapping-property approach provides a very useful packaging for topological details that might otherwise be
burdensome.

Thus, given topological spaces Xi with continuous maps

... ϕ21
/ X1 ϕ10
/ X0

let X and projections pi and Y and projections qi fit into diagrams

p0 q0

p1 q1
# $ # $
X ... ϕ21
/ X1 ϕ10
/ X0 Y ... ϕ21
/ X1 ϕ10
/ X0

such that, for all families of maps fi : Z → Xi such that all triangles commute in

... ϕ21
/ X1 / X0
ϕ10
E y <
yyy
f0 yy
yy
. . . f1 y
yy
yyy
yy
yy
Z

there are unique maps f : Z → X and g : Z → Y such that all triangles commute in both diagrams

p0 q0

p1 q1
# $ # $
XX2 ... ϕ21
/ X1 ϕ10
/ X0 Y X2 ... ϕ21
/ X1 ϕ10
/ X0
E y< E y <
2 yyy 2 yyy
2
f0 yy
yy 2
f0 yy
yy
2 . . . f1 y 2 . . . f1 y
f 2 y g 2 y
yy yy
2 yyy 2 yyy
2 yy 2 yy
y y
Z Z

Then
[39] The space R has a countable basis consisting of open balls with rational radii centered at rational points.
[40] Again, it is easy to see that a countable product of countably-based spaces is countably-based, and the projective

limit can be realized as a subspace of the product, so is countably-based.

17
 Paul Garrett: Solenoids (September 11, 2010)

[5.0.1] Claim: There is a unique isomorphism q : Y → X such that we have a commutative diagram
p0

p1
# $
XO ... ϕ21
/ X1 ϕ10
/ X0
q

Y ... ϕ21
/ X1 ϕ10
/ X0
< :
q1

q0

Proof: First, we prove that the only map of a projective limit to itself compatible with all projections is the
identity map. That is, using pi : X → Xi itself in the role of fi : Z → Xi , we find a unique map p : X → X
such that all triangles commute in

p0

p1
# $
XX2 ... ϕ21
/ X1 / X0
ϕ10
E y<
2 yyy
2
p0 yy
yy
2 . . . p1 y
p 2 yy
2 yyy
2 yyyy
X

Since the identity map idX fits the role of p, by uniqueness p can only be the identity on X.
Now we can do the main part of the proof. Let qi : Y → Xi take the role of fi : Z → Xi . Then there is a
unique q : Y → X such that all triangles commute in

p0

p1
# $
XX2 ... ϕ21
/ X1 ϕ10
/ X0
E y<
2 yyy
2
q0 yy
yy
2 . . . q1 y
q 2 yy
2 yyy
2 yyy
y
Y

To show that q is an isomorphism, reverse the roles of X and Y . Then there is a unique p : X → Y such
that all triangles commute in
q0

q1
# $
Y X2 ... ϕ21
/ X1 ϕ10
/ X0
E y <
2 yy
y
2
p0 yy
yy
2 . . . p1 y
p 2 y
yy
2 yyy
2 yyy
X

18
 Paul Garrett: Solenoids (September 11, 2010)

Then p ◦ q : Y → Y and q ◦ p : X → X are maps compatible with projections, so must be the identities, by
the first point of this argument. That is, these are mutually inverse maps, so q is an isomorphism. ///

[5.0.2] Remark: As usual in these categorical arguments, any continuity or other requirements on the
maps are packaged (or hidden) in the quantification over all families of maps fi : Z → Xi . That is, the
implicit specification that Z be a topological space and fi be continuous are what make the result relevant
to topological spaces and continuous maps. Thus, despite the lack of overt references to topology, the
uniqueness proven above yields topological isomorphisms, not merely set isomorphisms.

[5.0.3] Remark: As in our earlier discussion of the point that a projective limit of groups is a group, the
additional structure that must be demonstrated to have a group, as opposed to merely a set, is hidden in
the proof of existence of a projective limit. That is, in any case there is at most one, regardless of details,
but proof of existence invariable requires somewhat greater detail.

6. Appendix: topology of X  G/Gx

The point of this appendix is to prove that, with mild hypotheses, a topological space X acted upon
transitively by a topological group G is homeomorphic to the quotient G/Gx , where Gx is the isotropy group
of a chosen point x in X.
By the way, since we are not wanting to assume a pre-existing mastery of point-set topology, much less a
mastery of ideas about topological groups, several basic ideas will need to be developed in the course of the
proof. Everything here is completely standard and widely useful. The discussion includes a form of the Baire
Category Theorem [41] for locally compact Hausdorff spaces.

[6.0.1] Remark: Ignoring the topology, that is, as sets, the bijection G/Gx ≈ X is easy to see, and the
proof needs nothing. The topological aspects are not trivial, by contrast, and it should come as a surprise
that the topology of the group G completely determines the topology of the set X on which it acts.

[6.0.2] Proposition: Let G be a locally compact, Hausdorff topological group [42] and X a locally compact
Hausdorff topological space with a continuous transitive action of G upon X. [43] Suppose that G has a
countable basis. [44] Let x be any fixed element of X, and Gx the isotropy group [45] The natural map

G/Gx → X by gGx → gx

is a homeomorphism.

[41] The more common form of the Baire Category Theorem asserts that a complete metric space is not a countable

union of closed sets each containing no non-empty open set.

[42] As expected, this means that G is a group and is a topological space, the group multiplication is a continuous

map G × G → G, and inversion is continuous. The local compactness is the requirement that every point has an open
neighborhood with compact closure. The Hausdorff requirement is that any two distinct points x 6= y have open
neighborhoods U 3 x and V 3 y that are disjoint, that is, U ∩ V = φ.
[43] As expected, continuity of the action means that G × X → X by g × x → gx is continuous. The transitivity

means that for any x ∈ X the set of images of x by elements of G is the whole set X, that is, {gx : g ∈ G} = X.
[44] That is, there is a countable collection B (the basis) of open sets in G such that any open set is a union of sets

from the basis B.

[45] As usual, the isotropy (sub-) group of x in G is the subgroup of group elements fixing x, namely, G := {g ∈ G :
x
gx = x}.

19
 Paul Garrett: Solenoids (September 11, 2010)

Proof: We must do a little systematic development of the topology of topological groups in order to give a
coherent argument.

[6.0.3] Claim: In a locally compact Hausdorff space X, given an open neighborhood U of a point x, there
is a neighborhood V of x with compact closure V and V ⊂ U .

Proof: By local compactness, x has a neighborhood W with compact closure. Intersect U with W if
necessary so that U has compact closure U . Note that the compactness of U implies that the boundary [46]
∂U of U is compact. Using the Hausdorff-ness, for each y ∈ ∂U let Wy be an open neighborhood of y and
Vy an open neighborhood of x such that Wy ∩ Vy = φ. By compactness
T of ∂U , there is a finite list y1 , . . . , yn
of points on ∂U such that the sets Uyi cover ∂U . ThenS V = i Vyi is open and contains x. Its closure is
contained in U and in the complement of the open set i Wyi , the latter containing ∂U . Thus, the closure
V of V is contained in U . ///

[6.0.4] Claim: The map gGx → gx is a continuous bijection of G/Gx to X.

Proof: First, G × X → X by g × y → gy is continuous by definition of the continuity of the action. Thus,
with fixed x ∈ X, the restriction to G × {x} → X is still continuous, so G → X by g → gx is continuous.
The quotient topology on G/Gx is the unique topology on the set (of cosets) G/Gx such that any continuous
G → Z constant on Gx cosets factors through the quotient map G → G/Gx . That is, we have a commutative
diagram
/Z
y<
G
y
y
 y
G/Gx

Thus, the induced map G/Gx → X by gGx → gx is continuous. ///

[6.0.5] Remark: We need to show that gGx → gx is open to prove that it is a homeomorphism.
[6.0.6] Claim: For a given point g ∈ G, every neighborhood of g is of the form gV for some neighborhood
V of 1.

Proof: First, again, G × G → G by g × g → gh is continuous, by assumption. Then, for fixed g ∈ G, the

map h → gh is continuous on G, by restriction. And this map has a continuous inverse h → g −1 h. Thus,
h → gh is a homeomorphism of G to itself. In particular, since 1 → g · 1 = g, neighborhoods of 1 are carried
to neighborhoods of g, as claimed. ///

[6.0.7] Claim: Given an open neighborhood U of 1 in G, there is an open neighborhood V of 1 such that
V 2 ⊂ U , where
V 2 = {gh : g, h ∈ V }

Proof: The continuity of G × G → G assures that, given the neighborhood U of 1, the inverse image W of
U under the multiplication G × G → G is open. Since G × G has the product topology, W contains an open
of the form V1 × V2 for opens Vi containing 1. With V = V1 ∩ V2 , we have V 2 ⊂ V1 · V2 ⊂ U as desired.
///

[6.0.8] Remark: Similarly, but more simply, since inversion g → g−1 is continuous and is its own
(continuous) inverse, for an open set V the image V −1 = {g −1 : g ∈ V } is open. Thus, for example,

[46] As usual, the boundary of a set E in a topological space is the intersection E ∩ E c of the closure E of E and the

closure E c of the complement E c of E.

20
 Paul Garrett: Solenoids (September 11, 2010)

given a neighborhood V of 1, replacing V by V ∩ V −1 replaces V by a smaller symmetric neighborhood,

meaning that the new V satisfies V −1 = V .
The following result is not strictly necessary, but sheds some light on the nature of topological groups.

[6.0.9] Claim: Given a set E in G, \

closure E = E·U
U

where U runs over open neighborhoods of 1. [47]

Proof: A point g ∈ G is in the closure of E if and only if every neighborhood of g meets E. That is,
from just above, every set gU meets E, for U an open neighborhood of 1. That is, g ∈ E · U −1 for every
neighborhood U of 1. We have noted that inversion is a homeomorphism of G to itself (and sends 1 to 1),
so the map U → U −1 is a bijection of the collection of neighborhoods of 1 to itself. Thus, g is in the closure
of E if and only if g ∈ E · U for every open neighborhood U of 1, as claimed. ///

[6.0.10] Remark: This allows us to give another proof, for topological groups, of the fact that, given a
neighborhood U of 1 in G, there is a neighborhood V of 1 such that V ⊂ U . (We did prove this above for
locally compact Hausdorff spaces generally.)

Proof: First, from the continuity of G × G → G, there is V such that V · V ⊂ U . From the previous claim,
V ⊂ V · V , so V ⊂ V · V ⊂ U , as claimed. ///

[6.0.11] Remark: We can improve the conclusion of the previous remark using the local compactness of
G, as follows. Given a neighborhood U of 1 in G, there is a neighborhood V of 1 such that V ⊂ U and
V is compact. Indeed, local compactness means exactly that there is a local basis at 1 consisting of opens
with compact closures. Thus, given V as in the previous remark, shrink V if necessary to have the compact
closure property, and still V ⊂ V · V ⊂ U , as claimed.

[6.0.12] Corollary: For an open subset U of G, given g ∈ U , there is a compact neighborhood V of 1 ∈ G

such that gV 2 ⊂ U .

Proof: The set g−1 U is an open containing 1, so there is an open W 3 1 such that W 2 ⊂ g−1 U .
Using the previous claim and remark, there is a compact neighborhood V of 1 such that V ⊂ W . Then
V 2 ⊂ W 2 ⊂ g −1 U , so gV 2 ⊂ U as desired. ///

[6.0.13] Claim:
S Given an open neighborhood V of 1, there is a countable list g1 , g2 , . . . of elements of G
such that G = i gi V .

Proof: To see this, first let U1 , U2 , . . . be a countable basis. For g ∈ G, by definition of a basis,
[
gV = Ui
Ui ⊂gV

Thus, for each g ∈ G, there is an index j(g) such that g ∈ Uj(g) ⊂ gV . Do note that there are only countably
many such indices. For each index i appearing as j(g), let gi be an element of G such that j(gi ) = i, that is,

gi ∈ Uj(gi ) ⊂ gi · V

[47] This characterization of the closure of a subset of a topological group is very different from anything that happens

in general topological spaces. To find a related result we must look at more restricted classes of spaces, such as metric
spaces. In a metric space X, the closure of a set E is the collection of all points x ∈ X such that, for every ε > 0,
the point x is within ε of some point of E.

21
 Paul Garrett: Solenoids (September 11, 2010)

Then, for every g ∈ G there is an index i such that

g ∈ Uj(g) = Uj(gi ) ⊂ gi · V
This shows that the union of these gi · V is all of G. ///

A subset E of a topological space is nowhere dense if its closure contains no (non-empty) open set. [48]

[6.0.14] Claim: (Variant of Baire Category theorem) A locally compact Hausdorff topological space is not
a countable union of nowhere dense sets. [49]

Proof: Let Wn be closed sets containing no non-empty open subsets. Thus, any non-empty open U meets
the complement of Wn , and U −Wn is a non-empty open. Let U1 be a non-empty open with compact closure,
so U1 − W1 is non-empty open. From the discussion above, there is a non-empty open U2 whose closure is
contained in U1 − W1 . Continuing inductively, there are non-empty open sets Un with compact closures such
that
Un−1 − Wn−1 ⊃ Ūn
Certainly
Ū1 ⊃ Ū2 ⊃ Ū3 ⊃ . . .
T
Then Ūi 6= φ, by compactness. [50] [51]
Yet this intersection fails to meet any Wn . In particular, it cannot
be that the union of the Wn ’s is the whole space. ///

Now we can prove that G/Gx ≈ X, using the viewpoint we’ve set up.

Given an open set U in G and g ∈ U , let V be a compact S neighborhood of 1 such that gV 2 ⊂ U . Let
, . . . be a countable set of points such that G = i gi V . Let Wn = gn V x ⊂ X. By the transitivity,
g1 , g2S
X = i Wi .
We observed at the beginning of this discussion that G → X by g → gx is continuous, so Wn is compact,
being a continuous image of the compact set gn V . So Wn is closed since it is a compact subset of the
Hausdorff space X.
By the (variant) Baire category theorem, some Wm = gm V x contains a non-empty open set S of X. For
h ∈ V so that gm hx ∈ S,
gx = g(gm h)−1 (gm h)x ∈ gh−1 gm
−1
S
Every group element y ∈ G acts by homeomorphisms of X to itself, since the continuous inverse is given by
y −1 . Thus, the image gh−1 gm
−1
S of the open set S is open in X. Continuing,
gh−1 gm
−1
S ⊂ gh−1 gm
−1
gm V x ⊂ gh−1 V x ⊂ gV −1 · V x ⊂ U x
[48] The union of all open subsets of a given set is its interior. Thus, a set is nowhere dense if its closure has empty

interior.
[49] The more common verison of the Baire category theorem asserts the same conclusion for complete metric spaces.

The argument is structurally identical.

[50] In Hausdorff topological spaces X compact sets C are closed, proven as follows. Fixing x not in C, for each y ∈ C,

there are opens Uy 3 x and Vy 3 y with U ∩ V = φ, by the Hausdorff-ness. The Uy ’s cover C, so there is a finite
T
subcover, Uy1 , . . . , Uyn , by compactness. The finite intersection Wx = i Vyi is open, contains x, and is disjoint from
C. The union of all Wx ’s for x 6∈ C is open, and is exactly the complement of C, so C is closed.
[51] The intersection of a nested sequence C ⊃ C ⊃ . . . of non-empty compact sets C in a Hausdorff space X is
1 2 n
non-empty. Indeed, the complements Cnc = X − Cn are open (since compact sets are closed in Hausdorff spaces),
and if the intersection were empty, then the union of the opens Cnc would cover C1 . By compactness of C1 , there is
a finite subcollection C1c , . . . , Cnc covering C1 . But C1c ⊂ . . . ⊂ Cnc , and Cnc omits points in Cn , which is non-empty,
contradiction.

22
 Paul Garrett: Solenoids (September 11, 2010)

Therefore, gx is an interior point of U x, for all g ∈ U . ///

23