Chapter 0

Some Underlying Geometric Notions

Exercise 0.0.1. Construct an explicit deformation retraction of the torus with one point deleted onto a graph consisting
of two circles intersecting in a point, namely, longitude and meridian circles of the torus.

Solution. We know that the torus can be identified by the quotient space of
v b v

a e a

v b v.

Suppose the fundamental domain is [0, 1] × [0, 1]

Exercise 0.0.2. Construct an explicit deformation retraction of Rn − {0} onto S n−1 .

Solution. Define a map

π : Rn − {0} → S n−1
1
(x1 , · · · , xn ) 7→ p (x1 , · · · , xn ).
x21 + · · · + x2n

It is continuous because every coordinate component is continuous. Also, π|S n−1 = idS n−1 , therefore the map is
a deformation retraction.

Exercise 0.0.3. (i) Show that the composition of homotopy equivalences X → Y and Y → Z is a homotopy
equivalence X → Z . Deduce that homotopy equivalence is an equivalence relation.

(ii) Show that the relation of homotopy among maps X → Y is an equivalence relation.

(iii) Show that a map homotopic to a homotopy equivalence is a homotopy equivalence.

Solution. (i)

Exercise 0.0.4. A deformation retraction in the weak sense of a space X to a subspace A is a homotopy ft : X → X
such that f0 = idX , f1 (X) ⊆ A, and ft (A) ⊆ A for all t . Show that if X deformation retracts to A in this weak sense,
then the inclusion A ,→ X is a homotopy equivalence.

Solution.

Exercise 0.0.5. Show that if a space X deformation retracts to a point x ∈ X , then for each neighborhood U of x in
X there exists a neighborhood V ⊆ U of x such that the inclusion map V ,→ U is nullhomotopic.

5
6 CHAPTER 0. SOME UNDERLYING GEOMETRIC NOTIONS

Solution.

Exercise 0.0.6 (Exercise 0.10). Show that a space X is contractible iff every map f : X → Y , for arbitrary Y , is
nullhomotopic. Similarly, show X is contractible iff every map f : Y → X is nullhomotopic.

Solution. (i) Suppose X is contractible, then there is a point x0 , and maps h : X → {x0 }, g : {x0 } → X s.t.
g ◦ h ' id|X and h ◦ g ' id|{x0 } . We denote the homotopy as F : X × I → X where F |X×{0} = id and
F |X×{1} = g ◦ h. For any f : X → Y where Y is an arbitrary space, let y0 = f (g(x0 )), and let G := f ◦ F .
Thus G : X × I → Y is continuous since it is the composition of two continuous maps. G|X×{0} = f ◦ id = f
and G|X×{1} = f ◦ g ◦ h. But f ◦ g ◦ h(X) = y0 . Therefore f : X → Y is nullhomotopic.
Conversely, put Y = X, then we know that id : X → X is nullhomotopic. That is, we have a constant map
g : X → X and a homotopy F : X × I → X s.t. F |X×{0} = id and F |X×{1} = g. g being a constant map
means g(X) = {x0 } for some x0 ∈ X, so we say g is a map X → {x0 } and define f : {x0 } → X, x0 7→ x0 .
Thus g ◦ f = id{x0 } and f ◦ g = g. The existence of F implies f ◦ g ' id.
(ii) Suppose X is contractible, then there is a point x0 , and maps h : X → {x0 }, g : {x0 } → X s.t.
g ◦ h ' id|X and h ◦ g ' id|{x0 } . We denote the homotopy as F : X × I → X where F |X×{0} = id and
F |X×{1} = g ◦ h. Define G : Y × I → X, (y, t) 7→ F (f (y), t). Hence G|Y ×{0} = F (f (y), 0) = f (y) and
G|X×{1} = F (f (y), 1) = h(g(f (y))) = h(x0 ). Thus, f : X → Y is nullhomotopic.
Conversely, put Y = X, then we know that id : X → X is nullhomotopic. That is, we have a constant map
g : X → X and a homotopy F : X × I → X s.t. F |X×{0} = id and F |X×{1} = g. g being a constant map
means g(X) = {x0 } for some x0 ∈ X, so we say g is a map X → {x0 } and define f : {x0 } → X, x0 7→ x0 .
Thus g ◦ f = id{x0 } and f ◦ g = g. The existence of F implies f ◦ g ' id.

Exercise 0.0.7. Given positive integers v, e, and f satisfying v − e + f = 2, construct a cell structure on S 2 having v
0-cells, e 1-cells, and f 2-cells.

Solution. We do induction on v. Notice that v is at least 1, and f is at least 1 because S 2 is of dimension 2.

For v = 1,

Exercise 0.0.8 (Exercise 0.20). Show that the subspace X ⊆ R3 formed by a Klein bottle intersecting itself in a circle,
is homotopy equivalent to S 1 ∨ S 1 ∨ S 2 .

Exercise 0.0.9 (Exercise 0.23). Show that a CW complex X is contractible if it is the union of two contractible
subcomplexes whose intersection is also contractible.

Solution. Suppose X = A ∪ B and suppose A ∩ B is contractible. Hence by the first homotopy equivalence
criterion, {∗} ' B ' B/A ∩ B. The map ϕ̄ : X → B/A ∩ B induces a natural map

ϕ : X/A → B/A ∩ B,

where ϕ̄ maps every point x ∈ X − A to x itself in B/A ∩ B, and sends A to A ∩ B/A ∩ B, i.e. we have the
following

X
ϕ̄
π
ϕ
X/A B/A ∩ B.

By the definition of quotient topology, ϕ is continuous. In fact, if U is an open set in B/A ∩ B, then ϕ−1 (U ) =
π1 ◦ ϕ̄−1 ◦ π2−1 (U ) is also open. Similarly, the map ψ̄ : B → X/A induces a natural map

ψ : B/A ∩ B → X/A,

where ψ̄ maps every point x ∈ B − A to x itself in X/A, and sends A ∩ B to A/A. Also we have

B
ψ̄
π
ψ
B/A ∩ B X/A.
 7

The same argument shows ψ is continuous. Since ϕ ◦ ψ = id and ψ ◦ ϕ = id, we have a homeomorphism

X/A ∼
= B/A ∩ B.

Again by the first homotopy equivalence criterion, X ' X/A since A is contractible. Hence X ' X/A ∼
=
B/A ∩ B ' A ' {∗}.

Exercise 0.0.10 (Exercise 0.28). Show that if (X1 , A) satisfies the homotopy extension property, then so does every
pair (X0 tf X1 , X0 ) obtained by attaching X1 to a space X0 via map f : A → X0 .

Remark. This is a more general result: cofibrations are stable under taking pushouts.
Solution. Since (X1 , A) satisfies the homotopy extension property, we have a retraction r : X1 × I → X1 ×
{0} ∪ A × I. By the quotient map, we have
idX0 ×r
(X0 t X1 ) × I (X0 × I) t (X1 × {0} ∪ A × I)
f¯ f˜

(X0 tf X1 ) × I (X0 × I) tf (X1 × {0})

where f¯ attaches X1 to X0 and leaves I stable, and similarly f˜ attaches X0 × {0} and A × I to X0 × I. We say
that guarantees a map R : (X0 tf X1 ) × I → (X0 × I) tf (X1 × {0}) which makes the diagram commute. R
can be defined as following: for any (x, t) ∈ (X0 tf X1 ) × I, there is a (x̄, t) s.t. f¯(x̄, t) = (x, t). Hence define
R(x, t) = f˜ ◦ (idX0 × r)(x̄, t).
The map is well-defined. Actually, if (x0 , t), (x1 , t) are two different points that f¯(xi , t) = (x, t), then
x ∈ f (A) ⊆ X0 . W.l.o.g., assume xi ∈ Xi . It suffices to prove f˜ ◦ (idX0 × r)(x0 , t) = f˜ ◦ (idX0 × r)(x1 , t).
But f˜ ◦ (idX0 × r) and f¯ leave X0 × I stable, so x0 = x and f˜ ◦ (idX0 × r)(x0 , t) = (x0 , t) = (x, t). On the
other hand, f˜ ◦ (idX0 × r)(x1 , t) = (x1 , t) ∈ A × I. Thus f (x1 ) = x implies f˜(x1 , t) = (x, t).
Finally, R is a retract. It is continuous since the definition of quotient topology. We have seen it remains stable
on X0 ×I. For any (x1 , 0) ∈ (X1 −f (A))×{0} ⊆ (X0 tf X1 )×I, we still have only one (x¯1 , 0) ∈ (X1 −A)×I
s.t. f¯(x¯1 , t) = (x1 , t). Actually it is itself. But r|X1 ×{0} = id, hence f˜◦ (idX0 × r)(x¯1 , 0) = (x1 , 0) ∈ X1 × {0}.
In conclusion, (X0 ×I)tf (X1 ×{0}) is a retract of (X0 tf X1 )×I so (X0 tf X1 , X0 ) satisfies the homotopy
extension property.
8 CHAPTER 0. SOME UNDERLYING GEOMETRIC NOTIONS
Chapter 1

The Fundamental Group

1.1 Basic Constructions

Exercise 1.1.1 (Exercise 1.1.7). Define f : S 1 × I → S 1 × I by f (θ, s) = (θ + 2πs, s), so f restricts to the identity
on the two boundary circles of S 1 × I. Show that f is homotopic to the identity by a homotopy ft that is stationary on
one of the boundary circles, but not by any homotopy ft that is stationary on both boundary circles. [Consider what f
does to the path s 7→ (θ0 , s) for fixed θ0 ∈ S 1 .]

Solution. Let

F :S 1 × I × I → S 1 × I
(θ, s, t) 7→ (θ + 2πst, s).

Hence F |S 1 ×I×{0} = id and F |S 1 ×I×{1} = f , and F is continuous since every component of F is continuous,
therefore F is the homotopy from id to f . Again by definition, F |S 1 ×{0}×I = id so F is stationary on one of the
boundary circle.
Suppose we have a homotopy G : S 1 × I × I → S 1 × I giving ft ' id, s.t. G is stationary on both of
the boundary circles. Define a family of paths γθ0 : I → S 1 × I by s 7→ (θ0 , s). Explicitly the conditions are
G(γθ (s), 0) = γθ (s), G(γθ (s), 1) = f ◦ γθ (s), and G(γθ (0), t) = (θ, 0), G(γθ (1), t) = (θ, 1). Hence we have
a homotopy from path γθ0 to f ◦ γθ0 . Then we consider the projection π : S 1 × I → S 1 × {0}, (θ, s) 7→ (θ, 0).
Then π ◦ γθ and π ◦ f ◦ γθ are homotopy equivalent loops of S 1 . However, π ◦ γθ is a point so [π ◦ γθ ] = 0 but
[π ◦ f ◦ γθ ] = 1 since the projection is surjective, which leads a contradiction.

Exercise 1.1.2 (Exercise 1.1.12). Show that every homomorphism π1 (S 1 ) → π1 (S 1 ) can be realized as the induced
homomorphism ϕ∗ of a map ϕ : S 1 → S 1 .

Solution. Since π1 (S 1 ) ∼
= Z, every homomorphism ϕ∗ : π1 (S 1 ) → π1 (S 1 ) is uniquely determined by ϕ∗ (1)
since ϕ∗ (n) = nϕ∗ (1). Let k = ϕ∗ (1), then construct

ϕ :S 1 → S 1
z = (cos θ, sin θ) 7→ e2πikz = (cos kθ, sin kθ)

then is suffices to prove ϕ∗ is the homomorphism induced by ϕ. Denote the generator of π1 (S 1 ) as [ω(t)] where
ω : I → S 1 , t 7→ (cos t, sin t). Hence

ϕ∗ ([ω]) = [ϕ ◦ ω] = [(cos kθ, sin kθ)] = [ω]k .

Therefore ϕ∗ is the induced homomorphism.

Exercise 1.1.3 (Exercise 1.1.16). Given a map f : X → Y and a path h : I → X from x0 to x1 , show that
f∗ βh = βf h f∗ in the diagram

9
10 CHAPTER 1. THE FUNDAMENTAL GROUP

βh
π1 (X, x1 ) π1 (X, x0 )
f∗ f∗
βf h
π1 (Y, f (x1 )) π1 (Y, f (x0 )).

Solution. Suppose [ω] is an element in π1 (X, x1 ), then

βf h ◦ f∗ ([ω]) = βf h ([f ◦ ω]) = [(f h) ◦ f ◦ ω ◦ f¯h] = f∗ [h ◦ ω ◦ h̄] = f∗ βh ([ω]).

Hence the diagram commutes.

Exercise 1.1.4 (Exercise 1.1.16). Show that there are no retractions r : X → A in the following cases:

1. X = R3 with A any subspace homoeomorphic to S 1 .

2. X = S 1 × D2 with A its boundary torus S 1 × S 1 .

3. X = S 1 × D2 and A the circle shown in the book.

4. X = D2 ∨ D2 with A its boundary S 1 ∨ S 1 .

5. X a disk with two points on its boundary identified and A its boundary S 1 ∨ S 1 .

6. X the Möbius band and A its boundary circle.

Solution. Suppose we have a retraction r : X → A, then Proposition 1.17 tells us that the inclusion i : A ,→ X
induces an injection i∗ : π1 (A, x0 ) ,→ π1 (X, x0 ).
(i) Since X is contractible (F : X × I → X, (x, t) 7→ tx), π1 (X, x0 ) is trivial. But π1 (A, x0 ) is Z so we
cannot have an injection.
(ii) D2 is contractible, so π1 (X, x0 ) = π1 (S 1 × D2 , x0 ) = π1 (S 1 , x0 ) × π1 (D2 , x0 ) = Z. But π1 (A, x0 ) =
π1 (S 1 × S 1 , x0 ) = Z × Z, so there is no injection.
(iii) We know that π1 (A, x0 ) ∼ = Z. Suppose ω : I → A is a parametrization where ω(0) = ω(1), so [ω] is one
of the generators. But ω is contractible in X since A = ∂D1 for some D1 ⊆ X. So [ω] is 0 in π1 (X, x0 ). Hence
the map i∗ is never injective.
(iv) Suppose the base point is the intersection point of X = D2 ∨ D2 . Since each D2 is contractible,
π1 (X, x0 ) = π1 (D2 ∨ D2 , x0 ) = 0, but π1 (A, x0 ) = π1 (S 1 ∨ S 1 , x0 )(= Z ∗ Z). Instead of using van Kam-
pen’s theorem to prove it, it suffices to find an element which is not contractible. Suppose [ω] is an element in
π1 (S 1 ∨ S 1 , x0 ) where ω : I → S 1 , t 7→ e2πit . Since π1 (S 1 , x0 ) = Z, [ω] cannot be 0. Hence there cannot be an
injection.
(v) First X is homomorphic to a circle through the point on the boundary identified. We can do this by shrink
the surface from the two sides of the boundary to the circle we find. Thus π1 (X, x0 ) = π1 (S 1 , x0 ) = Z. Suppose
α, β are two loops s.t. [α], [β] are two generators of the two different circles in S 1 ∨ S 1 . The inclusion gives that
[α], [β] are homotopic since they are homotopically equivalent to the circle we find in X. But [α] 6= [β], so it is
not an injection.
(vi) It is clear that π1 (A, x0 ) is Z. Consider the retraction projecting the vertical lines onto the horizontal
brown line (see the figure below),

a a

b
1.2. VAN KAMPEN’S THEOREM 11

this gives a homotopy h : X ' S 1 and thus π1 (X) ∼ = Z where a generator is the circle. In the figure above, the
loop b is the subspace A, and suppose ω : I → A is the loop b so that [ω] is a generator of π1 (A). However h ◦ ω
gives the loop winding twice on S 1 , so via the homotopy along the stripe the image of the generator of π1 (A) is
[ω]2 , and hence i∗ : π1 (A) → π1 (X) maps 1 to 2.
Suppose one has a retraction r : X → A, then the composition

r∗ ◦ i∗ : π1 (A) → π1 (A)

is identity by the functoriality of π1 . However we’ve computed that i∗ is not surjective so there is not a retraction.

Exercise 1.1.5 (Exercise 1.1.18). Using the technique in the proof of Proposition 1.14, show that if a space X is
obtained from a path-connected subspace A by attaching a cell en with n ≥ 2, then the inclusion A ,→ X induces a
surjection on π1 . Apply this to show:
1. The wedge sum S 1 ∨ S 2 has fundamental group Z;
2. For a path-connected CW complex X the inclusion map X 1 ,→ X of its 1-skeleton induces a surjection
π1 (X 1 ) → π1 (X).
Solution. Suppose ω : I → X is S∞ a loop with ω(0) = ω(1) = x0 ∈ A. Since en is open in X, hence ω −1 (en ) is
−1 n
open in I. Assume ω (e ) = i=1 (ci , di ), where 0 < ci ≤ di < 1. (When ci = di we mean (ci , di ) is empty.)
Thus f (ci ), f (di ) ∈ A. Since A is path-connected, we have a path fi : [ci , di ] → A. Denote gi = ω|[ci ,di ] ,
then Fi : fi ' gi since π1 (Dn , x0 ) is trivial. Thus, combine all these homotopies, we have a homotopy F s.t.
F |(ci ) = Fi and F = id otherwise. Hence F is a homotopy from ω to a path in A.
(i) S 1 ∨ S 2 = S 1 ∪f e2 where f : ∂D2 → {x0 } ⊆ S 1 . Hence the inclusion A ,→ S 1 ∨ S 2 induces a surjection
Z → π1 (S 1 ∨ S 2 ). But Z → π1 (S 1 ∨ S 2 ) cannot be trivial since the circle S 1 is not nullhomotopic. Hence
π1 (S 1 ∨ S 2 ) = Z.
(ii) Suppose ω is a path in X. Then ω(I) is compact in X and hence by proposition A.1 we know it is included
in a finite CW complex, say X n . Therefore by previous proof ω is homotopic to a path in X 1 . Hence we have a
surjective map π1 (A, x0 ) → π1 (X, x0 ).

1.2 Van Kampen’s Theorem

Exercise 1.2.1 (Exercise 1.2.4). Let X ⊆ R3 be the union of n lines through the origin. Compute π1 (R3 − X).
Solution. Consider the standard sphere S 2 ⊆ R3 , then there is a radial deformation retraction R3 − X ' S 2 ∩
(R3 − X). Hence π1 (R3 − X) = π1 (S 2 ∩ (R3 − X)) where S 2 ∩ (R3 − X) is the space punctured sphere with
n holes. The punctured sphere with 1 hole is homeomorphic to Int D2 , hence S 2 ∩ (R3 − X) is homeomorphic
to punctured disk with 2n − 1 holes. Thus

π1 (S 2 ∩ (R3 − X)) ∼
= Fn−1

the free group generated by 2n − 1 elements.

Exercise 1.2.2 (Exercise 1.2.7). Let X be the quotient space of S 2 obtained by identifying the north and south poles
to a single point. Put a cell complex structure on X and use this to compute π1 (X).
Solution 1. Consider the CW structure on X given by.
Solution 2. It is easy to see X is homotopic to the space Y consisting of a sphere and a diameter connecting the
north pole and the south pole, since we have a deformation retract on the diameter into a point. Y is homotopic to
S 1 ∨ S 2 by Example 1.23 in Hatcher. Hence

π1 (X) ∼
= π1 (Y ) ∼
= π1 (S 1 ∨ S 2 ) = Z

by van Kampen’s theroem.

12 CHAPTER 1. THE FUNDAMENTAL GROUP

Exercise 1.2.3 (Exercise 1.2.9). In the surface Mg of genus g, let C be a circle that separates Mg into two compact
subsurfaces Mh0 and Mk0 obtained from the closed surfaces Mh and Mk by deleting an open disk from each. Show that
Mh0 does not retract onto its boundary circle C, and hence Mg does not retract onto C. But show that Mg does retract
onto the nonseparating circle C 0 in the figure.

Solution. First we prove that Mh0 does not retract onto its boundary circle C. By the argument in Chapter 0, the
CW complex of Mh consists of 1 1-cells, 2g 1-cells and a 2-cell. Mh0 is homeomorphic to cutting a hole inside
the 2-cell. Hence Mh0 is homotopic to S 1 ∨ · · · ∨ S 1 of 2g copies. Thus Mh0 = ha1 i ∗ · · · ∗ ha2g i.
Then we suppose we have an retraction r : Mh0 → C, then r induces an injection i∗ : π1 (C) → π1 (Mh0 ),
where i is the inclusion C ,→ Mh0 . Thus we have an injection of abelianization (i∗ )0 : π1 (C)/π1 (C)0 →
π1 (Mh0 )/π1 (Mh0 )0 . But π1 (C) = Z hence its abelianization is itself. But the loop C maps to a1 a2 a−1 −1 −1 −1
1 a2 · · · a2g−1 a2g a2g−1 a2g ,
0 0
whose image in π1 (Mh ) happens to be a commutator. Hence (i∗ ) cannot be an injection, a contradiction.
To see Mg retracts on to C 0 , we use the planar presentation of Mg , generated by 4g 1-cells (labelled by
a1 , b1 , · · · , ag , bg ) and 1 2-cell. Take the quotient of ai → a1 , bi → b1 , we get a map Mg → M1 . For M1 , the
map mapping all horizontal line in the figure

v b v

a e a

v b v.

into a point gives the retract to the circle represented by a, and the composition is the desired retraction.

Exercise 1.2.4 (Exercise 1.2.10). Consider two arcs α and β embedded in D2 × I as shown in the picture. The loop γ
is obviously nullhomotopic in D2 × I, but show that there is no nullhomotopy of γ in the complement of α ∪ β.

Solution. First we notice that the complement is homeomorphic to I 3 minus two straight lines which do not intersect,
since we can change the cylinder making the arcs straight. This can be deformation retracted to I × I minus two points,
which is homotopic to S 1 ∨ S 1 . Hence the fundamental group π1 (X) is Z ∗ Z = hai ∗ hbi.
Then we consider how did the loop γ change. In the original figure, the loop γ splits the intersections of the straight
lines in I 3 , after the homeomorphism and homotopy, [γ] = abab in π1 (X), hence it is not homotopic to 0.

Exercise 1.2.5 (Exercise 1.2.14). Consider the quotient space of a cube I 3 obtained by identifying each square face
with the opposite face via the right-handed screw motion consisting of a translation by one unit in the direction per-
pendicular to the face combined with a one-quarter twist of the face about its center point. Show this quotient space X
is a cell complex with two 0-cells, four 1-cells, three 2-cells, and one 3-cell. Using this structure, show that π1 (X) is
the quaterion group of order 8.

Solution. I 3 can be naturally seen a CW complex with 8 1-cells, 6 2-cells and 1 3-cell. We first consider the 1-skeleton
of X
· a *
db
b
* c ·
d

c
* d ·
a
ac
· b
*
1.2. VAN KAMPEN’S THEOREM 13

Thus the 1-skeleton is a graph consisting two points ∗, · and four edges a, b, c, d from · to ∗ respectively. The quotient
space has 1-skeleton generated by ab−1 , ac−1 , ad−1 , and the 2-cells are ac−1 db−1 , ad−1 bc−1 , bc−1 ad−1 . Denote
i = ab−1 , j = ac−1 , k = ad−1 , then we have a group presentation of π1 (X)
hi, j, k | jki−1 , kij −1 , ijk −1 i
by Prop 1.26. This is exactly a presentation of quaterion group of order 8.

Exercise 1.2.6 (Exercise 1.2.15). Given a space X with basepoint x0 ∈ X, we may construct a CW complex L(X)
having a single 0-cell, a 1-cell e1γ for each loop γ in X based at x0 , and a 2-cell for each map τ of a standard triangle
P QR into X taking the three vertices P, Q, R of the triangle to x0 . The 2-cell e2τ is attachd to the three 1-cells that
are the loops obtained by reconstructing τ to the three oriented edges P Q, P R and QR. Show that the natural map
L(X) → X induces an isomorphism π1 (L(X)) ' π(X, x0 ).
Solution. Since for every loop γ in X based at x0 , we have a 1-cell e1γ , the map L(X) → X induces a surjective map
π1 (L(X))  π(X, x0 ). Then it suffices to prove that the kernel is 0. Suppose ω is a nullhomotopic loop in X, then we
can find an open subset U of X s.t. ω([0, 1]) ∪ U ∼
= ω([0, 1]) ∪f e2 where f maps the boundary of the disk to the image
of ω by f . ω([0, 1]) ∪ U is homeomorphic to a triangle denoted by τ , hence we have an element [P Q ◦ QR ◦ RP ] in
π1 (L(X)) s.t. [ω] is the image of [P Q ◦ QR ◦ RP ]. But we have a 2-cell e2τ attached on it so [P Q ◦ QR ◦ RP ] = 0.
Therefore the induced homomorphism is an isomorphism.

Exercise 1.2.7 (Exercise 1.2.16). Show that the fundamental group of the surface of infinite genus shown below is free
on an infinite number of generators.
Solution. Let Y be a torus with two disjoint open disks removed. Since Y has a deformation retract on a circle with
a diameter, we know π1 (Y ) ∼
= π1 (S 1 ∨ S 1 ) ∼
= Z ∗ Z, and each boundary circle of the removed disk is a generator in
Z ,→ Z ∗ Z. The surface of infinite genus consists of infinitely many copies of Y , attached on the boundary circles. By
van Kampen’s theorem, after gluing two copies of Y together, the fundamental group becomes Z ∗ Z ∗ Z. Keep doing
the gluing, we know the fundamental group is free on an infinite number of generators.
Exercise 1.2.8 (Exercise 1.2.22). In this exercise we describe an algorithm for computing a presentation of the funda-
mental group of the complement of a smooth or piecewise linear knot K in R3 , called the Wirtinger presentation. To
begin, we position the knot to lie almost flat on a table, so that K consists of finitely many disjoint arcs αi where it
intersects the table top together with finitely many disjoint arcs βl where K crosses over itself. The configuration at
such a crossing is shown in the first figure below. We build a 2 dimensional complex X that is a deformation retract
of R3 − K by the following three steps. First, start with the rectangle T formed by the table top. Next, just above
each arc αi place a long, thin rectangular strip Ri , curved to run parallel to αi along the full length of αi and arched
so that the two long edges of Ri are identified with points of T , as in the second figure. Any arcs βl that cross over
αi are positioned to lie in Ri . Finally, over each arc βl put a square Sl , bent downward along its four edges so that
these edges are identified with points of three strips Ri , Rj , and Rk as in the third figure; namely, two opposite edges
of Sl are identified with short edges of Rj and Rk and the other two opposite edges of Sl are identified with two arcs
crossing the interior of Ri . The knot K is now a subspace of X, but after we lift K up slightly into the complement of
X, it becomes evident that X is a deformation retract of R3 − K.
1. Assuming this bit of geometry, show that π1 (R3 − K) has a presentation with one generator xi for each strip Ri
and one relation of the form xi xj x−1i = xk for each square Sl , where the indices are as in the figures above. [To
get the correct signs it is helpful to use an orientation of K.]
2. Use this presentation to show that the abelianization of π1 (R3 − K) is Z.
Solution. 1. We shall take the assumption that X is homotopic to R3 − K, so computing π1 (R3 − K) is actually
computing π1 (X).
First we compute the fundamental group of T ∪ R1 ∪ · · · ∪ Ri ∪ · · · . Each Ri gives a nontrivial loop, so by
Van Kampen’s theorem π1 (T ∪ R1 ∪ · · · ∪ Ri ∪ · · · ) = Fn where Fn is the free group generated by n elements,
and n is the number of such Ri ’s. Next, gluing Sl makes the loop xi xj x−1 −1
i xk trivial by Van Kampen’s theorem,
so the fundamental group is as described.
2. After taking abelianisation, the braid relation xi xj x−1
i = xk becomes xj = xk , hence the abelianisation is
Z.
14 CHAPTER 1. THE FUNDAMENTAL GROUP

1.3 Covering Spaces

Exercise 1.3.1.
Solution.

Exercise 1.3.2.
Solution.

Exercise 1.3.3.
Solution.

Exercise 1.3.4 (Exercise 1.3.4). Construct a simply-connected covering space of the space X ⊆ R3 that is the union
of a sphere and a diameter. Do the same when X is the union of a sphere and a circle intersecting it in two points.
Solution. Let Z be countably infinitely many disjoint union of closed unit ball in R3 centered at z-axis, and
connect their north pole N to the south pole S of the ball right above them. Let Y = ∂Z as figure 1.1. Then let
p : Y → X, where p send the spheres to sphere, send the connecting segments to the diameter inside the sphere,
and send north/south pole to the points where the diameter intersecting the sphere. Clearly Y is the covering
space.

Figure 1.1: The Covering Space of X

For another case, we still have the same space but we have different map p0 : Y → X 0 . First label the
connecting segments a, b consecutively, then p0 send the spheres to sphere, send the connecting segments labeled
as a to the half-circle inside the sphere and the connecting segments labeled as b to the half-circle outside the
sphere, and send north/south pole to the points where the circle intersecting the sphere.

Exercise 1.3.5 (Exercise 1.3.5). Let X be the subspace of R2 consisting of the four sides of the square [0, 1] × [0, 1]
together with the segments of the vertical lines x = 12 , 31 , 14 , · · · inside the square. Show that for every covering space
X̃ → X there is some neighborhood of the left edge of X that lifts homeomorphically to X̃. Deduce that X has no
simply-connected covering space.
1
Solution. Denote Xm be the subspace of R2 consisting of the four sides of the square [0, m ] × [0, 1] together with the
1 1 1
segments of the vertical lines x = m+1 , m+2 , m+3 ,
cdots inside the square. We first prove that for any integer m there is a homeomorphism X ∼= Xm . Consider
ϕ : X → Xm ,
1
x− k+1
where ϕ maps a point ( k1 , t) to ( k+m
1 1
, t) and ϕ maps a point (x, 0) to ( k+m 1
+ ( k+m − 1
k+m+1 ) k
1 1
− k+1
, 0) where
1 1
k+1 ≤ x ≤ for some integer k. Clearly ϕ and its inverse are piecewisely continuous on the intervals homeomorphic
k
to I, hence it is a homeomorphism.
Suppose Y ⊂ X is an open neighborhood of the left edge, then we have an open set U of R2 s.t. Y = U ∩ X.
For each point x in the left side, there is a ball Bx (δx ) centered at x with radius δx s.t. Bx (δx ) ⊆ U , therefore
we form an open covering of I. Since I is compact, there is a finite open covering Bx1 (δx1 ), · · · , Bxn (δxn ). Let
1
m < min{δx1 , · · · , δxn }, then we know Xm ⊆ U , hence Xm ⊆ Y .
Suppose p : X̃ → X is a covering space of X, then p−1 (Y ) has a connected piece homeomorphic to Y where
Y ⊂ X is an open neighborhood of the left edge. Thus we have a subspace homeomorphic to Xm ∼ = X. Therefore
we have a nontrivial element [ω] ∈ π1 (X̃) where ω : I → I ∪ [ 12 , 1] ∪ I ∪ [ 12 , 1]. Otherwise p∗ ([ω]) must be trivial in
π1 (X), which is a contradiction.
1.3. COVERING SPACES 15

Exercise 1.3.6.
Solution.

Exercise 1.3.7 (Exercise 1.3.7). Let Y be the quasi-circle, a closed subset of R2 consisting of a portion of the graph of
y = sin x1 , the segment [−1, 1] in the y-axis, and an arc connecting these two pieces. Collapsing the segment of Y in
the y-axis to a point gives a quotient map f : Y → S 1 . Show that f does not lift to the covering space R → S 1 even
though π1 (Y ) = 0. Thus local path-connectedness is a necessary hypothesis in the lifting criterion.
Solution. Without loss of generality, assume the quotient map f sends the segment [−1, 1] to the base point of
S 1 . Suppose there is a lift f˜
R
f˜ p
q

Y f
S1,

then f˜ has to send the segment to (one point of) Z. By the universal property of the quotient space, there is a map
q : S 1 → R s.t. f˜ = q ◦ f , thus f = p ◦ f˜ = p ◦ q ◦ f . Still by the universal property of quotient space (there
exists only one map), p ◦ q = id, a contradiction.

Exercise 1.3.8 (Exercise 1.3.8). Let X̃ and Ỹ be simply-connected covering spaces of the path-connected, locally
path-connected spaces X and Y . Show that if X ' Y , then X̃ ' Ỹ .
Solution.

Exercise 1.3.9 (Exercise 1.3.9). Show that if a path-connected, locally path-connected spaces X has π1 (X) finite, then
every map X → S 1 is nullhomotopic.
Solution. First we prove that if f : X → S 1 is a map then the induced map f∗ : π1 (X) → π1 (S 1 ) is the zero
map. Suppose n = |π1 (X)| > 0 and suppose that [ω] ∈ π1 (X) is an element, then [ω]n = 1 ∈ π1 (X). But
f∗ ([ω]n ) = n · f∗ ([ω]) = 0 ∈ π1 (S 1 ), and the only element k in π1 (S 1 ) = Z s.t. n · k = 0 is 0, so f∗ ([ω]) = 0,
i.e. f∗ : π1 (X) → π1 (S 1 ) is the zero map.
By the lifting criterion, we have a lift f˜ : X → R
R
f˜ p
f
X S1.

Since R is contractible, by Exercise 0.0.6, f˜ is contractible. Thus we have a F : X × I → R s.t. F |X×{0} = f

and F |X×{1} = constant. Therefore, G := p ◦ F is the homotopy from f˜ to a constant map.

Exercise 1.3.10 (Exercise 1.3.10). Find all the connected 2-sheeted and 3-sheeted covering space of S 1 ∨ S 1 , up to
isomorphism of covering spaces without basepoints.
Solution. By the textbook, the covering space X̃ of X = S 1 ∨ S 1 is a 2-oriented graph. While the map p : X̃ → X
maps vertices to the vertex in X, hence there are n vertices if and only if the covering space is n-sheeted.
For n = 2, there are two possibilities, where in one case there is an edge that connects a vertex to itself, while in
the other there is not. But the graph is connected, so there must be two edges connecting the two vertices. Hence there
are only two different covering space, shown in the textbook Page 58, (1) and (2). And because of the symmetry of the
two vertex, the orientation does not matter.
For n = 3, there are two possibilities, where in one case there is an edge that connects a vertex to itself, while in
the other there is not. But the graph is connected, so there must be two edges connecting some the two vertices. Thus
there must be a vertex without any single edge connecting to itself. So there are also two cases, where in one case
there is an edge that connects a vertex to itself, while in the other there is not (corresponding to the (3),(4) and (5),(6)
in the table on Page 58 respectively). There is also the consideration of orientation of the edges. Since for each vertex,
there are exactly two paths labeled as a (or b), and exactly one coming into the vertex. So for each case, there are two
nonisomorphic covering space, which are (3),(4) and (5),(6) in the table on Page 58.
16 CHAPTER 1. THE FUNDAMENTAL GROUP

Exercise 1.3.11.
Solution.

Exercise 1.3.12.
Solution.

Exercise 1.3.13.
Solution.

Exercise 1.3.14.
Solution.

Exercise 1.3.15.
Solution.

Exercise 1.3.16 (Exercise 1.3.16). Given maps X → Y → Z such that both Y → Z and the composition X → Z are
covering spaces, show that X → Y is a covering space if Z is locally path-connected. Show that this covering space is
normal if X → Z is normal.
Solution. Suppose the given maps are p : X → Y and q : Y → Z. Since q and q ◦ p are covering spaces, by
definition for each y ∈ Y , there is an open set U ⊆ Z containing q(x) s.t. both q −1 (U ) and (q ◦ p)−1 (U ) =
p−1 (q −1 (U )) are disjoint union of spaces that are homeomorphic to U via q and q ◦ p respectively. Since Z is
locally path-connected, one can take the neighborhood U small enough so that U is path-connected.
Since q is a covering space, there is an open subset V ⊆ q −1 (U ) s.t. q|V gives a homeomorphism between
spaces. Therefore one has a commutative diagram

p−1 (V ) X
p|p−1 (V ) p
fx |p−1 (V )

V Y
(q|V )−1
q|V q

U U i
Z.

For any x ∈ p−1 (V ) ⊆ X s.t. p(x) = y, there is a lift fx : U → X s.t.

i = q ◦ p ◦ fx : U → X → Y → Z
p q
where i : U ,→ X is the embedding, since X −
→Y −
→ Z is a covering space. Thus
p−1 (V ) ⊆ p−1 (q −1 (q(V ))) ⊆ p−1 (q −1 (U )),
`
which is a disjoint union of open sets, say j Wj . There is one such j = j0 s.t. x ∈ Wj0 . Because U is path-
connected (hence connected), there is a unique lift U → X to q ◦ p, which is fx as constructed. This means p|Wj0
is a homeomorphism. Furthermore, since x is arbitrary in the fibre of any point y ∈ Y , hence p−1 (V ) is a disjoint
union of open subsets in X homeomorphic to V via p.
For the second part, one has

X π1 (X)
p

Y π1 (Y )
q

Z π1 (Z),
1.3. COVERING SPACES 17

notice that since X → Z is normal, the normalisor of π1 (X) in π1 (Z) is the whole group, therefore the normalisor
of π1 (X) in π1 (Y ) is the intersection Nπ1 (X) (π1 (Z)) ∩ π1 (Y ), which is π1 (Y ). Hence X → Y is normal if X
is path-connected. Otherwise, notice that
1. Permutation of homeomorphic path-connected components of X is a deck transformation.
2. A deck transformation can only map a path-connected component to a component homeomorphic to it.
Therefore for any two points in the fibre of y ∈ Y , via a permutation of homeomorphic path-connected compo-
nents composed with a deck transformation within the path-connected component (shown before), one gets such
a deck transformation.

Exercise 1.3.17 (Exercise 1.3.17). Given a group G and a normal subgroup N , show that there exists a normal covering
space X̃ → X with π1 (X) = G, π1 (X̃) = N , and deck transformation group G(X̃) = G/N .
Solution. By Corollary 1.28, there is a 2-dimensional cell complex X s.t. π1 (X) = G. By Prop 1.36, we have a
covering space p : X̃ → X s.t. p∗ (π1 (X̃)) = N . Since N is a normal subgroup, by Prop 1.39, we have X̃ is a
normal covering space and the deck transformation group is G/N .
18 CHAPTER 1. THE FUNDAMENTAL GROUP

Exercise 1.3.18 (Exercise 1.3.18). For a path-connected, locally path-connected, and semilocally simply-connected
space X, call a path-connected covering space X̃ → X abelian if it is normal and has abelian deck transformation
group. Show that X has an abelian covering space that is a covering space of every other abelian covering space of X,
and that such a ‘universal’ abelian covering space is unique up to isomorphism. Describe this covering space explicitly
for X = S 1 ∨ S 1 and X = S 1 ∨ S 1 ∨ S 1 .

Solution. Since the space X is path-connected, locally path-connected, and semilocally simply-connected space,
there is a universal covering space X0 → X, whose deck transformation is exactly π1 (X).
Consider the normal group [π1 (X), π1 (X)], By Prop 1.36. there is a covering space p : X[π1 (X),π1 (X)] → X
s.t. p∗ (X[π1 (X),π1 (X)] ) = [π1 (X), π1 (X)]. By Prop 1.39., p : X[π1 (X),π1 (X)] → X is normal and the deck
transformation group is
π1 (X)/[π1 (X), π1 (X)] = π1 (X)ab .
To show that X̃ := X[π1 (X),π1 (X)] is universal among all abelian covering spaces, consider another abelian
covering space XK → X, with deck transformation group

π1 (X)/K = π1 (X)/p(π1 (XK ))

(where π1 (X)/K is only a notation of the group). Since π1 (X)/p(π1 (XK )) is abelian, there is a unique induced
group homomorphism g making the diagram

π1 (X) π1 (X)

g
π1 (X)ab π1 (X)/p(π1 (XK ))

commute, by the universal property of the abelianisation. The diagram implies that

p∗ (X[π1 (X),π1 (X)] ) = [π1 (X), π1 (X)] ⊆ π1 (XK ),

thus there is a “lift”

XK

X̃ X

by Prop 1.33. and the “lift” is unique by 1.34. since X is path-connected. The uniqueness comes from abstract
nonsense.
By discussion above, the universal abelian covering is the quotient space of the derived group acting on the
universal covering space. One knows that the universal covering space of S 1 ∨ S 1 (resp. S 1 ∨ S 1 ∨ S 1 ) is the
tree indexed by the free group F2 (resp. F3 ). Thus the universal abelian covering space is a quotient space of the
tree identifying the corresponding vertices (hence the edges) in the normal subgroup of the commutator subgroup.
These happen to give the lattice grids in R2 and R3 respectively.

Exercise 1.3.19 (Exercise 1.3.19). Use the preceding problem to show that a closed orientable surface Mg of genus g
has a connected normal covering space with deck transformation group isomorphic to Zn (the product of n copies of
Z) iff n ≤ 2g. For n = 3 and g ≥ 3, describe such a covering space explicitly as a subspace of R3 with translations
of R3 as deck transformations. Show that such a covering space in R3 exists iff there is an embedding of Mg in the 3
torus T 3 = S 1 × S 1 × S 1 such that the induced map π1 (Mg ) → π1 (T 3 ) is surjective.

Solution. By the discussion in Exercise 1.3.18, notice that

π1 (Mg ) = ha1 , b1 , · · · , ag , bg | [a1 , b1 ] · · · [ag , bg ] = 1i

and its abelianisation is Z2g . Thus, the universal abelian covering space X̃ → X has deck transformation group
Z2g . By the universal property, for any abelian covering X 0 → X, there is a normal covering X̃ → X 0 (by
Exercise 1.3.16)
1.3. COVERING SPACES 19

X̃ π1 (X̃) G(X̃/X̃) = {e}

X0 π1 (X 0 ) G(X̃/X 0 ) = π1 (X 0 )/π1 (X̃)

X π1 (X) G(X̃/X) = π1 (X)/π1 (X̃)

inducing the surjections among the (relative) deck transformation groups by Prop 1.39. Since G(X̃/X) =
π1 (X)/π1 (X̃) has 2g generators, the number of generators n in G(X̃/X 0 ) = π1 (X 0 )/π1 (X̃) is less than 2g.
Conversely, for n ≤ 2g, there is a surjection Z2g = (π1 (Mg ))ab → Z so one can construct an abelian covering
space with deck transformation group Zn as it is constructed in Exercise 1.3.18.
Before constructing the covering space, we first show that such a covering space in R3 with deck transforma-
tions being Z3 translations exists iff there is an embedding of Mg in the 3 torus T 3 = S 1 × S 1 × S 1 such that the
induced map π1 (Mg ) → π1 (T 3 ) is surjective. On the one hand, if there is such an embedding,

ĩ
M̃g R3
p q

i
Mg T3

then the composition M̃g → Mg → T 3 induces a composition of groups π1 (M̃g ) → π1 (Mg ) → π1 (T 3 ), where
the image of π1 (M̃g ) is the commutator subgroup, hence it maps to 0 since π1 (T 3 ) = Z3 is abelian. Therefore
there is a lift M̃g 99K R3 making the diagram commute. To show that this lift is injective, suppose there are
two points x̃0 , x̃1 are mapped to be the same point in R3 hence the same point in T 3 . Since Mg → T 3 is
injective, and the diagram is commutative, x̃0 , x̃1 are mapped to the same point in Mg , they are in the same fibre.
For any path γ : I → M̃g connecting these two points provided they are different, then by the unique lifting,
/ p∗ (π1 (M̃g )), implying [i ◦ p ◦ γ] ∈ π1 (T 3 ) is not trivial. However, [i ◦ p ◦ γ] = [q ◦ ĩ ◦ γ] is trivial since
[p ◦ γ] ∈
3
π1 (R ) being trivial, a contradiction. So ĩ is an embedding. Given any deck transformation σ, the composition

σ
→ M̃g ,→ R3
M̃g −

gives the same lift up to a choice of base point. Therefore

M̃g R3
σ

M̃g R3
p

Mg T3

is commutative. The way to find the deck transformations is to correspond a loop in π1 (Mg ) using the lift to give
the deck transformation. Since all covering spaces are normal, one can use the correspondence

Gal(M̃g /Mg ) ∼
= π1 (Mg )/p∗ (π1 (M̃g ))

to describe a deck transformation. For any deck transformation σ whose image in i∗ π1 (Mg ) ⊆ π1 (T 3 ) is not
zero, then its image in i∗ π1 (Mg ) ⊆ π1 (T 3 ) corresponds to a deck transformation σ̃ : R3 → R3 . However, all
diagrams given are commutative and the action of σ is constructed via the action on fibres, hence there is actually
a diagram
20 CHAPTER 1. THE FUNDAMENTAL GROUP

M̃g R3
σ σ̃

M̃g R3
p

Mg T 3.

This means all deck transformations of Gal(M̃g /Mg ) are transformations in Gal(R3 /Z3 ) = Z3 . Also because
π1 (Mg ) → π1 (T 3 ) is surjective, the transformations subgroup is the whole group.
On the other hand, suppose there is subspace

M̃g ,→ R3

with that all deck transformations of M̃g can be extended to be a unique action of Z3 on R3 . Then the embedding
induces a continuous map
Mg ∼ = M̃g /G = M̃g /Z3 → R3 /Z3 = T 3
by the group action. This map is injective because the map is quotienting out the same group action. Notice that
∼
π1 (Mg )/p∗ (π1 (M̃g )) ∼ → Gal(R3 /T 3 ) = π1 (T 3 )/p∗ (π1 (R3 )) = π1 (T 3 ),
= Gal(M̃g /Mg ) −

thus π1 (T 3 ) is a quotient group of π1 (Mg ) via the identification of deck transformations, hence via the inclusion
π1 (T 3 ) → π1 (Mg ) is surjective.
Finally we want to construct one embedding for g ≥ 3. For the case of g = 3, notice that Mg is the boundary
of a tubular neighbourhood of S 1 ∨ S 1 ∨ S 1 embedded in R3 . We have done the case of S 1 ∨ S 1 ∨ S 1 for a abelian
covering space embedded in R3 with three generators commuting. With “these” three generators commuting for
M3 , the abelian covering space should be the boundary of a tubular neighbourhood of the abelian covering space
of S 1 ∨ S 1 ∨ S 1 embedded in R3 , with the deck transformations being the translations by the lattice. For the case
g > 3, we should not have more deck transformations as described above, hence other “genus” should be trivial
under all the deck transformations, where the way to construct is to put g − 3 handles to each unit of the deck
transformation action.

Exercise 1.3.20 (Exercise 1.3.23). Show that if a group acts freely and properly discountinuously on a Hausdorff
space X, then the action is a covering space action. (Here ’properly discountinuously’ means that each x ∈ X has a
neighborhood U such that {g ∈ G | U ∩ gU 6= ∅} is finite). In particular, a free action on a finite group on a Hausdorff
space is a covering space action.

Solution. First we shall prove that for each x ∈ X, there is a neighborhood V s.t. V ∩ gV 6= ∅ implies that g = e.
By the proper-discountinuity, there are finitely many open neighborhood g1 , · · · , gn s.t. all elements g ∈ G − {e}
but them satisfy that
gU ∩ U = ∅.
Since X is Hausdorff, for each 1 ≤ i ≤ n, there are open sets Ui 3 x and Vi 3 gi · x s.t. Ui ∩ Vi = ∅. Take

V := U ∩ U1 ∩ · · · ∩ Un ∩ g1−1 V1 ∩ · · · ∩ gn−1 Vn ,

then V ⊆ Ui and gi V ⊆ Vi , hence V is the open neighborhood we want.

Then we shall prove that the action is a covering space action. For any two g1 , g2 s.t. g1 V ∩ g2 V 6= ∅, then
V ∩ g2−1 g1 V 6= ∅. Thus g2−1 g1 = e and g1 = g2 . By Prop 1.40, this is a covering space action.

Exercise 1.3.21 (Exercise 1.3.25). Let ϕ : R2 → R2 be the linear transformation ϕ(x, y) = (2x, y/2). This generates
an action of Z on X = R2 − {0}. Show this action is a covering space action and compute π1 (X/Z). Show the orbit
space X/Z is non-Hausdorff, and describe how it is a union of four subspaces homeomorphic to S 1 × R, coming from
the complementary components of the x-axis and the y-axis.
1.3. COVERING SPACES 21

Solution. First to show that this is a covering space action, which is to find an open neighbourhood U s.t. gU ∩
U 6= ∅ if and only if g = e. For any (x0 , y0 ) ∈ R2 − {(0, 0)}, W.L.O.G. one can assume x0 6= 0 then take
U := {(x, y) | x0 − |x40 | < x < x0 + |x40 | }, then
 
|x0 | |x0 |
< 2−n x < x0 + = (x, y) 2n x0 − 2n−2 |x0 | < x < 2n x0 + 2n−2 |x0 | ,

n · U = (x, y) x0 −
4 4
thus n · U ∩ U 6= ∅ iff n = 0.
There is an algebraic way of computing π1 (X/Z). By Prop 1.40, X is a covering space of X/Z with the deck
transformation group Z, such that
π1 (X/Z) ∼
= Z.
p∗ (π1 (X))
Therefore one has a S.E.S. of groups
π1 (X/Z) ∼
0 → π1 (X) ∼
= Z → π1 (X/Z) → =Z→0
p∗ (π1 (X))
where the first one is injective since it is induced by the covering map. Therefore π1 (X/Z) has two generators,
denoted by [ω], [γ] respectively, where [ω] comes from the image of π1 (X) under p∗ and [γ] is mapped to be the
generator of pπ∗1(π(X/Z)
1 (X))
. On the one hand, one can take ω : I → X, t 7→ exp 2πit then this gives the generator
[ω]. On the other hand, [γ] is a class in π1 (X/Z) but not a class in p∗ (π1 (X)), then if one takes a representative
γ : I → X/Z, it is a loop in X/Z. By the path lifting property (Prop 1.30), there is a (unique) lift γ̃ : I → X.
However, [γ] ∈ / p∗ (π1 (X)) means that the lift cannot be [ω n ] for some nonzero n, therefore γ̃ is a contractible
path in X. Hence there is a homotopy in X s.t.
γ̃ · ωγ̃ −1 ' ω,
therefore it induces a homotopy in X/Z s.t. [ω] commutes with [γ]. In conclusion, π1 (X/Z) ∼ = Z2 .
To show that the orbit space is not Hausdorff, it suffices to find a sequence of points with two different limit
points. Consider the sequences {(1, 2n )} ∼ {(2n , 1)}, where they are the same sequence under the group action.
However, the first sequence converges to (1, 0) as n → −∞ while the second one converges to (0, 1). Two limit
points are not in the same equivalence.
Finally consider the curve
{(x, y) | xy = c, x > 0}
for some constant c ∈ R+ , then under the group action this becomes an S 1 . Therefore the first quadrant is one
piece of S 1 × R, where the four cylinder are given by the quadrants, with axis giving four S 1 ’s.
Remark. There is a way of visualising the fundamental group of X/Z.
Take x0 = (1, 0) to be the base point of X and its image under the quotient to be the base point of X/Z. Thus both
of the paths below
y axis

x axis

are loops in X/Z. All these are homotopic to [γ] and the unit circle is ω.