Topology (H) Lecture 18

Lecturer: Zuoqin Wang

Time: May 18, 2020

THE FUNDAMENTAL GROUPS (II)

Last time we learned:

• path homotopy (homotopy fixing endpoints) the fundamental groupoid
• the fundamental group π1 (X, x0 ) = path homotopy classes of loops based at x0
– path λ from x0 to x1 group isomorphism Λ : π1 (X, x0 ) → π1 (X, x1 ).
– π1 (“star like domain”) = {e} and π1 (S n ) = {e} for n ≥ 2.
– π1 (X × Y ) = π1 (X) × π1 (Y ). [In PSet]
– f ∈ C(X, Y ) group homomorphism f∗ : π1 (X, x0 ) → π1 (Y, f (x0 )).
– Elements in π1 (X) “distinguished” non-contractible loops.
Relations in the group relations between the loops up to path homotopy.
• We have a sequence of functors πn a sequence topological invariants
πn (X, x0 ) = [(S n , p), (X, x0 )], n ≥ 0,
where π0 (X, x0 ) is a set, π1 (X, x0 ) is a group, and πn (X, x0 ) (n ≥ 2) are abelian
groups.
Today we continue to study the fundamental groups.

1. Homotopy invariance of π1 (X, x)

Recall that two topological spaces X, Y are homeomorphic, denoted by X ' Y , if
∃f ∈ C(X, Y ), g ∈ C(Y, X) s.t. f ◦ g = IdY and g ◦ f = IdX .
We have seen that if X ' Y, then
π1 (X, x) ' π1 (Y, f (x))
and in fact πn (X, x) ' πn (Y, f (x)) for all n.
On the other hand, we also have examples with X 6' Y but π1 (X) ' π1 (Y ), e.g.
π1 (Rk ) = π1 ({p}) = {e}, ∀k,
and in fact πn (Rk ) = πn ({p}) = {e} for all n and all k.
Note that “Homeomorphism” is a very strong equivalence relation: If X ' Y, then
X, Y have the same topological properties (compactness, connectedness, metrizability,
countability, separation properties etc.) Now we define a much weaker equivalence
relation between topological spaces.

1
2 THE FUNDAMENTAL GROUPS (II)

Definition 1.1. We say X and Y are homotopy equivalent, denoted by X ∼ Y , if

∃f ∈ C(X, Y ), g ∈ C(Y, X) s.t. f ◦ g ∼ IdY and g ◦ f ∼ IdX .
Such maps f and g are called homotopy equivalences between X and Y . [So f and g
are the “inverse” to each other at the homotopy level.]
Example.
(1) Obviously if X ' Y , then X ∼ Y . [And obviously the converse is not true.]
(2) Recall: a topological space X is contractible if IdX : X → X is null-homotopic.
It follows that for any map f : X → {pt}, g : {pt} → X, we have
f ◦ g = Id{pt} and g ◦ f = const ∼ IdX .
In other words, any contractible space X ∼ {pt}. [This includes Rk for all
k, and in fact all star-like (including all convex) domains in Rk , and the cone
C(X) = X × [0, 1]/X × {0} of any topological space X.]
Conversely, if X ∼ {pt}, then there exist
f : X → {pt} and g : {pt} → X
s.t. g ◦ f ∼ IdX . Since g ◦ f is a constant map, we conclude that IdX is
null-homotopic. In other words, we have
Proposition 1.2. X is contractible ⇐⇒ X ∼ {pt}.
(3) More examples of homotopy equivalence (but not homeomorphic):

(4) In PSet 6-3-3(b), we defined the conception of “deformation retract”: A ⊂ X is

a deformation retract if there is a retraction r : X → A ⊂ X such that r ∼ IdX .
Proposition 1.3. If A ⊂ X is a deformation retract, then A ∼ X.
Proof. Let ι : A → X be the inclusion map. Then r◦ι = IdA and ι◦r = r ∼ IdX . 
In particular, since S n is a deformation retract of Rn+1 \ {0}, we get
S n ∼ Rn+1 \ {0}.
 THE FUNDAMENTAL GROUPS (II) 3

Remarks.
(1) It is easy to check that the homotopy equivalence is an equivalence relation.
(2) Most topological properties (including compactness, (Ti), (Ai), metrizability
etc) are NOT preserved under homotopy equivalence.

However: it is easy to check that path-connectedness is preserved under homotopy

equivalence there is a good chance that the fundamental group (and homotopy
groups) will be preserved under homotopy equivalence. This is in fact the case:
Theorem 1.4. Let f : X ∼ Y be a homotopy equivalence, then
π1 (X, x0 ) ' π1 (Y, f (x0 )).

To prove the theorem, we first prove that homotopic maps induce “the same” group
homomorphisms. More precisely, suppose f1 , f2 ∈ C(X, Y ), and f1 ∼ f2 . Fix x0 ∈ X.
Let F : [0, 1] × X → Y be a homotopy between f1 and f2 . Then
λ(t) := F (t, x0 ), 06t61
is a path from y1 = f1 (x0 ) to y2 = f2 (x0 ). Let
Λ : π1 (Y, y1 ) → π1 (Y, y2 ), [γ]p 7→ [λ−1 ∗ γ ∗ λ]p
be the “base-point changing” group isomorphism induced by the path λ.
Lemma 1.5. (f2 )∗ = Λ ◦ (f1 )∗ (as group homomorphisms from π1 (X, x0 ) to π1 (Y, y2 )).

Proof. Take any [γ]p ∈ π1 (X, x0 ). We want to construct a homotopy between f2 ◦ γ and
λ−1 ∗ (f1 ◦ γ) ∗ λ. Let
G : [0, 1] × [0, 1] → Y, G(t, s) = F (t, γ(s)).
Let λt be the path from λ(t) to λ(1) = y2 along λ (via a linear reparametrization).
Then λt−1 ∗ G(t, s) ∗ λt is the path homotopy between λ−1 ∗ (f1 ◦ γ) ∗ λ and f2 ◦ γ. 
Now we prove the homotopy invariance of the fundamental group:

Proof of Theorem 1.4. From f ◦ g ∼ IdY , we get

Λ ◦ f∗ ◦ g∗ = Id : π1 (Y, y) → π1 (Y, y),
so f∗ is surjective. Conversely g ◦ f ∼ IdX implies
Ü ◦ g∗ ◦ f∗ = Id : π1(X, x) → π1(X, x).
Λ
It follows f∗ is injective. 
Corollary 1.6. π1 (Rn \ {0}) = {e} for n > 3.
4 THE FUNDAMENTAL GROUPS (II)

2. π1 (S 1 )
We have seen many topological spaces with trivial fundamental groups. Now we
will compute the first non-trivial fundamental group:
π1 (S 1 ) ' Z.
Before we prove this, we give several applications:
Application 1. Compute the fundamental groups of many simple spaces:
• π1 (S 1 × Rn ) ' Z, ∀n,
| × ·{z· · × S }) ' Z| ⊕ ·{z· · ⊕ Z},
1 1
• π1 (S
r r
• π1 (R3 \ “z-axis” ) ' Z
• π1 (Möbius strip) ' Z.
And in fact, many more complicated fundamental groups will be computed via π1 (S 1 ).
Application 2. Distinguishing topological spaces.
Proposition 2.1. For n > 3, Rn and R2 are not homeomorphic.
Proof. Suppose Rn ' R2 , then Rn \ {0} ' R2 \ {0}. But
Rn \ {0} ∼ S n−1 , Rn \ {0} ∼ S 1
and π1 (S n−1 ) = {e}, π1 (S 1 ) ' Z, contradiction! 
Remark. • Using path connectivity (i.e. π0 ), R is not homeomorphic to Rn for n > 2.
• Using π1 , we just saw R2 is not homeomorphic to Rn for n > 3.
• In general, using πk one can prove Rk is not homeomorphic to Rn for n > k + 1.
Application 3. Non-retraction.
Theorem 2.2. The circle S 1 is not a retract of the unit disk D.
Proof. By PSet 6-3-3(a), if r : D → S 1 is a retraction, then r∗ : π1 (D) → π1 (S 1 ) is
surjective, which is impossible. 
Similarly S 1 (as the equator) is not a retract of S 2 .
Application 4. Brouwer’s fixed point theorem (n = 2)
Theorem 2.3 (Brouwer’s fixed point theorem, n = 2). For any continuous map f :
D → D, there exists p ∈ D s.t. f (p) = p.
Proof. Suppose f (p) 6= p for all p ∈ D. Then we can define g : D → S 1 by
g(p) = the intersection point of S 1 with the ray f (p)p.
One can check that the map g is continuous, and g|S 1 = Id|S 1 . So S 1 is a retract of D,
which is a contradiction! 
 THE FUNDAMENTAL GROUPS (II) 5

To compute π1 (S 1 ) = π1 (S 1 , 1), we identify S 1 ⊂ R2 = C. Intuitively, we can “see”

π1 (S 1 ) ' Z, under the correspondence:
“rotate n cycles counterclockwise along S 1 ” ! n ∈ Z.
The correspondence should preserve the group structures. To prove the desired con-
clusion, we associate to each n ∈ Z an element in π1 (S 1 , 1). It is quite obvious how to
construct such a map:
Φ : Z → π1 (S 1 , 1), n 7→ [γn ]p ,
where γn is the loop
γn : [0, 1] → S 1 , t 7→ e2πint .
Theorem 2.4. The map Φ is a group isomorphism.
Remark. So not only we get π1 (S 1 ) ' Z, but also we know that π1 (S 1 ) is generated by
either [γ1 ]p or [γ−1 ]p .

Since it will be a bit messy to directly work on S 1 , we will “lift” loops γn on S 1

to “paths” γfn on R, which are much easier to handle since R is simply connected. (It
turns out that the same idea applies to many other examples!) Note that we can regard
S 1 as the quotient space R/2πZ, with the quotient map
p : R → S 1, x 7→ e2πix .
The “lifted” map γfn is a path γfn : [0, 1] → R with γfn (0) = 0 s.t. the diagram commutes:
: R
Ü
γn
p
γn

I = [0, 1] / S1
Comparing the expressions of γn and p, we get the following very simple expression:
γfn (t) = nt.
Proof of Theorem 2.4.
Step 1. Φ is a group homomorphism.
By definition, Φ(m + n) is represented by γm+n (t) = e2πi(m+n)t . To relate the
“lifting” of γm+n with the liftings of γm and γn , we introduce the translation map
Tm : R → R, x 7→ x + m.
Then by definition, γem+n ∼ γem ∗ (Tm ◦ γen ). [Since R is simply connected.] So
p

γm+n = p ◦ γem+n ∼ p ◦ (γem ∗ (Tm ◦ γen )) = γm ∗ γn .

p

In other words,
Φ(n + m) = [γn+m ]p = [γm ]p · [γn ]p = Φ(m) · Φ(n).
6 THE FUNDAMENTAL GROUPS (II)

Step 2. Φ is surjective.
Let γ : [0, 1] → S 1 be any path with γ(0) = γ(1) = 1. We will need:
Path Lifting: ∃ unique γe : [0, 1] → R with γe (0) = 0 s.t. p ◦ γe = γ.
Suppose this is true, then the fact 1 = γ(1) = p ◦ γe (1) implies γe (1) ∈ Z. In other words,
∃n ∈ Z s.t. γe (0) = 0, γe (1) = n.
Since R is contractible, we must have γe ∼ γen . Let FÜ : [0, 1] × [0, 1] → R be a path
homotopy connecting γe and γen . Then
p

F = p ◦ FÜ : [0, 1] × [0, 1] → S 1
is a path homotopy connecting γ and γn . So [γ]p = [γn ]p = Φ(n).
Step 3. Φ is injective.
Suppose Φ(n) = Φ(m), i.e. ∃ path homotopy F : [0, 1] × [0, 1] → S 1 connecting γn
and γm . The idea is to “lift” anything on S 1 to R. So we need
Homotopy Lifting: ∃ unique FÜ : [0,1]×[0,1] → R with FÜ(t,0) ≡ 0, p◦ FÜ = F.
Assuming this, then
p ◦ FÜ(0, t) = γn (t), p ◦ FÜ(1, t) = γm (t).
Since FÜ(0, 0) = 0 and FÜ(1, 0) = 0, by uniqueness of path lifting above, we must have
FÜ(0, t) = γen (t), FÜ(1, t) = γem (t).
Observe that F (s, 1) = 1 implies FÜ(s, 1) ∈ Z. But FÜ is continuous, so
n = γen (1) = FÜ(0, 1) = FÜ(s, 1) = FÜ(1, 1) = γem (1) = m.

Remark. In Lecture 20 we will extend the same idea to compute the fundamental group
for topological spaces that are more complicated than S 1 but have a similar “covering
structure”.
Remark. Let f : S 1 → S 1 be any continuous map. Consider the induced map
f∗ : π1 (S 1 ) → π1 (S 1 ).
Since π1 (S 1 ) ' Z, there exists n ∈ Z such that f∗ (1) = n. The integer n is called the
degree of the map f , and is denoted by deg(f ). One can prove:
(1) If f, g ∈ C(S 1 , S 1 ), then deg(f ◦ g) = deg(f )deg(g).
(2) f is homotopic to g if and only if deg(f ) = deg(g).
 THE FUNDAMENTAL GROUPS (II) 7

3. The Lifting Lemmas

In what follows we denote I = [0, 1]. It remains to prove the two lifting lemmas.
Path Lifting: ∃ unique γe : I → R with γe (0) = 0 s.t. p ◦ γe = γ.
Homotopy Lifting: ∃ unique FÜ : I × I → R with FÜ(t,0) ≡ 0 s.t. p ◦ FÜ = F.
Since they look similar, we will give a unified proof. We will prove the following
stronger result [so that the path lifting is the special case where P is a single point set,
and the homotopy lifting is a special case where P = I = [0, 1]].
Lemma 3.1 (The Lifting Lemma). Given any continuous map F : P × I → S 1 , and
any “lifting” FÜ0 : P × {0} → R of a given map
F0 = F |P ×{0} : P × {0} → S 1 ,
there is a unique lifting FÜ : P × I → R of F such that FÜ0 = FÜ|P ×{0} .
Proof.
: R ; R
Ü
F0
p
e
F
p
 
P × {0}
F0
/ S1 P ×I
F / S1
[Idea behind the proof.] We want to construct a lifting, i.e. a map
FÜ s.t. p ◦ FÜ = F. If p is invertible, then we can take FÜ = p−1 ◦ F. But
of course, p is NOT invertible!
Key observation: Locally p is invertible!

More precisely, we can find a cover of S 1 , say

U1 = S 1 \ {1}, U2 = S 1 \ {−1},
such that
• p−1 (Ui ) =
S V i is a disjoint union of open
j
j∈Z
sets in R, where
1 1
Vj1 = (j − 1, j), Vj2 = (j − , j + ).
2 2
• Each pij = p|Vji i
: Vj → Ui is a homeomor-
phism.
So we shall first try to construct “local liftings” via the local invertibil-
ity, and then “glue” them together to get the demanded “global lifting”.
[Roughly speaking, p is the “exponential map”, while pij is the “loga-
rithm map” on the imaginary axis.] [Geometrically, one can regard R
as a “helix” above a circle.]
8 THE FUNDAMENTAL GROUPS (II)

Step 1. Local lifting near s0 ∈ P .

Fix any s0 ∈ P . By Lebesgue number lemma and the tube lemma, there exists a
partition 0 = t0 < t1 < · · · < tn+1 = 1 and open neighbourhood V of s0 s.t.
F (V × [ti , ti+1 ]) ⊂ either U1 or U2 .
For simplicity, for 0 ≤ i ≤ n we denote
8
<U1 if F (V × [ti , ti+1 ]) ⊂ U1 ,
Ui = :
U2 if F (V × [ti , ti+1 ]) ⊂ U2 .

Similarly we change the indices of Vji ’s and pij ’s to 0 6 i 6 n, j ∈ Z. Since

p ◦ FÜ0 (s0 , 0) = F0 (s0 , 0) = F (s0 , 0) ⊂ U0 ,
there must exists j ∈ Z s.t. F Ý0(s0, 0) ⊂ Vj0. By continuity, there exists open V0 ⊂ V
s.t. s0 ∈ V0 and FÜ0 (V0 , 0) ⊂ Vj0 .
Now define
FÜ|V0 ×[t0 ,t1 ] = (p0j )−1 ◦ F : V0 × [t0 , t1 ] → Vj0 .
Then FÜ is continuous since
• F is continuous on V0 × [t0 , t1 ],
• F (V0 × [t0 , t1 ]) ⊂ U0 ,
• p0j : Vj0 → U0 is homeomorphism.
Moreover, FÜ|V0 ×{0} = FÜ0 |V0 ×{0} since ∀s ∈ V0 ,
• FÜ0 (s, 0) ∈ Vj0 , FÜ(s, 0) ∈ Vj0 ,
• p0j ◦ FÜ0 (s, 0) = F0 (s, 0) = F (s, 0) = p0j ◦ FÜ(s, 0).
Thus FÜ0 (s, 0) = FÜ(s, 0). [So FÜ is a super-local lifting: local near s0 and local in t.]
We can continue this construction, with “initial value” FÜ0 |V0 ×{0} replaced by FÜ|V0 ×{t1 } ,
to get open set V1 ⊂ V0 and a continuous map F̃ defined on V1 × [0, t2 ] such that p ◦ F̃
on V1 × [0, t2 ] and FÜ|V1 ×{0} = FÜ0 |V1 ×{0} .
Here is how: Since p ◦ FÜ(s0 , t1 ) = F (s0 , t1 ) ∈ U1 , there exists j ∈ Z
s.t. FÜ(s0 , t1 ) ∈ Vj1 . So one can find open V1 ⊂ V0 s.t. s0 ∈ V1 and
FÜ(V1 , t1 ) ⊂ Vj1 . Define
FÜ|V1 ×[t1 ,t2 ] = (p1j )−1 ◦ F : V1 × [t1 , t2 ] → Vj1 ⊂ R.
As before one can check that FÜ|V1 ×[t0 ,t2 ] is continuous.
Repeating the same construction, we finally get an open neighbourhood V Ü of s0, and
a continuous lifting FÜ : V
Ü × [0, 1] → R of F : VÜ × [0, 1] → S 1 s.t. FÜ|Ve ×{0} = FÜ0|Ve ×{0}.
 THE FUNDAMENTAL GROUPS (II) 9

Step 2. Uniqueness of lifting for P = {pt}.

If P is a one-point set, then by Step 1, the lifting FÜ exists on the whole P × I = I.
We want to show that such a lift is unique. Suppose FÜ0 is another lifting. Then
FÜ(0) = FÜ0 (0) = FÜ0 (0). [Note: In this case, FÜ, FÜ0 : [0, 1] → R, and F : [0, 1] → S 1 .]
Again we pick 0 = t0 < t1 < · · · < tn+1 = 1 s.t. F ([ti , ti+1 ]) ⊂ Ui as before. Since
• p ◦ FÜ([0, t1 ]) ⊂ U0 and p ◦ FÜ0 ([0, t1 ]) ⊂ U0 ,
• FÜ([0, t1 ]) and FÜ0 ([0, t1 ]) are connected,
• FÜ(0) = FÜ0 (0),
we must have FÜ|[0,t1 ] = FÜ0 |[0,t1 ] . The same argument applies to each interval [tk , tk+1 ].
So we conclude FÜ = FÜ0 on [0, 1].
Step 3. Existence and Uniqueness of lifting for general P .
We have shown in Step 1 that for any s ∈ P, there exists open neighbourhood Vs
of s in P and a “lifting” FÜs : Vs × I → R of Fs = F |Vs ×I : Vs × I → S 1 s.t.
FÜs |Vs ×{0} = FÜ0 |Vs ×{0} .
Now suppose s0 ∈ Vs1 ∩ Vs2 . Then
FÜs1 ,s0 := FÜs1 |{s0 }×I : I → R and FÜs2 ,s0 := FÜs2 |{s0 }×I : I → R
are both liftings of Fs0 = F |{s0 }×I : I → S 1 , and
FÜs1 ,s0 (0) = FÜ0 (s0 , 0) = FÜs2 ,s0 (0).
By Step 2, FÜs1 ,s0 = FÜs2 ,s0 . So if we define FÜ : P × I → R by
FÜ|Vs ×I = FÜs ,
then FÜ is well-defined, continuous, and is a lifting of F satisfying the given initial
Ý0.
condition F
Finally, such lifting must be unique, because if FÜ and FÜ0 are two liftings of F , then
for each s ∈ P , FÜ|{s}×I and FÜ0 |{s}×I are both liftings of F |s×{0} , and thus by Step 2
they must be the same.