The n-dimensional complex projective space Pn over C is the space Cn+1\ {0} modulo the relation ∼ defined below. Two elements x and y in Cn+1\ {0} are said to be equivalent if there exists λ ∈ C \ {0} so that x = λy. If x and y are equivalent, we write x ∼ y. The equivalent class of an element x = (x0, · · · , xn) of Cn+1\ {0} is denoted by [x] = (x0: · · · : xn). Let π : Cn+1\ {0} → Pn be the quotient map. We equip Pn with the quotient topology, i.e. U ⊆ Pn is open if and only if π−1(U ) is open in Cn+1\ {0}. Since π is continuous and Cn+1\ {0} is connected (it is path connected and hence connected), Pn= π(Cn+1\ {0}) is connected. (Continuous functions send connected sets to connected sets).
The unit sphere
S2n+1= {(x0, · · · , xn) ∈ Cn:
n
X
i=0
|xi|2= 1}
is contained in Cn+1\ {0}. Let π0 be the restriction of π to S2n+1. Then π0 : S2n+1 → Pn is a surjective continuous map. Since S2n+1 is closed and bounded in Cn, it is compact by Heine-Borel Theorem. Since continuous functions map compact sets to compact sets and S2n+1 is compact, Pn= π0(S2n+1) is compact. We conclude that
Proposition 1.1. The n-dimensional complex projective space Pn is a compact and connected space.
For each 0 ≤ i ≤ n, let Ui be the set of all points (x0 : · · · : xn) in Pn with the property that there exists a representative (x0, · · · , xn) of (x0: · · · : xn) such that xi6= 0. Let us check that this set is well-defined. Let (x00, · · · , x0n) be another representative of (x0: · · · : xn). Then there exists a nonzero complex number λ in C so that x0j = λxifor 0 ≤ j ≤ n. Since xi6= 0 and λ 6= 0, x0i6= 0. (C is a field.) Therefore Ui is well-defined. Let us show that Ui is open.
Let 0 ≤ i ≤ n. We define πi : Cn+1 → C by (x0, · · · , xn) to xi. Then πi is smooth for any 0 ≤ i ≤ n. Let Ui0 be the subset of Cn+1 consisting of vectors (x0, · · · , xn) so that xi 6= 0. Then Ui0= πi−1(C \ {0}) and hence Ui0 is open by the continuity of πi and the openess of C \ {0}. Since
π−1(Ui) = Ui0∩ (Cn+1\ {0}), Ui is open in Pn. One can easily see thatSn
i=0Ui = Pnk. Hence {U0, · · · , Un} forms an open cover for Pn.
For each 0 ≤ i ≤ n, we define
ϕi: Ui→ Cn, (x0: · · · : xn) = x0 xi
, · · · ,xn xi
.
This map is well-defined (left to the reader) and is a bijection. Let us prove that ϕi is a homeomor- phism.
To show that ϕiis continuous, we show that ϕ−1i (V ) is open in Uifor any open set V of Cn. Since Ui is open in Pn, to show that ϕ−1i (V ) is open in Ui, we only need to show that ϕ−1i (V ) is open in Pn. To show that ϕ−1i (V ) is open in Pn, we show that π−1(ϕ−1i (V )) is open in Cn+1\ {0}. Let Ui0 be as above. The composition ϕi◦ π : Ui0 → Cn is given by
(ϕi◦ π)(x0, · · · , xn) = x0 xi
, · · · ,xn xi
.
It is easy for us to see that ϕi◦π : Ui0→ Cnis continuous. Hence (ϕi◦π)−1(V ) is open in Ui0and hence open in Cn+1\{0} (we use the fact that Ui0is open in Cn+1\{0}.) Since π−1(ϕ−1i (V )) = (ϕi◦π)−1(V ) is open in Cn+1\ {0}, ϕ−1i (V ) is open in Pn.
Let us prove that ϕiis an open mapping, i.e. ϕi(W ) is open in Cnfor any open set W of Ui. Since W is open in Ui and Ui is an open subset of Pn, W is also open in Pn. Therefore W0 = π−1(W ) is open in Cn+1\ {0} and hence open in Cn+1. We only need to show that ϕi(π(W0)) is open. In fact, we can show that ϕi◦ π : Ui0→ Cn is an open mapping.
1
Since open sets of the form I0× I1× · · · × In generates the topology on Cn+1, where I0, · · · , In
are open subsets of C, open sets of the form I0× I1× · · · × In∩ Ui0 generates the topology of Ui0. Hence open sets of the form I0× I1× · · · × In with Ii open in C \ {0} generates the topology of Ui0. More precisely, W0 is a union of open sets of the form I0× I1× · · · × In where Ii is open in C \ {0}.
Lemma 1.1. Let f : X → Y be any function. Suppose {Aα: α ∈ Λ} is a family of subsets of X.
Then
f [
α∈Λ
Aα
!
= [
α∈Λ
f (Aα).
Since any union of open sets is open, if we can show that (ϕi◦π)(I0×I1×· · ·×In) is open in Cnfor any open set of the form I0×I1×· · ·×Inwith Iiopen in C\{0}, Lemma 1.1 implies that (ϕi◦π)(W0) is open for any open subset W of Uiso that W0 = π−1(W ). Let Jk= {xk/xi: xi∈ Ii, xk ∈ Ik} for any k 6= i. Then
(ϕi◦ π)(I0× · · · × In) = J1× · · · × Ji−1× Ji+1× · · · × Jn.
If we can show that all Jk are open in C for any k 6= i, by the fact that the product of open sets is open, J1× · · · × Ji−1× Ji+1× · · · × Jn is open in Cn and hence (ϕi◦ π)(I0× · · · × In) is open in Cn.
For each µ ∈ Ii, we see that
Jk = [
µ∈Ii
{xk/µ : xk∈ Ik} = [
µ∈Ii
µ−1Ik.
Since Ik is open in C, µ−1Ik is open for any µ ∈ Ii. Since any union of open sets is open, Jk is open in C. We complete the proof of our assertion. We conclude that ϕi: Ui→ Cn is a homeomorphism for all 0 ≤ i ≤ n.
Let 0 ≤ j < i ≤ n. Then
ϕi(Ui∩ Uj) = {(z1, · · · , zn) ∈ Cn: zj6= 0}.
Hence ϕi(Ui∩ Uj) is open. The transition functions are given by (ϕj◦ ϕ−1i ) : ϕi(Ui∩ Uj) → Cn
(ϕj◦ ϕ−1i )(z1, · · · , zn) = z1 zj
, · · · ,zj−1 zj
,zj+1 zj
, · · · ,zi zj
, 1 zj
,zi+1 zj
, · · · ,zn zj
Hence ϕj◦ ϕ−1i is biholomorphic on ϕi(Ui∩ Uj).
Before proving Pn is Hausdorff, let us recall some basic facts from topology.
Lemma 1.2. Let X be a Hausdorff space and K be a compact subset of X. Then K is closed.
Proof. To show that Y is closed, we show that X \ K is open. Let us prove that every point of X \ K is an interior point of X \ K. Let x ∈ X \ K. For any y ∈ K, x 6= y. Since X is Hausdorff, we can find an open neighborhood Uy of x and an open neighborhood Vy of y such that Uy∩ Vy = ∅.
Then {Vy: y ∈ K} forms an open cover for K. Since K is compact, we can find y1, · · · , yk so that {Vyi : 1 ≤ i ≤ k} covers K. Take U =Tk
i=1Uyk. Then U is an open neighorhood of x. Let us show that U is contained in X \ K.
Let z ∈ U. If z ∈ K, then z ∈ Vyi for some 1 ≤ i ≤ k. Then z ∈ Vyi∩ U ⊆ Vyi∩ Uyi = ∅ which is impossible. Therefore z 6∈ K and hence z ∈ X \ K. We see that U ⊆ X \ K. We find that x is an
interior point of X \ K.
Corollary 1.1. Let X be a compact space and Y be a Hausdorff space. Suppose f : X → Y is a bijective continuous function. Then f is a homeomorphism.
Proof. To show that f is a homeomorphism, we only need to prove that f is a closed mapping. Let A be any closed subset of X. Since X is compact and A is a closed subset of X, A is compact. Since f is continuous and A is compact, f (A) is compact. Since Y is closed and f (A) is compact, f (A) is
a closed subset of Y.
Definition 1.1. A topological space X is normal if given any disjoint closed subsets E and F, there exist open neighborhood U of E and V of F such that U ∩ V = ∅.
Proposition 1.2. Any compact Hausdorff space is normal.
Proof. Let X be a compact Hausdorff space. Let B and K be disjoint closed subset of X. Since X is compact, both B and K are also compact. Let b ∈ B. By the proof of Lemma 1.2 and the compactness of K, we can choose an open neighborhood V =Sk
i=1Vyiof K for some y1, · · · , yk∈ K and an open neighborhood Ubof b such that Ub∩ V = ∅. Since B is compact and {Ub: b ∈ B} forms an open cover for B, there exist b1, · · · , bl∈ B so that {Ubi : 1 ≤ i ≤ l} forms an open cover for B.
Let U =Sl
i=1Ubi. Then U is an open neighborhood of B. Claim U ∩ V = ∅. Let z ∈ U ∩ V. Then z ∈ V and z ∈ Ubi for some i. Hence z ∈ V ∩ Ubi= ∅ which is not possible. Proposition 1.3. Let X and Y be compact spaces. Then the product space X × Y is also compact.
To prove that Pn is Hausdorff, we show that Pn is homeomorphic to S2n+1/S1. We will show that S2n+1/S1 is a Hausdorff space.
Define an action of S1 on S2n+1 by
S1× S2n+1→ S2n+1, (λ, x) 7→ λx.
Let us show that this action is continuous. Let α : C × Cn → Cn be the function (λ, x) 7→ λx.
Then α is continuous. The action of S1 on S2n+1 is the restriction of α to S1× S2n+1; hence it is continuous. Two points x and y of S2n+1 are equivalent if there exists λ ∈ S1 so that x = λy.
The quotient space of S2n+1 modulo this relation is denoted by S2n+1/S1. The equivalent class of x ∈ S2n+1modulo this relation is denoted by [x]S1. The quotient map S2n+1→ S2n+1/S1is denoted by q.
Definition 1.2. A group with a topology is called a topological group if the function G × G → G, (a, b) → ab−1
is continuous. Here G × G is equipped with the product topology.
Example 1.1. C∗= {z ∈ C : z 6= 0} with the subspace topology induced from C is a commutative topological group called a noncompact (one dimensional) complex torus. The subset S1 of C∗ consisting of complex numbers z so that |z| = 1 forms a compact subgroup of C.
To show that S2n+1/S1is Hausdorff, we need the following Lemma.
Lemma 1.3. Let X be a compact Hausdorff space and G be a compact topological group. Suppose G acts on X continuously, i.e. the function
m : G × X → X, (g, x) 7→ gx is continuous. Then
(1) the quotient map π : X → X/G is an open mapping.
(2) the quotient map π : X → X/G is a closed mapping.
(3) the orbit space X/G is Haudorff.
Proof. At first, let us prove that the quotient map π : X → X/G is an open mapping. To show that π is an open mapping, we need to show that π(U ) is open in X/G for any open set U of X. To show that π(U ) is open in X/G, we need to show that π−1(π(U )) is open in X. Claim that
(1.1) π−1(π(U )) = [
g∈G
g(U ).
Since g : X → X is a homeomorphism for any g ∈ G, g(U ) is open in X for any open subset U of X.
If the above equation is true, then π−1(π(U )) is open (any union of open subsets of X is open). Let y ∈ π−1(π(U )), then π(y) ∈ π(U ) and hence π(y) = π(z) for some z ∈ U. Therefore y = gz ∈ g(U ) for some g ∈ G. Hence y ∈S
g∈Gg(U ). Hence π−1(π(U )) ⊆S
g∈Gg(U ). Conversely, if y ∈S
g∈Gg(U ),
then y ∈ g(U ) for some g ∈ G. Hence y = gz for some z ∈ U. Therefore π(y) = π(z) ∈ π(U ) which implies that y ∈ π−1(π(U )). HenceS
g∈Gg(U ) ⊆ π−1(π(U )). We conclude that (1.1) holds.
Since X is compact, any closed subset of X is also compact. Let A be a closed subset of X.
Then A is compact. Since any product of compact spaces is compact, G × A is compact. Since G(A) = m(G, A) and m is continuous, G(A) =S
g∈Gg(A) is compact. To show that π is a closed mapping, we show that π−1(π(A)) is closed in X. In fact, π−1(π(A)) = G(A) is a compact subset of a Hausdorff space X, it is closed.
The equivalence class of a point x ∈ X is the orbit [x] = Gx = {y ∈ X : y = gx, g ∈ G}. Since G × X → X is continuous and G is compact, the orbit Gx is compact. If [x] 6= [y], then Gx and Gy are disjoint compact subsets of X. Since X is Hausdorff, we can find disjoint open sets U and V of X so that Gx ⊆ U and Gy ⊆ V. Since V is a closed subset of X and π is a closed mapping, π(V ) is closed in X/G. Let U0 = (X/G) \ π(V ). Then U0 is an open subset of X/G. Since π is an open mapping, V0 = π(V ) is open in X/G. We obtain disjoint open sets {U0, V0}. Since Gy ⊆ V, [y] ∈ π(V ) = V0. Let us show that [x] ∈ U0.
Since X is compact and V is a closed subset of X, V is compact. Since Gx is compact and V is compact, we may choose disjoint open sets U and W so that Gx ⊆ U and V ⊆ W. Since U ∩ W = ∅, Gx ∩ W = ∅. Hence [x] 6∈ π(W ) which implies that [x] 6∈ π(V ), i.e. [x] ∈ U0 = (X/G) \ π(V ).
We find {U0, V0} disjoint open subsets of X/G so that [x] ∈ U0 and [y] ∈ V0 for any [x], [y] in X/G with [x] 6= [y]. Therefore X/G is a Hausdorff space.
This lemma implies that S2n+1/S1is a (compact) Hausdorff space.
Lemma 1.4. The complex projective space Pn is homeomorphic to S2n+1/S1. Proof. Let r : Cn+1\ {0} → S2n+1 be the function r(z) = z
kzk for z ∈ Cn+1\ {0}. Then r is continuous. Define f : Pn → S2n+1/S1 by
f ([z]) = [r(z)]S1
where z is a representative of [z]. Let us check that this map is well-defined. Suppose z and z0 are equivalent. Then z = µz0 for some nonzero complex number µ. Then
r(z) = z
kzk = µz0
|µ|kz0k = λr(z0) for λ = µ/|µ|. In other words, [r(z)]S1= [r(z0)]S1.
If [r(z)]S1 = [r(z0)]S1, then
z
kzk= λ z0
kz0k =⇒ z = λkzk kz0kz0.
Hence [z] = [z0]. This shows that f is injective. Let [x]S1 be a point in S2n+1/S1. Choose a repre- sentative x of [x]S1. Then x ∈ S2n+1⊆ Cn+1\ {0} and r(x) = x. Then f ([x]) = [r(x)]S1 = [x]S1. We prove that f is surjective.
By definition, we have the following commutative diagram:
Cn+1\ {0} −−−−→r S2n+1
π
y
y
q
Pn
−−−−→ Sf 2n+1/S1.
To show that f is continuous, we show that f−1(V ) is open in Pnfor any open subset V of S2n+1/S1. To show that f−1(V ) is open in Pn, we show that π−1(f−1(V )) is open in Cn+1\ {0}. By the above commutative diagram,
π−1(f−1(V )) = (f ◦ π)−1(V ) = (q ◦ r)−1(V ) = r−1(q−1(V )).
Since V is open in S2n+1/S1, q−1(V ) is open in S2n+1. Since r is continuous and q−1(V ) is open in S2n+1, r−1(q−1(V )) is open in Cn+1\ {0}. This proves that π−1(f−1(V )) is open in Cn+1\ {0}.
Hence f−1(V ) is open in Pn. We conclude that f is continuous.
Since f is a bijective continuous map and Pn is compact and S2n+1/S1 is Hausdorff, f is a
homeomorphism.
Lemma 1.5. Let X and Y be two spaces. Suppose f : X → Y is a homeomorphism. Then X is Hausdorff if and only if Y is Hausdorff.
Proof. Let us assume that X is Hausdorff. Suppose y1and y2are two points of Y such that y16= y2. Since f is surjective, there exist x1and x2in X so that f (xi) = yi for i = 1, 2. Since f is a function, x16= x2. (If x1= x2, y1= f (x1) = f (x2) = y2.) Since X is Hausdorff, there exist open neighborhoods Ui of xi so that U1∩ U2 = ∅. Let Vi = f (Ui) for i = 1, 2. Then Vi are open neighborhood of yi for i = 1, 2. Claim V1∩ V2= ∅. Suppose not. Take z ∈ V1∩ V2. By surjectivity of f, we can find x ∈ X so that z = f (x). Then x ∈ f−1(V1∩ V2) ⊆ U1∩ U2= ∅ which is impossible.
Since Pn is homeomorphic to S2n+1/S1 and S2n+1/S1 is Hausdorff, Pn is also Hausdorff.