0% found this document useful (0 votes)
12 views11 pages

Main

Uploaded by

Prabina Gautam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
12 views11 pages

Main

Uploaded by

Prabina Gautam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

The Inverse Galois Problem over C(t)

Swayam Chube

December 1, 2024

Abstract
This is an attempt to present a self-contained proof of the Inverse Galois Prob-
lem over C(t). The only result used without proof is Riemann’s Existence Theorem
(Theorem 3.1).

§1 R IEMANN S URFACES AND H OLOMORPHIC M APS


D EFINITION 1.1. Let X be a two-dimensional manifold. A complex chart on X is a homeo-
morphism φ : U → V of an open subset U ⊆ X onto an open subset V ⊆ C.
Two complex charts φi : Ui → Vi , i = 1, 2 are said to be holomorphically compatible if the
map
φ2 ◦ φ1−1 : φ1 (U1 ∩ U2 ) → φ2 (U1 ∩ U2 )
is biholomorphic.
A complex atlas on X is a system A = { φi : Ui → Vi | i ∈ I } of charts which are
holomorhpically compatible and which cover X. Two complex atlases A and A′ on X are
said to be analytically equivalent if every chart of A is holomorphically compatible with
every chart of A′ .
A complex structure on a two-dimensional manifold X is an equivalence class of analyti-
cally equivalent atlases on X.
A Riemann surface is a pair ( X, Σ) where X is a connected two-dimensional manifold
and Σ is a complex structure on X.
E XAMPLE 1.2. Consider P1 := C ∪ {∞}, the one-point compactification of C. Let U1 =
C ⊆ P1 and U2 = C∗ ∪ {∞}. Consider the charts φ1 : U1 → C, the identity map, and
φ2 : U2 → C given by (
1
z z ∈ C∗
φ(z) =
0 z = ∞.
These are compatible charts since the transition function is z 7→ 1
z on C∗ .
E XAMPLE 1.3. If X is a Riemann surface and Y ⊆ X is a connected open set, then every
chart of X restricts to a chart on Y (by restriction of the domain) and these are still holo-
morphically comptible. Thus, Y inherits a natural Riemann surface structure from X. In
particular, every open subset of C is a Riemann surface.

1
D EFINITION 1.4. A map f : X → Y of Riemann surfaces is said to be holomorphic if for
every pair of charts ψ1 : U1 → V1 on X and ψ2 : U2 → V2 on Y with f (U1 ) ⊆ U2 , the
mapping ψ2 ◦ f ◦ ψ1−1 : V1 → V2 is holomorphic.
A holomorphic function on X means a holomorphic function f : X → C. These form a
ring denoted by O ( X ).

T HEOREM 1.5 (R IEMANN ’ S R EMOVABLE S INGULARITIES T HEOREM ). Let X be a Rie-


mann surface and f ∈ O ( X \ { a}). If f is bounded in a neighborhood of a, then f can be
extended to a holomorphic function fe ∈ O ( X ).

Proof. Follows from the analogous statement in elementary complex analysis. ■


T HEOREM 1.6 (I DENTITY T HEOREM ). Let X and Y be Riemann surfaces and f 1 , f 2 : X → Y
be two holomorphic mappings which coincide on a set A ⊆ X having a limit point a ∈ X.
Then f 1 = f 2 .

Proof. Let

B = { x ∈ X : there is a neighborhood W of x such that f 1 |W = f 2 |W }.

By definition, B is open. By continuity, note that a ∈ A. Considering charts centered at


a and f 1 ( a) = f 2 ( a), and using the identity theorem from elementary complex analysis,
it is not hard to see that a ∈ B, that is, B ̸= ∅. Finally, suppose bn → b ∈ X. Then, by
continuity, y = f 1 (b) = f 2 (b). Consider charts centered at b and y. Note that bn lies in the
chart centred at b for sufficiently large n and hence, it would follow that b ∈ B. Thus, B is
a clopen nonempty subset of X. Owing to the connectedness of X, B = X. This completes
the proof. ■
D EFINITION 1.7. A meromorphic function on a Riemann surface X is a holomorphic function
f : X ′ → C, where X ′ ⊆ X is an open subset, such that the following hold:

(a) X \ X ′ is discrete.

(b) For every p ∈ X \ X ′ ,


lim | f ( x )| = ∞.
x→ p

The points of X \ X ′ are called the poles of f . The set of all meromorphic functions on X is
denoted by M ( X ).

P ROPOSITION 1.8. There is a canonical correspondence between M ( X ) and the set of


holomorphic functions X → P1 .

Proof. Straightforward. ■
C OROLLARY. M ( X ) is a field.

2
§§ Local Normal Form
T HEOREM 1.9. Let X and Y be Riemann surfaces and f : X → Y a non-constant holomor-
phic map. Suppose a ∈ X and b = f ( a) ∈ Y. Then, there exists an integer k ⩾ 1 and charts
φ : U → V on X and ψ : U ′ → V ′ on Y with the following properties:
(i) a ∈ U, φ( a) = 0, b ∈ U ′ and ψ′ (b) = 0.
(ii) f (U ) ⊆ U ′ .
(iii) The diagram
U / U′

φ ψ
 
V / V′
z7→zk
commutes. The number k is called the multiplicity of f at a.
Proof. Begin with two charts φ1 : U → V1 and ψ1 : U ′ → V1′ satisfying (i) and (ii). The
induced map V1 → V1′ takes 0 to 0 and hence, is of the form zk g(z) for some k ⩾ 1 and
holomorphic g : V1 → V1′ with g(0) ̸= 0. Shrinking all the open sets if necessary, we may
suppose that g(z) = h(z)k for some holomorphic function h : V1 → C. Note that zh(z)
must be injective and non-constant on V1 whence maps V1 biholomorphically onto some
V ⊆ C. We obtain the following commutative diagram
f
U / U′

φ1 ψ1
 (zh(z))k 
V1 / V′
> 1
zh(z)
 z7→zk
V
thereby completing the proof. ■
T HEOREM 1.10 (O PEN M APPING T HEOREM ). A non-constant holomorphic map between
Riemann surfaces is open.
Proof. Since being open is a local property, this follows immediately from Theorem 1.9. ■
C OROLLARY. Let f : X → Y be an injective holomorphic map of Riemann surfaces. Then
f is a biholomorphic mapping of X onto Z = f ( X ).
Proof. Due to Theorem 1.10, Z ⊆ Y is open. Since f is injective, it follows from Theorem 1.9
that k = 1 at each point of X. In particular, f is a local homeomorphism onto Z. The
conclusion follows. ■
T HEOREM 1.11. If X is a compact Riemann surface and f : X → Y a non-constant
holomorphic map of Riemann surfaces, then f is surjective.
Proof. The image of f is both open and closed in Y. ■
C OROLLARY. If X is a compact Riemann surface, then O ( X ) consists of only constant
functions.

3
§§ Branched and Unbranched Coverings
D EFINITION 1.12. Let p : Y → X be a non-constant holomorphic map of Riemann surfaces.
A point y ∈ Y is said to be a branch point or ramification point of p, if there is no neighborhood
V of y such that p|V is injective. The map p is called an unbranched holomorphic map if it has
no branch points.

T HEOREM 1.13. A non-constant holomorphic map p : Y → X is unbranched if and only if


p is a local homeomorphism, i.e., every point y ∈ Y has an open neighborhood V which is
mapped homeomorphically by p onto an open set U in X.

Proof. Immediate from the definition since an injective map of Riemann surfaces is a
biholomorphism onto its image. ■
T HEOREM 1.14. Let X be a Riemann surface, Y a connected Hausdorff topological space,
and p : Y → X a local homeomorphism. Then there is a unique complex structure on Y
such that p is holomorphic.

Proof. Suppose φ1 : U1 → V1 ⊆ C is a chart of the complex structure of X such that there is


an open subset U ⊆ Y with p|U : U → U1 a homeomorphism. Then, φ := φ ◦ p : U → V
is a complex chart on Y. Let A be the set of all complex charts on Y obtained in this way. It
is easy to see that the charts of A cover Y and are holomorphically compatible. Thus, we
have defined a complex structure on Y and it follows that p is a holomorphic map when Y
is equipped with this structure.
Suppose (Y, Σ) and (Y, Σ′ ) are two complex charts such that p is holomorphic, then
id : (Y, Σ) → (Y, Σ′ ) is a bijective holomorphic map, whence a biholomorphism. This
shows uniqueness. ■
T HEOREM 1.15. Let X, Y, Z be Riemann surfaces, p : Y → X an unbranched holomorphic
map and f : Z → X any holomorphic map. Then, every continuous lift g : Z → Y of f is
holomorphic.

Proof. Let z ∈ Z, x = f (z), and y = g(z). There is a neighborhood V of y in Y such that


p|V is injective. Let U = p(V ) ⊆ X, which is open and biholomorphic to V through p. If
−1
W = g−1 (V ), then g|W = p|V ◦ f |W whence g is holomorphic. ■
D EFINITION 1.16. A continuous map f : X → Y of topological spaces is said to be proper
if f −1 (K ) is compact in X for every compact subset K of Y. The map f is said to be discrete
if every fiber is discrete in X.

L EMMA 1.17. A proper map between locally compact Hausdorff spaces is closed.

Proof. Follows from the fact that a subset of an LCH space is closed if and only if its
intersection with every compact subset is closed. ■
C OROLLARY. A proper holomorphic map between Riemann surfaces is surjective.

Proof. The image is both closed and open. ■

4
L EMMA 1.18. Let X and Y be locally compact Hausdorff. If p : Y → X is a proper, discrete
map then:

(a) for every x ∈ X, the set p−1 ( x ) is finite.

(b) if x ∈ X and V is a neighborhood of p−1 ( x ), then there is a neighborhood U of x


with p−1 (U ) ⊆ V.

Proof. (a) Compact discrete sets must be finite.

(b) Since Y \ V is closed, due to the preceding lemma, A = p(Y \ V ) is closed in X and
x∈/ A. Hence, U = X \ A is an open neighborhood of x such that p−1 (U ) ⊆ V. ■

T HEOREM 1.19. Let X and Y be locally compact Hausdorff spaces and p : Y → X a proper
local homeomorphism. Then p is a covering map.

Proof. Choose any x ∈ X and let p−1 ( x ) = {y1 , . . . , yn }. Since p is a local homeomorphism,
we can inductively choose disjoint neighborhoods Wi of yi and a neighborhood V of x
such that the restriction p|Wi : Wi → V is a homeomorphism. It follows that p is a covering
map. ■
P ROPOSITION 1.20. The set of branch points of a non-constant holomorphic map between
Riemann surfaces is a discrete closed set.

Proof. Let f : X → Y be a non-constant holomorphic map. Let a ∈ X be a branch point


and b = f ( a). Then due to Theorem 1.9, there are charts φ : U → V and φ′ : U ′ → V ′
centered at a and b respectively such that the induced map V → V ′ is z 7→ zk for some
positive integer k ⩾ 2 (since a is a branch point). But for any 0 ̸= z ∈ V ′ , the map V → V ′
is a local homeomorphism and hence, the set of branch points forms a discrete set.
To see that it is closed, let a ∈ X not be a branch point. Then, there is a neighborhood
V of a on which f is injective and hence, none of the points in V are branch points. This
shows that the set of branch points is also closed. ■
D EFINITION 1.21. Let f : X → Y be a proper holomorphic map. As we have seen earlier,
f is surjective. Let A ⊆ X be the set of branch points of f . Since f is proper, the set
B = f ( A) ⊆ Y is closed and discrete (use the Local Normal Form). One calls B the set of
critical values of f .

With notation as above, let Y ′ = Y \ B and X ′ = f −1 (Y ′ ) ⊆ X \ A. The restric-


tion f : X ′ → Y ′ is a proper unbranched holomorphic covering map since it is a local
homeomorphism (owing to the fact that all branch points have been removed). It has a
well-defined finite number of sheets, say n. Thus, every value c ∈ Y ′ is taken precisely n
times. We would like to extend this notion to critical values.
For x ∈ X, denote by V ( f , x ), the multiplicity of f at x in the sense of Theorem 1.9. We
say that f takes the value c ∈ Y, counting multiplicities, m times on X, if

m= ∑ v ( f , x ).
x ∈ f −1 ( c )

5
T HEOREM 1.22. Let f : X → Y be a proper non-constant holomorphic map between
Riemann surfaces. Then there exists a natural number n such that f atkes every value
c ∈ Y, counting multiplicities, n times.

Proof. Using the notation as in the preceding paragraph, let n be the number of sheets
of the unbranched covering f : X ′ → Y ′ . Suppose b ∈ B is a critical value, p−1 (b) =
{ x1 , . . . , xr } and k i = v( f , xi ). Due to Theorem 1.9, there are disjoint neighborhoods Uj
of x j and Vj of b such that for every c ∈ Vj \ {b} the set p−1 (c) ∩ Uj consists of exactly k j
points. Due to Lemma 1.18, we can find a neighborhood V ⊆ V1 ∩ · · · ∩ Vr of b such that
p−1 (V ) ⊆ U1 ∪ · · · ∪ Ur . Then for every point c ∈ V ∩ Y ′ , we have that p−1 (c) consists of
k1 + · · · + kr points. On the other hand, the cardinality of p−1 (c) must be the number of
sheets, n and hence, n = k1 + · · · + kr , thereby completing the proof. ■
R EMARK 1.23. A proper non-constant holomorphic map between Riemann surfaces will
be called an n-sheeted holomorphic covering map, where n is the integer found in the above
result. Note that holomorphic covering maps are allowed to have branch points.

Let D denote the unit disk in C and D ∗ = D \ {0}.

T HEOREM 1.24. Let f : X → D ∗ be an unbranched holomorphic covering map. Then one


of the following holds:

(a) If the covering has an infinite number of sheets, then there exists a biholomorphic
mapping φ : X → H of X onto the left half plane such that
φ
X / H

exp
f }
D∗
commutes.

(b) If the covering is k-sheeted with k < ∞, then there exists a biholomorphic mapping
φ : X → D ∗ such that
φ
X / D∗

f } z7→zk
D∗
commutes.

Proof. Follows from the Galois theory of covers and the fact that H is the universal cover
of D ∗ and
Deck( H/D ∗ ) = {τn : n ∈ Z},
where τn (z) = z + 2nπi. ■

6
T HEOREM 1.25. Let f : X → D be a proper non-constant holomorphic map which is
unbranched over D ∗ = D \ {0}. Then there is a natural number k ⩾ 1 and a biholomorphic
map φ : X → D such that
φ
X / D

f ~ z7→zk
D
Proof. The preceding theorem furnishes a k ⩾ 1 making
φ
X / D∗

f } z7→zk
D∗

commute. Let pk : D → D denote the map z 7→ zk . If we show that f −1 (0) is a singleton,


then we would be done since we could extend φ : X → D making the required diagram
commute.
Suppose f −1 (0) consists of n points b1 , . . . , bn , where n ⩾ 1. Then due to Lemma 1.18
there are disjoint open neighborhoods Vi of bi and a disk D (r ) = {z ∈ C : |z| < r },
0 < r ⩽ 1 such that
f −1 ( D (r )) ⊆ V1 ∪ · · · ∪ Vn .

Let D ∗ (r ) = D (r ) \ {0}. Since f −1 ( D ∗ (r )) is homeomorphic to p−k
1
( D ∗ (r )) = D ∗ ( k r ), it is

connected. Since every point bi is in the closure of f −1 ( D ∗ (r )), f −1 ( D (r )) is also connected.


Hence, n = 1. This completes the proof. ■

§2 A LGEBRAIC F UNCTIONS
D EFINITION 2.1. Let π : Y → X be an n-sheeted unbranched holomorphic covering of
Riemann surfaces and f ∈ M (Y ). Every point x ∈ X has an open neighborhood U such
that π −1 (U ) is the disjoint union of open sets V1 , . . . , Vn and π : Vv → U is biholomorphic
for v = 1, . . . , n. Let τv : U → Vv denote the inverse of the restricted map π : Vv → U and
let f v = τv∗ f := f ◦ τv ∈ M (U ).
Define the elementary symmetric functions c1 , . . . , cn ∈ M (U ) as

cv = (−1)v σv ( f 1 , . . . , f n ) ,

where σv is the v-th elementary symmetric polynomial in n indeterminates.


This same construction can be carried out about every point in X and it is hard to not
see that the the elementary symmetric functions glue to global meromorphic functions in
M ( X ). These are known as the elementary symmetric functions corresponding to f .

T HEOREM 2.2. Let π : Y → X be an n-sheeted branched holomorphic covering map.


Suppose A ⊆ X is a closed discrete subset containing all the critical values of π and let
B = π −1 ( A). Suppose f is a holomorphic (resp. meromorphic) function on Y \ B and

7
c1 , . . . , cn ∈ O ( X \ A) (resp. ∈ M ( X \ A)) are the elementary symmetric functions of f .
Then f may be continued holomorphically (resp. meromorphically) to Y precisely if all the
cv may be continued holomorphically (resp. meromorphically) to X.
Proof. Suppose a ∈ A and b1 , . . . , bm are the preimages of a. Suppose (U, z) is a relatively
compact coordinate neighborhood centered at a and U ∩ A = { a}. Note that V ⊆ V ⊆
π −1 (U ), which is compact since π is proper. It follows that V is relatively compact and
contains all the bµ ’s.

Case 1. Suppose f ∈ O (Y \ B)

(a) Suppose f can be continued holomorphically to all the points bµ . Then f is


bounded on V and hence, on V \ {b1 , . . . , bm }. This implies that all the cv ’s are
bounded on U \ { a}. Thus by Riemann’s theorem on removable singularities,
they may all be continued holomorphically to a.
(b) Suppose all the cv can be continued holomorphically to a; then they are all
bounded on U \ { a}. Note that for any y ∈ V \ {b1 , . . . , bm }, if x = π (y), then
f (y) is a root of the polynomial

T n + c 1 ( x ) T n −1 + · · · + c n ( x ),

whose coefficients are uniformly bounded, whence f is bounded in a neighbor-


hood of every bµ and hence, can be continued there.

Case 2. Now suppose f ∈ M (Y \ B).

(a) Assume first that f can be continued meromorphically to all points bµ . The
function φ = π ∗ z = z ◦ π ∈ O (V ) vanishes at all the points bµ . Thus, φk f may
be continued holomorphically to all the points bµ if k is sufficiently large. The
elementary symmetric functions of φk f are zkv cv and by the first part of the
proof, they may be continued holomorphically to a. Thus, all the cv may be
continued meromorphically to a.
(b) Suppose now that all the cv can be continued meromorphically to a. There is a
sufficiently large k such that all the zkv cv can be continued holomorphically to
a. Thus due to the first case, φk f admits a holomorphic continuation to all the
points bµ . This completes the proof. ■

T HEOREM 2.3. Let π : Y → X be a branched holomorphic n-sheeted covering map. If


f ∈ M (Y ) and c1 , . . . , cn ∈ M ( X ) are the elementary symmetric functions of f , then

f n + (π ∗ c1 ) f n−1 + · · · + (π ∗ cn−1 ) f + π ∗ cn = 0.

• The morphism π ∗ : M ( X ) ,→ M (Y ) is an algebraic field extension of degree ⩽ n.

• Moreover, if there exists an f ∈ M ( X ) and an x ∈ X with preimages y1 , . . . , yn ∈ Y


such that the values f (yv ) for v = 1, . . . , n are all distinct, then the field extension
π ∗ : M ( X ) ,→ M (Y ) has degree n.

8
Proof. The fact that f solves the equation follows immediately from the definition of
the elementary symmetric functions. Let L = M (Y ) and K = M ( X ). Choose f 0 ∈ L
maximizing n0 = [K ( f 0 ) : K ] ⩽ n. Let f ∈ L be arbitrary. Then, K ( f 0 , f ) is a finite extension
of K and hence, is of the form K ( g0 ) due to the Primitive Element Theorem. But then

n 0 ⩾ [ K ( g0 ) : K ] = [ K ( f 0 , f ) : K ] ⩾ [ K ( f 0 ) : K ] = n 0 ,

whence f ∈ K ( f 0 ), that is, K ( f 0 ) = L and hence, [ L : K ] = n0 ⩽ n.


Now, consider f as in the second part of the theorem and suppose its minimal polyno-
mial over K looks like

f m + (π ∗ d1 ) f m−1 + · · · + (π ∗ dm ) = 0,

where d1 , . . . , dm ∈ K. Under π, y1 , . . . , yn map to x and hence,

f (yi )m + d1 ( x ) f (yi )m−1 + · · · + dm ( x ) = 0,

but since the f (yi )’s are distinct, we must have m ⩾ n, and hence, m = n. This completes
the proof. ■
T HEOREM 2.4. Suppose X is a Riemann surface, A ⊆ X is a closed discrete subset and let
X ′ = X \ A. Suppose Y ′ is another Riemann surface and π ′ : Y ′ → X ′ a proper unbranched
holomorphic covering. Then π ′ extends to a branched covering of X, i.e., there exists
a Riemann surface Y, a proper holomorphic mapping π : Y → X and a biholomorphic
mapping φ : Y \ π −1 ( A) → Y ′ making the diagram
φ
Y \ π −1 ( A ) / Y′

π
& } π′
X\A

Proof. For every a ∈ A, choose a coordinate neighborhood (Ua , z a ) on X such that z a ( a) =


0, z a (Ua ) is the unit disk in C and Ua ∩ Ua′ = ∅ if a ̸= a′ . Let Ua∗ = Ua \ { a}. Since
π ′ : Y ′ → X ′ is proper, π ′−1 (Ua∗ ) consists of a finite number of connected components Vav
∗,

v = 1, . . . , n( a).
For every v, the restricted mapping π ′ : Vav ∗ → U ∗ is an unbranched covering. Let its
a
covering number be k av . Due to Theorem 1.24 there are biholomorphic maps ζ av : Vav ∗ →

D ∗ such that
/
∗ ζ av
Vav D∗
π′ π av
 
Ua∗ / D∗
za

commutes.
Next, let p av for a ∈ A and v = 1, . . . , n( a) be fresh points disjoint from Y ′ and set

Y = Y ′ ∪ { p av : a ∈ A, v = 1, . . . , n( a)}.

9
We now topologize Y. If Wi , i ∈ I is a neighborhood basis of a, then { p av } ∪ π ′−1 (Wi ) ∩ Vav
∗ ,


i ∈ I is set as a neighborhood basis of p av along with the fact that Y ′ retains its topology as
a subspace of Y. Define π : Y → X by π (y) = π ′ (y) if y ∈ Y ′ and π ( p av ) = a.
Next, we make Y a Riemann surface. Add to the charts of the complex structure of Y ′
the following charts. Let Vav = Vav ∗ ∪ { p } and let ζ : V → D be the continuation of
av av av
the aforementioned ζ av obtained by setting ζ av ( p av ) = 0. These charts are holomorphically
compatible and everything works out nicely. ■
T HEOREM 2.5. Let π : Y → X and τ : Z → X be proper holomorphic covering maps. Let
A ⊆ X be a closed discrete set and X ′ = X \ A, Y ′ = π −1 ( X ′ ) and Z ′ = τ −1 ( X ′ ). Then
every biholomorphic mapping σ′ : Y ′ → Z ′ making
σ′ /
Y′ Z′

π ~ τ
X′
commute can be extended to a biholomorphic mapping σ : Y → Z making
Y
σ / Z
π   τ
X
commute. In particular, Deck(Y/X ) ∼
= Deck(Y ′ /X ′ ) via this extension.
Proof. Suppose a ∈ A and (U, z) is a coordinate neighborhood of a such that z( a) = 0 and
z(U ) is the unit disk. Let U ∗ = U \ { a}. We may also assume that U is so small that π and τ
are unbranched over U ∗ . Let V1 , . . . , Vn (resp. W1 , . . . , Wm ) be the connected components of
π −1 (U ) (resp. τ −1 (U )). Then Vv∗ = Vv \ π −1 ( a) (resp. Wµ∗ ) are the connected components
of π −1 (U ∗ ) (resp. τ −1 (U ∗ )).
Since σ′ : π −1 (U ∗ ) → τ −1 (U ∗ ) is biholomorphic, n = m and one may renumber so
that σ′ (Vv∗ ) = Wv∗ . The restriction π : Vv∗ → U ∗ is a finite sheeted unbranched covering
of something biholomorphic to the punctured unit disk. It follows from Theorem 1.25
that Vv ∩ π −1 ( a) (resp. Wv ∩ τ −1 ( a)) consists of only one point bv (resp. cv ). Hence,
σ′ : π −1 (U ∗ ) → τ −1 (U ∗ ) can be continued to a bijection π −1 (U ) → τ −1 (U ). This contin-
uation is a homeomorphism. Also recall that the Vv and Wv ’s are biholomorphic to the
unit disk and hence, by Riemann’s Removable Singularities Theorem, this extension is
biholomorphic. If one applies this construction to every exceptional point a ∈ A, then one
gets the desired continuation σ : Y → Z.
Note that there is a canonical restriction map Deck(Y/X ) → Deck(Y ′ /X ′ ) which is
surjective because of what we have proved above. The injectivity is a trivial consequence
of the identity theorem. ■

§3 T HE I NVERSE G ALOIS P ROBLEM OVER C(t)


T HEOREM 3.1 (R IEMANN E XISTENCE T HEOREM ). Meromorphic functions on a compact
Riemann surface separate points.

10
T HEOREM 3.2. Every finite group can be realised as the Galois group of a field extension
of C(t).

Proof. Let G be a finite group having n elements. There is a surjection  Fn ↠ G, where


Fn is the free group on n elements. Recall that π1 P \ { x0 , . . . , xn+1 ∼
1
= Fn whence due
to the Galois theory of covering spaces for manifolds, there is a topological n-sheeted
covering π : Y ′ → P1 \ { x0 , . . . , xn+1 }. Note that this covering endows Y ′ with a unique
Riemann surface structure. Since the covering has finitely many sheets, Y is compact. Due
to Theorem 2.4, π can be extended to a branched covering π : Y → P1 .
For any σ ∈ Deck(Y/P1 ), the induced map σ∗ on M (Y ) is an element of Aut(M (Y )/M (P1 )).
This gives a natural group homomorphism:

Deck(Y/P1 ) −→ Aut(M (Y )/M (P1 )), σ 7→ σ∗ .

We contend that this map is injective. Indeed, suppose σ∗ is the identity map for some
σ ̸= 1. This is equivalent to stating that f = f ◦ σ for every f ∈ M (Y ), which is impossible
due to Theorem 3.1.
Due to Theorem 2.5, the cardinality of Deck(Y/P1 ) is precisely the cardinality of
Deck(Y ′ /P1 \ { x0 , . . . , xn+1 }), which is equal to n. Further, using Theorem 3.1 and Theo-
rem 2.3, note that [M (Y ) : M (P1 )] = n. Injectivity of the aforementioned map forces the
cardinality of Aut(M (Y )/M (P1 )) to be n whence the extension is Galois and the map is
an isomorphism. This gives Aut(M (Y )/M (P1 )) ∼ = G, thereby completing the proof. ■

11

You might also like