All the Riemann surfaces in this note are assumed to be connected and com- pact.
Let X be a Riemann surface of genus g ≥ 0 and K(X) be the field of meromorphic functions on X. The free abelian group generated by points of X is denoted by Div(X).
Elements of Div(X) are called divisors on X. They are of the form
(1.1) D = n1P1+ · · · + nkPk
for some n1, · · · , nk ∈ Z and for some points P1, · · · , Pk ∈ X and for some k ≥ 1. Or equivalently, D can be written as
D = X
P ∈X
nPP,
where n : X → Z is a function so that nP = 0 except a finite number of P ∈ X.
Given a divisor D of the form (1.1), its degree denoted by deg D is defined to be
(1.2) deg D = X
P ∈X
nP.
The map deg : Div(X) → Z sending a divisor D to its degree deg D is a group homomor- phism.
Let f be a nonconstant meromorphic function on X and P be a point of X. In a neigh- borhood of P, we write f = g/h with f, g holomorphic. We set
ordP f = ordPg − ordPh.
We say that f has a zero of order ordPf if ordPf > 0 and that f has a pole of order ordPf if ordPf < 0. If P is neither a zero nor a pole, we set ordPf = 0. We define a divisor
(f ) = X
P ∈X
(ordPf )P.
If f = g/h with g, h holomorphic and relatively prime, the divisors of zero (f )0 is defined to be
(f )0=X
P
(ordPg)P and the divisor of poles is defined to be
(f )∞= X
P ∈X
(ordP h)P.
Clearly, these divisors are well-defined as long as we require that g, h are relatively prime, and
(f ) = (f )0− (f )∞.
It is clear that the map (·) : K(X)∗ → Div(X) sending f to its associated divisor (f ) is a group homomorphism. A divisor D is called principal if there exists f ∈ K(X)∗ such that D = (f ).
Lemma 1.1. Let f ∈ K(X)∗. Then deg(f ) = 0.
Proof. If f is a constant function, (f )0 = (f )∞= 0. Hence deg(f ) = 0 is obvious. Assume that f is not a constant function. Since X is compact, there are only finitely many zeros and poles of f on X. Write (f )0= n1P1+ · · · + nkPkand (f )∞= m1Q1+ · · · + mlQl where
1
P1, · · · , Pk, Q1, · · · , Ql ∈ X and ni, mj ≥ 1. Since a nonconstant meromorphic function f on X is a holomorphic mapping f : X → P1, the degree of f is defined to be
deg f = X
P ∈f−1(Q)
eP.
Here eP is the ramification index of f at P (or the multiplicity). When we choose Q = 0, we get deg f = n1+ · · · + nk. When we choose Q = ∞, we obtain deg f = m1+ · · · + ml. Then
deg(f ) = (n1+ · · · + nk) − (m1+ · · · + ml) = deg f − deg f = 0.
A meromorphic one-form ω on X is also called an abelian differential. Given an abelian differential ω, we can also define its associated divisor
(ω) = X
P ∈X
(ordPω)P.
A divisor of this form is called a canonical divisor. In the next lemma, we are going to see that any two canonical divisors are linearly equivalent.
Lemma 1.2. Let ω1 and ω2 be two abelian differentials on X with ω1 6= 0. Then there is a unique geomorphic function f on X such that ω2 = f ω1. Hence (ω1) is linearly equivalent to (ω2).
Proof. Let φ : U ⊂ X → V ⊂ C be a local chart on X. Write ωi◦ φ(z) = gi(z)dz for some meromorphic function gi on V. Set h = g2/g1. Then f = h ◦ φ is a meomorphic function on U. Then f can be glued to a global meromorphic function on X such that ω2 = f ω1. This implies that (ω2) = (f ) + (ω1). In other words, (ω2) − (ω1) = (f ) is a principal divisor.
Hence (ω1) and (ω2) are linearly equivalent.
Notice that by deg(f ) = 0, we have
deg(ω2) = deg(f ) + deg(ω1) = deg(ω1).
This formula tells us that every canonical divisor has the same degree. Now, let us compute the degree of a canonical divisor on a compact Riemann surface. In order to do this, we need the following lemma.
Lemma 1.3. Let f : X → Y be holomorphic mapping between Riemann surfaces and ω be a meromorphic one-form on Y. For a point p ∈ X, we have
ordp(f∗ω) = (1 + ordf (p)ω)ep(f ) − 1.
Proof. Let (ψ, U ) and (ϕ, V ) be local charts around p and f (p) with f (U ) ⊂ V, and ψ(p) = 0, and ϕ(f (p)) = 0 such that the local representation F = ϕ ◦ f ◦ ψ−1 of f is given by the local normal form z 7→ zn where m = ep(f ). Assume that ω is represented by ωU(w) = (ckwk+ ck+1wk+1+ · · · )dw in the local chart (ϕ, V ) with ck6= 0 and k = ordf (p)ω. Hence the local representation of f∗ω in local chart (ψ, U ) is represented by
F∗ωV = F∗(ckwk+ ck+1wk+1+ · · · )dF∗w
= (ckznk+ ck+1zn(k+1)+ · · · )dzn
= (ckznk+ ck+1zn(k+1)+ · · · )nzn−1dz
= (nckznk+n−1+ nck+1cn(k+1)+n−1+ · · · )dz.
This implies that ordp(f∗ω) = nk + n − 1 = (ordf (p)ω + 1)ep(f ) − 1 which proves our assertion.
Proposition 1.1. Let X be a compact Riemann surface of genus g and KX be a canonical divisor. Then deg KX = 2g − 2.
Proof. Let f be a nonconstant meromorphic function on X. Then f : X → P1 is a noncon- stant holomorphic mapping and thus a ramified covering of P1. Denote by d the degree of f. The Riemann-Hurwitz formula implies
X
p∈X
bp(f ) = 2g − 2 + 2d.
Let us define a one-form ω on P1as follows. Write P1 = C ∪ {∞}. Let z be the coordinate on C and w = 1/z be a coordinate around ∞. Define ω = dz. Then ω = −dw/w2 as a pole of order two around ∞ and no other zeros or poles. Hence (ω) = −2∞. Denote η = f∗ω.
Then η is a meromorphic one form on X. Now let us compute the degree of (η).
deg(η) = X
p∈X
(ordpη)
= X
p∈X
(ordpf∗ω)
= X
p∈X
((1 + ordf (p)ω)ep(f ) − 1)
= X
p∈X, f (p)6=∞
((1 + ordf (p)ω)ep(f ) − 1) + X
p∈X, f (p)=∞
((1 + ordf (p)ω)ep(f ) − 1) We know ordqω = 0 if q 6= ∞ and ordqω = −2 if q = ∞. Hence we have
deg(η) = X
p∈X, f (p)6=∞
(ep(f ) − 1) + X
p∈X, f (p)=∞
(−ep(f ) − 1)
= X
p∈X, f (p)6=∞
(ep(f ) − 1) − X
p∈X, f (p)=∞
(ep(f ) + 1)
= X
p∈X, f (p)6=∞
(ep(f ) − 1) + X
p∈X,f (p)=∞
(ep(f ) − 1) − X
p∈X,f (p)=∞
(ep(f ) − 1) − X
p∈X, f (p)=∞
(ep(f ) + 1)
=X
p∈X
(ep(f ) − 1) − 2 X
p∈f−1(∞)
ep(f )
=X
p∈X
bp(f ) − 2d
= 2g − 2 + 2d − 2d
= 2g − 2.
Since any two canonical divisors have the same degree, we obtain that deg KX = 2g − 2. Let f : X → Y be a nonconstant holomorphic mapping between Riemann surfaces. Given a point q ∈ Y, we define the pull back divisor f∗(q) by
f∗(q) = X
p∈f−1(q)
ep(f )p.
In general, given a divisor D =P
q∈Y nqq, the pull back divisor is defined to be f∗(D) =X
q∈Y
nqf∗(q).
Lemma 1.4. Let f : X → Y be a nonconstant holomorphic map between Riemann surfaces.
(1) f∗: Div(Y ) → Div(X) is a group homomorphism.
(2) The pull back of principal divisor is principal. In fact, if h is a meromorphic function on Y, then f∗(h) = (h ◦ f ).
(3) If X and Y are compact, so that divisors have degrees, we have deg f∗D = deg f · deg D.
Proof. It follows from the definition that f∗is a group homomorphism. Let h be a meromor- phic function on Y. Then h ◦ f is a meromorphic function on X. Write (h) =P
q∈Y(ordqh)q.
Then
f∗(h) = X
q∈Y
(ordqh)f∗q = X
q∈Y
X
p∈f−1(q)
(ordqh)ep(f )p
= X
p∈X
ordp(h ◦ f )p = (h ◦ f ).
Here we use the fact that ep(h ◦ f ) = ep(f ) ordf (p)h. Thus (b) is proved.
Since f∗ is linear, if D =P
q∈Y nqq, then f∗D = X
q∈Y
nqf∗q =X
q∈Y
X
p∈f−1(q)
nqep(f )p.
Hence
deg f∗D =X
q∈Y
X
p∈f−1(q)
nqep(f ) =X
q∈Y
nq
X
p∈f−1(q)
ep(f )
. Since P
p∈f−1(q)ep(f ) = deg f, we find
deg f∗D = deg f ·X
q∈Y
nq = deg f · deg D.
If f : X → Y is a nonconstant holomorphic mapping, the ramification divisor of f is the divisor on X defined by
Rf =X
p∈X
bp(f )p.
The branched divisor of f is the divisor on Y defined by Bf =X
q∈Y
X
p∈f−1(q)
bp(f )
q.
Proposition 1.2. Let f : X → Y be nonconstant holomorphic map between Riemann surfaces. Let ω be a nonzero merormorphic one forms on Y. Then
(f∗ω) = f∗(ω) + Rf.
Proof. The proof of this equation is based on the fact that
ordpf∗ω = (1 + ordf (p)ω)ep(f ) − 1 = (ordf (p)ω)ep(f ) + (ep(f ) − 1).
We obtain
(f∗ω) = X
p∈X
(ordpf∗ω)p
= X
p∈X
(ordf (p)ω)ep(f )p +X
p∈X
(ep(f ) − 1)p
=X
q∈Y
X
p∈f−1(q)
(ordqω)ep(f )p + Rf
=X
q∈Y
(ordqω)f∗q + Rf
= f∗(ω) + Rf.
Corollary 1.1. As above, when X and Y are compact, then
2g(X) − 2 = (deg f )(2g(Y ) − 2) + deg(Rf), where g(X) and g(Y ) are genus of X and Y respectively.
Proof. Computing the degree of (f∗ω) and using deg(ω) = 2g(Y ) − 2, we find deg(f∗ω) = deg f∗(ω) + deg Rf
= deg f · deg ω + deg Rf
= (deg f )(2g(Y ) − 2) + deg Rf.
Since ω is a meromorphic one form on Y, f∗ω is a meromorphic one form on X. Since the degree of any canonical divisor on X has degree 2g − 2, deg(f∗ω) = 2g − 2. We prove our
assertion.
Usually we denote deg Rf by B. Corollary 1.1 gives us the second proof of the Riemann- Hurwitz formula:
g(X) = (deg f )(g(Y ) − 1) + 1 + B 2.