33 (20160722) Sect. 5 Sect.

5 (20160722) 34

5 The Weierstrass representation Holomorphic differential forms on Riemann surfaces.

Let M 2 be a Riemann surface, i.e., a 1-dimensional complex
Complex Analysis again. A holomorphic function f on a manifold and let {(Uα , zα )} be a complex atlas of it. A notion
domain D ⊂ C is said to be having an isolated singularity at p of holomorphic function on M 2 is defined in a usual way: a
if there exists a neighborhood Up of p such that Up ⊂ D. function f : M 2 → C is holomorphic if f |Uα is holomorphic in
Fact 5.1 (The Laurent expansion). For a holomorphic func- zα for each α. A meromorphic function on M 2 is a holomorphic
tion f having an isolated singularity at p, there exists a positive function on M 2 \ Σ, where Σ is a discrete subset of M 2 , such
number ε and complex numbers an (n ∈ Z) such that that each point p ∈ Σ is at most a pole of f |Uα for a chart Uα
containing p. The order of a pole at p ∈ Σ is defined as the order
+∞
∑ of f |Uα at p, which does not depend on a choice of coordinates.
(5.1) f (z) = an (z − p)n (Dp,ε := {z ; 0 < |z − p| < ε}). A form “f (z) dz” on a local complex chart (U, z) is called
n=−∞
a holomorhpic 1-form if f (z) is a holomorphic function in z.
The convergence of the right-hand side is uniform on any com- A holomorphic 1-form on M 2 is a collection of holomorhpic 1-
pact subset of Dp,ε forms {fα dzα } satisfying the compatibility
Definition 5.2. The coefficient a−1 in (5.1) is called the residue dzα
of f at an isolated singularity p, and denoted by fα (zα ) = fβ (zβ (zα )) .
dzβ
Res f (z) := a−1 .
z=p The collection {fα dzα } is a meromorphic 1-form if each fα is
Definition 5.3. An isolated singularity p of holomorphic func- meromorphic.
tion f is a pole of (at most) order k if a−m = 0 holds in (5.1) Definition 5.5. Let ω = {fα dzα } be a meromorphic 1-form on
for m > k. If {m ; am ̸= 0} is unbounded, p is said to be an M 2 and p a pole of ω. The residue of ω at p is defined as
essential singularity.
Proposition 5.4. If p is a pole of order at most k of a holo- Res ω := Res fα (zα ),
p zα =p
morphic function f , the residue is computed as
were (Uα , zα ) is a complex chart around p.
1 dk−1 { }
Res f (z) = lim k−1 (z − p)k f (z) .
z=p (k − 1)! z→p dz Remark 5.6. The definition of the residue does not depend on a
22. July, 2016. choice of coordinate charts.
35 (20160722) Sect. 5 Sect. 5 (20160722) 36

Let C be a curve in a coordinate neighborhood (U, z) of M 2 , is a holomorphic map satisfying (4.2) and (4.3), cf., Proposi-
and z = z(t) (a ≦ t ≦ b) is a parametrization of it. Then the tion 4.4, where (U, z) is a complex coordinate chart.
integration of a holomorphic 1-form f (z) dz on U is defined as Though ϕ depends on a choice of coordinate charts,
∫ ∫ b (5.4) ϕ̂ := ϕ(z) dz
( ) dz(t)
(5.2) f (z) dz = f z(t) dt. does not depend on coordinates. In fact, if one take another
C a dt
complex coordinate chart (V, w),
Noticing that this definition does not depend on coordinate ∂f ∂f dz ∂f
charts and parametrizations of C, one can define the line in- dw = dw = dz.
∂w ∂z dw ∂z
tegral of a holomorphic 1-form ω on M 2 along a curve C. The
following is a corollary of Cauchy’s theorem of complex integra- Proposition 5.8 (The Weierstrass representation). For a con-
tions: formal minimal immersion f : M 2 → R3 of a Riemann surface
M 2 , there exists a meromorphic function g and a holomorphic
Fact 5.7 (The residue principle). Let C is a closed curve of M 2 1-form ω on M 2 such that, up to translations in R3 ,
which bound a simply connected domain D ⊂ M 2 , and ω be a ∫
( )
meromorhpic 1-form on a neighborhood of D ∪ C which have the (5.5) f (z) = Re (1 − g 2 ), i(1 + g 2 ), 2g ω
Cz
only pole p ∈ D. Then
∫ holds, where Cz is a path on M 2 joining a base point z0 and z.
ω = 2πi Res ω. Proof. Define ϕ = (ϕ1 , ϕ2 , ϕ3 ) as in (5.3). If ϕ1 − iϕ2 is equiva-
C p
lently zero, ϕ3 = 0 because of (4.2). In this case, the surface is
a horizontal plane, and g = 0, ω = a dz satisfy the conclusion.
The Weierstrass representation formula. Let M 2 be an ϕ3
Otherwise, let g := ϕ1 −iϕ and ω = ϕ1 − iϕ2 .
2
orientable manifold and f : M 2 → R3 be an immersion. By Since g does not depend on a choice of complex charts, g is
Corollary 3.11, there exists a structure of Riemann surface on a meromorphic function on M 2 . On the other hand, by (5.4)
M 2 such that any complex coordinate is isothermal. So, without does not depend on coordinates, ω can be considered as a holo-
loss of generality, we may assume that M 2 is a Riemann surface morphic 1-form on M 2 . By (4.2), we have
and f is a conformal immersion. Moreover, if f is minimal,
0 = ϕ̂21 + ϕ̂22 + ϕ̂23 = (ϕ̂1 − iϕ̂2 )(ϕ̂1 + iϕ̂2 ) + ϕ23
∂f
(5.3) ϕ := : U −→ C3 = (ϕ̂1 − iϕ̂2 )ω + g 2 ω 2 ,
∂z
37 (20160722) Sect. 5 Sect. 5 (20160722) 38

which (implies ϕ̂1 − iϕ̂2 = −g)2 ω, where ϕ̂ = ϕ dz. Hence we have Theorem 5.10 (The Weierstrass representation). Let M 2 be a
ϕ̂ = 12 (1 − g 2 ), i(1 + g 2 ), 2g ω. Equation 5.5 holds because simply connected Riemann surface, and let g and ω be a pair
∫ of a meromorphic function and a holomorphic 1-form on M 2
∂ ( )
(5.6) F (z) := ϕ̂, then F (z) + F (z) = ϕ̂. such that ds2 in (5.7) is positive definite4 . Then (5.5) gives a
Cz ∂z minimal immersion.
Proof. The integration (5.6) does not depend on a choice of
Corollary 5.9. Let f be as in (5.5), the first fundamental form paths Cz , and then it gives a map F : M 2 → C3 .
ds2 , the unit normal vector field ν, and the second fundamental
form II are expressed as
Examples.
(5.7) ds2 = (1 + |g|2 )2 |ω|2 , Example 5.11. Let M 2 = C, (g, ω) = (z, dz). Then
1 ( ) ∫
(5.8) ν= 2 Re g, 2 Im g, |g|2 − 1 = π −1 (g), ( )
1 + |g|2 f : = Re 1 − z 2 , i(1 + z 2 ), 2z dz
(5.9) II = −ω dg − ω dg, ( )
u3 2 2 v3 2 2
= u− + uv , −v − u v + , u − v
where π : S → C ∪ {∞} is the stereographic projection.
2
3 3
Proof. Let z = u + iv be a complex coordinate. Then by the is a minimal surface, where z = u + iv (Figure 3,left). This
proof of (4.3), ds2 = E(du2 + dv 2 ) = E dz dz̄, where E = surface is known as Enneper’s surface.
2(|ϕ1 |2 + |ϕ2 |2 + |ϕ3 |2 ) proving the first assertion. The second
Example 5.12. Let M 2 = C \ {0} (not simply connected) and
assertion was the homework 4-1. The third assertion follows
set (g, ω) = (z, i dz/z 2 ). Then f in (5.5) is represented by
since the second fandamental form is expressed as
(( ) ( ) )
1 1
II = (fzz · ν) dz 2 + 2(fzz̄ · ν) dz dz̄ + (fz̄z̄ · ν) dz̄ 2 . f= r− sin θ, r − cos θ, −2θ (z = reiθ )
r r
which is not well-defined on M 2 but on the universal cover of
As seen in Corollary 5.9, the meromorphic function g : M 2 → M 2 . The surface is congruent to the helicoid (Example 5.12).
C∪{∞} can be identified with ν via the streographic projection. 4 This condition is equivalent that the set of the zeros of ω is the set of
So we call g the Gauss map of f . poles of g, and for each pole p of g, the order of the pole p of g is exactly
The following is the converse assertion of Proposition 5.8. half of the order of zero of ω.
39 (20160722) Sect. 5 Sect. 5 (20160722) 40

holds for all loops γ on M 2 . Then f in (5.5) is well-defined on

M 2 and gives a minimal immersion of M 2 into R3 .
Corollary 5.15. Let M 2 = C ∪ {∞} \ {p1 , . . . , pn }, and (g, ω)
as in Proposition 5.14, and assume
( )
Im Res 1 − g 2 , i(1 + g 2 ), 2g ω = 0 (j = 1, . . . , n).
pj

Then f as in (5.5) is a minimal immersion defined on M 2 .

Example 5.16. Let n ≧ 2 be an integer, and
Example 5.11 Example 5.16 (n = 3) M 2 = C ∪ {∞} \ {1, ζ, . . . , ζ n−1 }, ζ = e2πi/n .
( )
Figure 3: Examples of minimal surfaces.
Then (g, omega) = z n−1 , (zndz −1) 2 satisfies the assumptions
of Corollary 5.15, and hence there exists a minimal immersion
Example 5.13. Let M 2 = C \ {0} and set (g, ω) = (z, dz/z 2 ). f : M 2 → R3 with (g, ω). Such a series of minimal surfaces are
Then f in (5.5) is represented by, with z = reiθ , called the Jorge-Meeks’ surfaces.
( ( ) ( ) )
1 1
f = − r+ cos θ, − r + sin θ, 2 log r : M 2 → R3 , References
r r
[5-1] R. Osserman, A survey of minimal surfaces, Dover Publ.
which is the catenoid (Example 5.13). [5-2] L. P. Jorge and W. H. Meeks, III, The topology of complete minimal
The phenomenon as in Example 5.13 is generalized as surfaces of finite total Gaussian curvature, Topology 22 (1983), 203–
221.
Proposition 5.14. Let M 2 be a (not necessarily simply con-
nected) Riemann surface, and let (g, ω) be a pair of a meromor- Exercises
phic function and a holomorphic 1-form on M 2 such that ds2
in (5.7) is positive definite. Assume 5-1H Verify Example 5.16 for n = 3.
∫
( )
Re 1 − g 2 , i(1 + g 2 ), 2g ω = 0
γ