Page 196 -
P. 196

§3. The Weierstrass ℘-Function  173

        Proof. Every elliptic function f (z) is the sum of an even and an odd elliptic function

                                f (z) + f (−z)  f (z) − f (−z)
                         f (z) =            +             .
                                     2             2

        Since ℘ (z) times an odd elliptic function is an even elliptic function, it suffices to
        show that C(℘(z)) is the field of even elliptic functions.
           If f (z) is an even elliptic function, then ord 0 f (z) = 2m is even and f (z) =
        ℘(z) −m g(z), where g(z) is an even elliptic function with no zeros or poles on the
        associated lattice L.If a is a zero of ℘(z) − c,thensois ω − a for ω ∈ L,and if a is
        a zero or pole of g(z),thensois ω − a.If2a ∈ L, then the zero (or pole) is of order




        at least 2 since g (−z) =−g (z) and so g (a) = g (−a) =−g (a). Thus


                                         (℘(z) − ℘ (a i ))
                                        i
                             g(z) = c ·               ,
                                         (℘(z) − ℘ (b i ))
                                        i
        where {a i ,ω − a i } are the zeros of g(z) and {b i ,ω − b i } are the poles of g(z) in a
        fixed fundamental domain. This proves the theorem.
           Let M L denote the field of elliptic meromorphic functions for the lattice L. This
        can be viewed as the field of meromorphic functions on the complex torus C/L = T
        as in (2.7).
        (3.4) Definition. Adivisor D on C/L is a finite formal integral linear combination

        D =     n u (u) of points u in C/L, and its degree deg(D) =  n u is an integer.
               u                                             u
        The group Div(C/L) of all divisors on C/L is the group of all divisors. A principal

        divisor D is one of the form ( f ) =  ord u ( f )(u), where f is a nonzero meromor-
                                       u
        phic function in M L .WedenotebyDiv 0 (C/L) and Div p (C/L), respectively, the
        subgroups of divisors of degree zero and principal divisors.
           We have an exact sequence
                                                  deg
                         0 → Div 0 (C/L) → Div(C/L) → Z → 0,
        and Div p (C/L) is a subgroup of Div 0 (C/L) by (2.5).


        (3.5) Theorem (Abel–Jacobi). The function f (  n u (u)) =  n u · ufrom
                                                   u            u
        Div(C/L) to C/L induces a function also denoted f :Div 0 (C/L)/Div p (C/L) →
        C/L which is an isomorphism of groups.
        Proof. Since u = f ((u)−(0)) for any u in C/L, the group homomorphism is surjec-
        tive. The subgroup Div p (C/L) is carried to zero by f from (2.5). The construction

        following the proof of Theorem (3.3) above shows that  n u · u = 0in C/L is suf-
                                                      u
        ficient for the existence of an elliptic function f with ( f ) =  u  n u (u). This proves
        the theorem.
           Another proof that Div p (C/L) is thekernelof f results by using the σ-function
        introduced in Exercises 3, 4, and 5 at the end of this section.
   191   192   193   194   195   196   197   198   199   200   201