Page 199 -
P. 199

176    9. Elliptic and Hypergeometric Functions

        is a relation between elliptic functions for the lattice L with all poles at points of L.
        The relation is an equality because the difference between the two sides is an elliptic
        function without poles which is zero at 0, and, hence, the difference is zero.

        (4.2) Differential Equation for ℘(z).

                                       3
                            2

                      ℘ (z, L) = 4℘(z, L) − g 2 (L)℘(z, L) − g 3 (L),

        where g 2 (L) = 60G 2 (L) and g 3 (L) = 140G 3 (L). Thus the points (℘(z), ℘ (z)) ∈
          2
        C lie on the curve defined by the cubic equation with g 2 = g 2 (L) and g 3 = g 3 (L)
                                  2    3
                                 y = 4x − g 2 x − g 3 .
        For a basis L = Zω 1 +Zω 2 and ω 3 = ω 1 +ω 2 , let e i = ℘(ω i /2), where i = 1, 2, 3.
        Then the elliptic function ℘(z) − e i = f (z) has a zero at ω i /2 which must be of
        even order by (2.5), and thus

                                     ω i        ω i
                                f      = ℘       = 0
                                    2        2
        for i = 1, 2, 3. By comparing zeros and poles, we obtain the factorization of (4.2):
                          2

                      ℘ (z) = 4 (℘(z) − e 1 )(℘(z) − e 2 )(℘(z) − e 3 ) ,
                                        3
        where e 1 , e 2 ,and e 3 are the roots of 4x − g 2 x − g 3 . Since ℘(z) takes the value e i
        with multiplicity 2 and has only one pole of order 2 modulo L, we see that e i  = e j
                                                             3
        for i  = j. We are led to the following result since the cubic 4x − g 2 x − g 3 has
        distinct roots.
        (4.3) Theorem. The function h : C/L → E(C), where E is the elliptic curve over
        C with equation
                              2
                                             2
                                   3
                           wy = 4x − g 2 (L)w x − g 3 (L)w 3
        and h(z mod L) = (1,℘(z), ℘ (z)) for z /∈ L, h(0 mod L)= (0, 0, 1), is an analytic

        groupisomorphism.
        Proof. Clearly h(z mod L) = (0, 0, 1) ∈ E(C) is the zero element of E(C) if and
                                                            3
                                                     3
                                                 3

        only if z mod L = 0 ∈ C/L,and h(z mod L) = (z : z ℘(z) : z ℘ (z)) is analytic
        at 0 ∈ C/L with values in the projective plane.
           To see that h : C/L − 0 → E(C) − 0 is an analytic isomorphism, we consider

        (x, y) ∈ E(C)−0. There are two zeros z 1 , z 2 of the function ℘(z)−x with ℘ (z 1 ) =

        −℘ (z 2 ) =±y from the equation of E. Thus h(−z mod L) =−h(z mod L) since

        z 1 +z 2 ∈ L by (3.5). Note further that z 1 ≡ z 2 mod L if and only if ±℘ (z 1 ) = y = 0.
        Hence h is an analytic isomorphism commuting with the operation of taking inverses.
           Finally, to see that h or equivalently h −1  preserves the group structure, we study
        the intersection points of the line y = λx + ν with the cubic curve E as in 1, §1. The

        elliptic function f (z) = ℘ (z) − (λ℘(z) + ν) has a pole of order 3 at each point of
   194   195   196   197   198   199   200   201   202   203   204