Page 151 -
P. 151

128    6. Proof of Mordell’s Finite Generation Theorem

        §3. Finiteness of (E E E(Q) :2E E E(Q)) for E E E = E E E[a a a,b b b]

        Using the factorization of multiplication by 2 on the elliptic curve E = E[a, b] con-
        sidered in 4(5.3), we will show by relatively elementary means that (E(Q) :2E(Q))
        is finite for an elliptic curve of this form over the rational numbers. Note E[a, b]is
                       3
                            2
                  2
        defined by y = x + ax + bx for a, b ∈ k.
                                                     ∗ 2
                                                  ∗
           Recall from 4(5.6) the function α : E[a, b] → k /(k ) defined by
                             α(0) = 1,
                                           ∗ 2
                         α((0, 0)) = b  mod(k ) ,
                                            ∗ 2
                         α((x, y)) = x  mod(k )   for x  = 0,
        and having the property, see 4(5.7), that the sequence
                                ϕ         2      α      ∗ 2
                                                    ∗
                         E[a, b] → E[−2a, a − 4b] → k /(k )
        is exact. In the case k = Q, the field of rational numbers, the quotient group
              ∗ 2
        Q /(Q ) is additively an F 2 -vector space with a basis consisting of −1 and all
          ∗
        the positive prime numbers p.
        (3.1) Proposition. Let E = E[a, b] be an elliptic curve over the rational numbers
        Q. The homomorphism
                                                 ∗ 2
                                             ∗
                                α : E[a, b] → Q /(Q )
        has image im(α) contained in the F 2 -vector subspace generated by −1 and all
        primes p dividing b. If r distinct primes divide b, then the cardinality of im(α) is
        less than 2 r+1 .
                         2
                                      3
        Proof. Let x = m/e and y = n/e be the coordinates of (x, y) on E[a, b] with
        rational coefficients where the representations are reduced to lowest terms. Note we
        can always choose denominators in this form. The equation of the curve gives
                                          2 2
                               2
                                    3
                              n = m + am e = bme   4
                                              2
                                                   4
                                       2
                                 = m(m + ame + be ).
                        2
                 2
                             4
        If m and m + ame + be are relatively prime, then it follows that m is a square up
        to sign and α(x, y) =±1.
                                                              2
                                                                     2
                                                                          4
           More generally, let d be the greatest common divisor of m and m +ame +be .
                      4
        Then d divides be , and since m and e are relatively prime, the integer d also divides
                         2
                                                           ∗ 2
                                                       ∗
        b. Moreover, m = M d up to sign, and α(x, y) = d mod Q /(Q ) . This proves the
        proposition.
                                                       4 2
                                                            2
        (3.2) Theorem. Let a and b be two integers with   = 2 b (a − 4b) unequal to
        zero, and let r be the number of distinct prime divisors of b and s the number of
         2
        a − 4b. Then for E = E[a, b] we have
                                                r+s+2
                               (E(Q) :2E(Q)) ≤ 2    .
   146   147   148   149   150   151   152   153   154   155   156