Page 124 -
P. 124

§6. Examples of Noncyclic Subgroups of Torsion Points  101

        §6. Examples of Noncyclic Subgroups of Torsion Points

        Under the curve E[a, b] studied in the previous section, we look for examples of
        subgroups of E(k) of the form Z/nZ ⊕ Z/2Z. We start with the cases n = 2, 4, and
        8 always assuming that k is of characteristic different from 2.

        (6.1) Example. For n = 2wehave Z/2Z ⊕ Z/2Z ⊂ E(k) for E = E[a, b]ifand
                                   2                      2   3     2
        only if the quadratic polynomial x + ax + b factors, that is, y = x + ax + bx =
        x(x −r 1 )(x −r 2 ) and {0,(0, 0), (r 1 , 0), (r 2 , 0)} is the subgroup of 2-division points.
                            2
           Now assuming that x + ax + b factors,wegoonto n = 4 by applying 1(4.1) to
        see when there exists a point P ∈ E(k) with 2P = (0, 0).

        (6.2) Example. For n = 4wehave Z/4Z ⊕ Z/2Z ⊂ E(k) for E = E[a, b] =
                                                                   2    3
        E[−r 1 − r 2 ,r 1 r 2 ] with some P satisfying 2P = (0, 0) if and only if y = x +
                          2
                                 2
          2
        ax + bx = x(x + r )(x + s ),thatis, −r 1 and −r 2 are squares in k. The four
        solutions to the equation 2P = (0, 0) are (rs, ±rs(r − s)) and (−rs, ±rs(r + s)).
                                                                   2
           Assuming that 2P = (0, 0) has a solution as in (6.2), then 2P = (−r , 0) has a
                                                                  2
                                                       2
                                              2
                                                            2
        solution by (4.1) if and only if the elements −r − 0 =−r , −r − (−r ) = 0, and
           2
                 2
                           2
                       2
        −r − (−s ) = s − r are all squares in k. This cannot happen in Q or even R,as
        we know, but over Q(i), the Gaussian numbers, we need only choose s and r such
                     2
                 2
             2
        that s = r + t , and this takes us back to Pythagorean triples again.
           Continuing with the curve given in (6.2), we ask when is it true that
                 (rs, ±rs(r − s)) ∈ 2E(k)  or (−rs, ±rs(r + s)) ∈ 2E(k)?
        Again we apply 1(4.1) to see that 2P = (rs,rs(r − s)) has a solution if and only if
                                    2
                2
        rs,rs + r = r(r + s),and rs + s = s(r + s) are all three squares in k. This is the
                                                       2
                            2
                                                   2
                                              2
                                         2
                      2
        case when r = u , s = v ,and r + s = t for t = u + v .
                                                                        2 2
                                                                     2
                                                                 2
        (6.3) Example. For n = 8wehave Z/8Z⊕Z/2Z ⊂ E(k) for E = E[r +s ,r s ]
                                     2
                                2
                          2
                                          2
                   2
        when r = u , s = v , and u + v = t in k. This is equivalent to E is a curve
                     4
                          4
                             4 4
        of the form E[u + v , u v ]. In particular, over the rational numbers such curves
        correspond to Pythagorean triples (u,v, t).
           Applying 1(4.1) to the equation 2P = (−rs,rs(r + s)), we obtain almost the
                                                 2
                                     2
        same result, that is, −rs, −rs + r ,and −rs + s must be squares. This can be
                                           2
                                                    2
                                                2
                               2
                       2
        realized for r = u , −s = v ,and r − s = t = u + v .
           For the question of a point of order 3 on E[a, b] we recall from (5.3) that if
                                           2
                                               2
                                      2
        (x 2 , y 2 ) = 2(x 1 , y 1 ), then x 2 = (x − b) /4y . The case 2(x 1 , y 1 ) = (x 1 , −y 1 )
                                     1         1
                                                                 2
                                                         2
                                           2
                                                2
                                     2
        occurs for x = x 1 = x 2 , that is, 4xy = (x − b) .For x = t , y = st we see that
             3
                                   2
                              2
                                        3
        b = t (t − 2s) and from ax = y − x − bx, we obtain
                                  2    2           3
                        a = (s − t) − 3t  and b = t (t − 2s).
   119   120   121   122   123   124   125   126   127   128   129