Page 116 -
P. 116

§4. Tate’s Normal Form for a Cubic with a Torsion Point  93

        In particular, (0, 0) is not a point of order 2. By (2.2) the point (0, 0) has order 3 if
        and only if a 2 = 0and a 3  = 0.
           Assume that (0, 0) is not of order 2 or 3, that is , both a 2 and a 3 are each unequal
                             2
                                        3
        to 0. By changing x to u x and y to u y, we can make a 3 = a 2 =−b, and we
        obtain the following form of the equation of the curve which now depends on two
        parameters. Observe that all of the above discussion is carried out in a given field k,
        and the changes of variable did not require field extensions.
        (4.1) Definition. The Tate normal form of an elliptic curve E with point P = (0, 0)
        is
                                   2                   3    2
                      E = E(b, c) : y + (1 − c)xy − by = x − bx .
           Further a calculation, using the relations in 3(3.3), yields the following formula
        for the discriminant   =  (b, c) of E(b, c):
                                                           2 4
                                               3 3
                                    4 3
                       (b, c) = (1 − c) b − (1 − c) b − 8(1 − c) b
                                           4      4     5
                                + 36(1 − c)b − 27b + 16b .
        (4.2) Remark. The Tate normal form gives a description in terms of equations for
        the set of pairs (E, P) consisting of an elliptic curve E together with a point P on E
        where P,2P, and 3P are all unequal to zero. These pairs correspond to pairs (b, c)
        with both b  = 0and  (b, c)  = 0. The curve in the family corresponding to (b, c)
        is E(b, c) and the point is P = (0, 0). In the two-parameter Tate family E(b, c)
        there will be many cases where curves in different fibres E(b, c) are isomorphic, for
        example, E(b, 1) and E(b, −1) are isomorphic curves.
           If we require, further, that nP = 0 for some integer n > 3, then there is a
        polynomial equation f n (b, c) = 0 over Z which b and c must satisfy. The result is
        that the relations

                       T n : f n (b, c) = 0,  b  = 0,   (b, c)  = 0
        define an open algebraic curve with a family E(b, c) of elliptic curves over it together
        with a given n division point P, i.e., a cross-section P in the family of order n.We
        will determine f n explicitly in several cases, see (4.6).

        (4.3) Remarks. Since this family contains, up to isomorphism, all elliptic curves E
        with torsion point P of order n, the curve T n maps onto the open curve Y 1 (n) which
        is the parameter space for isomorphism classes of pairs (E, P) of elliptic curves E
        together with a point P of order n. The curve Y 1 (n) has a completion X 1 (n) which
        is nonsingular where the completing points, called cusps, correspond to degenerate
        elliptic curves. We return to this later.
        (4.4) Remark. There is an elliptic curve E over a field k with torsion point P of
        order n over the field k if and only if the open algebraic curve T n has k rational points,
        or in other words, the set T n (k) is nonempty. This is equivalent to the statement
        that Y 1 (n)(k) is nonempty. This is also equivalent to X 1 (n)(k) having noncuspidal k
        rational points. For further discussion of Y 1 (n) and X 1 (n) see Chapter 11, §2and §3.
   111   112   113   114   115   116   117   118   119   120   121