Page 127 -
P. 127

104    5. Reduction mod p and Torsion Points
                                             u
                                          n
                                     a = p ·  ,
                                             v
        where p does not divide either u or v and n is an integer uniquely determined by a.
        Let ord p (a) = n denote the order function associated with p. Let r p (a) =¯a denote
        the canonical reduction mod p defined R → k(p).If R (p) denotes the subring of
        all a in k with ord p (a) ≥ 0, then the mod p reduction function is well defined on
        R (p) → k(p).
           The order function satisfies the following properties:
        (V1)                   ord p (ab) = ord p (a) + ord p (b)
        and
        (V2)                 ord p (a + b) ≥ min{ord p (a), ord p (b)}.
        The second property can be refined in the case where ord p (a)< ord p (b) in which
        case ord p (a) = ord p (a + b). This refinement can be deduced directly from the
        second property of ord p (a) using an elementary argument. The function is a special
        case of a discrete valuation (of rank 1) and R (p) is the valuation ring associated to
        ord p . It is a principal ring with one irreducible p and one maximal ideal R (p) p where
        r p induces an isomorphism R (p) /R (p) p → k(p).
           There are some cases, especially over a number field, where it is confusing to use
        p for the irreducible or local uniformizing element of R, in those cases we will tend
        to use t or π.
           The reduction mod p function r p : R (p) → k(p) can be defined on affine space
        by taking products
                                  n
                                                 n
                                 k ⊃ R n  → k(p) ,
                                       (p)
        but it is only defined on the points x = (x 1 ,... , x n ) all of whose coordinates x j
        have positive ord p . Then the formula is r p (x 1 ,... , x n ) = (¯x 1 ,... , ¯x n ). On the other
        hand, reduction mod p is defined on the entire projective space.
        (1.2) Definition. The reduction mod p function r p : P n (k) → P n (k(p)) is defined
        by the relation
                            r p (y 0 : ··· : y n ) = ( ¯y 0 : ··· : ¯y n ),

        where (y 0 : ··· : y n ) is the homogeneous coordinates of a point in P n (k) with all
        y i in R and without a common irreducible factor. Such a representative of a point is
        called reduced.

           Observe that r p (y 0 : ··· : y n ) = ( ¯y 0 : ··· : ¯y n ) is defined when each ord p (y i ) ≥
        0 and some ord p (y j ) = 0 so that ¯y j  = 0. Such a representative (y 0 : ··· : y n )
        of a point in P n (k) is called p-reduced. The reduced representatives are unique up
        to multiplication by a unit in R, and the p-reduced representatives are unique up to
        multiplication by a unit in R (p) .
           We saw in Chapter 2 that a good intersection theory had to be formulated in pro-
        jective space. The above construction of mod p reduction shows another advantage
        of projective space over affine space which is arithmetic instead of geometric.
   122   123   124   125   126   127   128   129   130   131   132