Page 284 -
P. 284

§3. Definition of Supersingular Elliptic Curves  261

                                                                        2
        (3.5) Proposition. In characteristic p > 2 the curve in Legendre form E λ : y =
        x(x − 1)(x − λ) is supersingular if and only if λ is a root of the Hasse invariant
                                       m 2  i            p − 1
                                   m

                                 m
                     H p (λ) = (−1)        λ ,  where m =     .
                                       i                   2
                                  i=0
        Proof. We must calculate the coefficient of (wxy) p−1  in f  p−1  where
                                       2
                          f (w, x, y) = wy − x(x − w)(x − λw).
        In f (w, x, y) p−1  the term (wxy) p−1  will appear only in the middle of the binomial
                             p−1    2 m                m             p−1
        expansion, i.e., the term (  )(wy ) [x(x − w)(x − λw)] . Observe that (  ) =
                              m                                       m
            m
                                                        p
                                                             p
        (−1) (mod p) from the relation (x + y) p−1 (x + y) = x + y (mod p). Since
                                                                   m
        2m = p − 1, the coefficient of (wxy) p−1  is just the coefficient of (xw) in (x −
                    m
          m
                               m
        w) (x − λw) up to (−1) and by Lemma (3.4) this is H p (λ) which proves the
        proposition.
           In Chapter 9 we saw how the hypergeometric function F(1/2, 1/2, 1; λ) played
        a basic role in describing the period lattice of an elliptic curve over C. This same
        function is related to the Hasse invariant H p (λ). To see this, observe that the coeffi-
        cients

                         1    3
                 1     −    −    ...(1 − 2k/2)       k

               −         2    2                  −1   1 · 3 · 5 ...(2k − 1)
                 2
                    =                        =
                k               k!               2           k!
                               k
                                  2k
                             1
                                                1
                    = (−1) k            are in Z   .
                             2     k            2
        Hence reducing modulo p, we see that the formal series F(1/2, 1/2, 1; λ) in F p [[λ]]
        is defined, and moreover we denote by
                               m 2         1 1
                           m
                                    i                          p
                   G p (λ) =       λ ≡ F    , , 1; λ   mod (p,λ )
                                i          2 2
                           i=0
        The polynomial G p (λ) is called the Deuring polynomial. Since for m = (p − 1)/2,
        we have the congruence
           m      1   (p − 1)(p − 3)...(p − 2k + 1)  −1   1 · 3 · 5 ··· (2k − 1)
         
          k                                    k
              =                                  ≡
           k      2               k!                 2           k!

                    1
                  − 2
              =          (mod p) for k < p.
                   k
        (3.6) Remark. In the ring of formal series F p [[λ]] the following relation holds:
                         1 1                    p       p 2

                      F   , , 1; λ = G p (λ) · G p (λ ) · G p (λ )... .
                         2 2
   279   280   281   282   283   284   285   286   287   288   289