Abstract. We study the behaviour near s = 1 2 of zeta functions of varieties over finite fields Fq with q a square. The main result is an Euler-characteristic formula for the square of the special value at s = 1 2. The Euler-characteristic is constructed from the Weil-étale cohomology of a certain supersingular elliptic curve.
IntroductionLet V be a integral scheme of finite type over Spec Z. The special values a V (n) at integers s = n of the zeta function ζ(V, s) of V are conjecturally related to deep arithmetical invariants of V . One may ask if the special values of ζ(V, s) at non-integral values of s, e.g. s = 1 2 , also admit an arithmetical interpretation. The simplest example -and perhaps the most interesting -arises from the central value of the Riemann zeta function ζ(s) = ζ(Spec Z, s):Is there a motivic interpretation of ζ( The motivic philosophy indicates that a V (n) depends on the interaction of the motive h(V ) of V with Z(±n) (power of the Tate motive). This leads us to suspect that the special value at s = 1 2 is governed by an exotic object (unknown to exist, as yet): a square root Z( It is clear that the investigation of special values at s = 1 2 should begin with the important case of varieties over finite fields. Namely, consider the zeta function ζ(X, s) of a smooth projective variety X over a finite field k = F q of characteristic p > 0. The function Z(X, t), defined by Z(X, q −s ) := ζ(X, s), is a rational function of t with integer coefficients. For any integer n ≥ 0, the order of the pole ρ n := −ord s=n ζ(X, s) at s = n and the special value a X (n) of Z(X, t) at t = q −n conjecturally admit a motivic interpretation.The Tate conjecture (Conjecture 2) predicts thatin the category of motives and h 2n (X) is part of the motive of X. A related variant is that ρ n is the rank of the Chow group CH n (X) of algebraic cycles of codimension n on X. The Lichtenbaum-Milne conjecture (Conjecture 3) expresses a X (n) as an Euler-characteristic ofétale motivic cohomology H * (X, Z X (n)); i.e., the cohomology of the (étale) motivic complexes Z X (n) of S. Lichtenbaum; Z X (0) is the constant sheaf Z and Z X (1) = G m [−1] is the sheaf G m in degree one. Their conjecture is known for n = 0 (unconditionally) and for n = 1 (modulo the Tate conjecture for divisors on X); cf. [11, 19]. For n = 0, 1, it takes the formLichtenbaum [12] has provided another elegant interpretation of a X (0) using his Weil-étale topology (cf. Theorem 5).Let us now turn to the special value at s = 1 2 or t = 1/ √ q. One can ask for a motivic description of • the order of vanishing ρ X := ord s= 1 2 ζ(X, s) and • the corresponding special value c X at s = 1 2 , viz., c X := lim t→1/ √ q (1 − √ qt) −ρX Z(X, t). The main result of this paper provides such a description, under the condition that q = p 2f , i.e., that F p 2 ⊂ F q . We note that c X may not be rational, if the condition on q is dropped.Our paper is an exploration, using the methods of [11,19,12,21], of a beautiful suggestion of Yuri Manin that "a certain supersingu...