arXiv:1202.5922v2 [math.AG] 19 May 2013 Over all non-prime finite - PDF document

Towers of Function Fields over Non-prime Finite Fields Alp Bassa ∗ , Peter Beelen † , Arnaldo Garcia ‡ , and Henning Stichtenoth § Abstract arXiv:1202.5922v2 [math.AG] 19 May 2013 Over all non-prime finite fields, we construct some recursive towers of function fields with many rational places. Thus we obtain a substantial improvement on all known lower bounds for Ihara’s quantity A ( ℓ ), for ℓ = p n with p prime and n > 3 odd. We relate the explicit equations to Drinfeld modular varieties. 1 Introduction Investigating the number of points on an algebraic curve over a finite field is a classical subject in Number Theory and Algebraic Geometry. The origins go back to Fermat, Euler and Gauss, among many others. The basic result is A. Weil’s theorem which is equivalent to the validity of Riemann’s Hypothesis in this context. New impulses came from Goppa’s construction of good codes from curves with many rational points, and also from applications to cryptography. For information, we refer to [9, 21]. One of the main open problems in this area of research is the determination of Ihara’s quantity A ( ℓ ) for non-square finite fields; i.e., for cardinalities ℓ = p n with p prime and n odd. This quan- tity controls the asymptotic behaviour of the number of F ℓ -rational points (places) on algebraic curves (function fields) as the genus increases. This is the topic of our paper. Let F be an algebraic function field of one variable over F ℓ , with the field F ℓ being algebraically closed in F . We denote N ( F ) = number of F ℓ -rational places of F, and g ( F ) = genus of F. The Hasse–Weil upper bound states that √ N ( F ) ≤ 1 + ℓ + 2 ℓ · g ( F ) . √ This upper bound was improved by Serre [23] who showed that the factor 2 ℓ can be replaced √ above by its integer part ⌊ 2 ℓ ⌋ . Ihara [16] was the first to realize that the Hasse–Weil upper bound becomes weak when the genus g ( F ) is large with respect to the size ℓ of the ground field F ℓ . He introduced the quantity ∗ Alp Bassa is supported by T¨ ubitak Proj. No. 112T233. Part of this work was done while Alp Bassa was at CWI, Amsterdam (Netherlands). † Peter Beelen gratefully acknowledges the support from the Danish National Research Foundation and the National Science Foundation of China (Grant No.11061130539) for the Danish-Chinese Center for Applications of Algebraic Geometry in Coding Theory and Cryptography. ‡ Arnaldo Garcia was partially supported by CNPq (Brazil), Sabancı University and T¨ ubitak (Turkey). § Henning Stichtenoth was partially supported by T¨ ubitak Proj. No. 111T234. 1

N ( F ) A ( ℓ ) = lim sup g ( F ) , (1) g ( F ) →∞ where the limit is taken over all function fields F/ F ℓ of genus g ( F ) > 0. By Hasse–Weil it holds √ √ that A ( ℓ ) ≤ 2 ℓ , and Ihara showed that A ( ℓ ) < 2 ℓ . The best upper bound known is due to Drinfeld–Vl˘ adut ¸ [2]. It states that √ A ( ℓ ) ≤ ℓ − 1 , for any prime power ℓ. (2) √ If ℓ is a square, the opposite inequality A ( ℓ ) ≥ ℓ − 1 had been shown earlier by Ihara using the theory of modular curves (see [15]). Hence √ A ( ℓ ) = ℓ − 1 , when ℓ is a square. (3) For all other cases when the cardinality ℓ is a non-square, the exact value of the quantity A ( ℓ ) is not known. Tsfasman–Vl˘ adut ¸–Zink [25] used Equation (3) to prove the existence of long linear codes with relative parameters above the Gilbert–Varshamov bound, for finite fields of square cardinality ℓ = q 2 with q ≥ 7. They also gave a proof of Equation (3) in the cases ℓ = p 2 or ℓ = p 4 with p a prime number. To investigate A ( ℓ ) one introduces the notion of (infinite) towers of F ℓ -function fields: F = ( F 1 ⊆ F 2 ⊆ F 3 ⊆ . . . ⊆ F i ⊆ . . . ) , where all F i are function fields over F ℓ , with F ℓ algebraically closed in F i , and g ( F i ) → ∞ as i → ∞ . Without loss of generality one can assume that all extensions F i +1 /F i are separable. As follows from Hurwitz’ genus formula, the limit below exists (see [6]) and it is called the limit of the tower F : N ( F i ) λ ( F ) := lim g ( F i ) . i →∞ Clearly, the limit of any tower F over F ℓ provides a lower bound for A ( ℓ ); i.e., for any F ℓ -tower F . 0 ≤ λ ( F ) ≤ A ( ℓ ) , So one looks for towers with big limits in order to get good lower bounds for Ihara’s quantity. Serre [23] used Hilbert classfield towers to show that for all prime powers ℓ , A ( ℓ ) > c · log 2 ( ℓ ) , with an absolute constant c > 0 . (4) One can take c = 1 / 96, see [21, Theorem 5.2.9]. When ℓ = q 3 is a cubic power, one has the lower bound A ( q 3 ) ≥ 2( q 2 − 1) , for any prime power q. (5) q + 2 When q = p is a prime number, this bound was obtained by Zink [26] using degenerations of modular surfaces. The proof of Equation (5) for general q was given by Bezerra, Garcia and Stichtenoth [1] using recursive towers of function fields; i.e., towers which are given in a recursive way by explicit polynomial equations. The concept of recursive towers turned out to be very fruitful for constructing towers with a large limit. An F ℓ -tower F = ( F 1 ⊆ F 2 ⊆ F 3 ⊆ . . . ) is recursively defined by f ( X, Y ) ∈ F ℓ [ X, Y ] when 2

(i) F 1 = F ℓ ( x 1 ) is the rational function field, and (ii) F i +1 = F i ( x i +1 ) with f ( x i , x i +1 ) = 0, for all i ≥ 1. For instance, when ℓ = q 2 is a square, the polynomial (see [6]) f ( X, Y ) = (1 + X q − 1 )( Y q + Y ) − X q ∈ F q 2 [ X, Y ] defines a recursive tower over F q 2 whose limit λ ( F ) = q − 1 attains the Drinfeld–Vl˘ adut ¸ bound. When ℓ = q 3 is a cubic power one can choose the polynomial (see [1]) f ( X, Y ) = Y q ( X q + X − 1) − X (1 − Y ) ∈ F q 3 [ X, Y ] (6) to obtain a recursive tower over F q 3 with limit λ ( F ) ≥ 2( q 2 − 1) / ( q + 2) ; this is the proof of Equation (5) above. The case q = 2 of Equation (6) is due to van der Geer–van der Vlugt [11]. In the particular case of a prime field, no explicit or modular tower with positive limit is known; only variations of Serre’s classfield tower method have been successful in this case, see [3, 21]. All known lower bounds for A ( p n ), with a prime number p and an odd exponent n > 3, are rather weak, see [18, 21]. For example, one has for q odd and n ≥ 3 prime (see [19]) 4 q + 4 A ( q n ) ≥ . (7) ⌊ 3+ ⌊ 2 √ 2 q +2 ⌋ ⌋ + ⌊ 2 √ 2 q + 3 ⌋ n − 2 The main contribution of this paper is a new lower bound on A ( p n ) that gives a substantial improvement over all known lower bounds, for any prime p and any odd n > 3. For large n and ¸ upper bound for A ( p n ). Moreover, small p , this new bound is rather close to the Drinfeld–Vl˘ adut it is obtained through a recursive tower with an explicit polynomial f ( X, Y ) ∈ F p [ X, Y ]. Our lower bound is: Theorem 1.1 Let p be a prime number and n = 2 m + 1 ≥ 3 an odd integer. Then A ( p n ) ≥ 2( p m +1 − 1) with ǫ = p − 1 p m − 1 . p + 1 + ǫ In particular A ( p 2 m +1 ) > p m − 1, which shows that a conjecture by Manin [10, 20] is false for all odd integers n = 2 m + 1 ≥ 3. The bound of Drinfeld–Vl˘ adu ¸ t can be written as A ( p 2 m +1 ) ≤ p m · √ p − 1 . Fixing the prime number p , we get ¸ bound → 2 √ p our lower bound p + 1 , as m → ∞ . Drinfeld–Vl˘ adut For p = 2 we have 2 √ p/ ( p + 1) ≈ 0 . 9428 . . . , hence our lower bound is only around 6 % below ¸ upper bound, for large odd-degree extensions of the binary field F 2 . the Drinfeld–Vl˘ adut Now we give the defining equations for the several recursive towers F that we consider in this paper. Let F ℓ be a non-prime field and write ℓ = q n with n ≥ 2. Here the integer n can be even or odd. For every partition of n in relatively prime parts, n = j + k with j ≥ 1 , k ≥ 1 and gcd( j, k ) = 1 , (8) 3