Finite Fields and Their Applications 14 (2008) 721 726 http://www.elsevier.com/locate/ffa On the elliptic curve analogue of the sum-product problem Igor Shparlinski Department of Computing, Macuarie University, Sydney, NSW 2109, Australia Received 25 September 2007; revised 11 December 2007 Available online 10 January 2008 Communicated by Gary L. Mullen Abstract Let E be an elliptic curve over a finite field F of elements and x(p) to denote the x-coordinate of a point P = (x(p ), y(p )) E. Let denote the group operation in the Abelian group E(F ) of F -rational points on E. We show that for any sets R, S E(F ) at least one of the sets { x(r) + x(s): R R, S S } and { x(r S): R R, S S } is large. This uestion is motivated by a series of recent results on the sum-product problem over F. 2008 Elsevier Inc. All rights reserved. Keywords: Sum-product problem; Elliptic curves; Character sums 1. Introduction We fix a finite field F of elements and an elliptic curve E over F givenbyanaffine Weierstraß euation with some a 1,...,a 6 F, see [11]. E: y 2 + (a 1 x + a 3 )y = x 3 + a 2 x 2 + a 4 x + a 6, E-mail address: igor@comp.m.edu.au. 1071-5797/$ see front matter 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.ffa.2007.12.002
722 I. Shparlinski / Finite Fields and Their Applications 14 (2008) 721 726 We recall that the set of all points on E forms an Abelian group, with the point at infinity O as the neutral element, and we use to denote the group operation (we also use in its natural meaning). As usual, we write every point P O on E as P = (x(p ), y(p )). Let E(F ) denote the set of F -rational points on E. We show that for any sets R, S E, at least one of the sets U = { x(r) + x(s): R R, S S }, V = { x(r S): R R, S S }, (1) is large. This uestion is motivated by a series of recent results on the sum-product problem over F which assert that for any sets A, B F, at least one of the sets G ={a + b: a A, b B} and H ={ab: a A, b B} is large, see [1 8] for the background and further references. In fact, our approach is a combination of the argument of M. Garaev [4] and an estimate of [10] of certain bilinear character sums over points of E(F ). We recall the idea of [4] (extended to the case of two distinct sets instead of just A = B as in [4]) to obtain upper and lower bounds on the number of solutions (a 1,a 2,g,h)to the euation ha 1 1 + a 2 = g, a 1,a 2 A, g G, h H. (2) This euation obviously has at least (#A) 2 #B as the solutions of the form (a 1,a 2,a 2 + b,a 1 b), a 1,a 2 A, b B. There are now several ways to get the upper bound (#A) 2 #G#H/ + O( 1/2 #A(#G#H) 1/2 ) on the number of solutions. For example, one can simply use the result of A. Sárközy [9] (see also [4]). Throughout the paper, the implied constants in the symbols O and may depend on an integer parameter ν 1. We recall that X Y and X = O(Y) are both euivalent to the ineuality X cy with some constant c>0. 2. Sum-product estimate for elliptic curves Theorem 1. Let R and S be arbitrary sets of E(F ). Then for the sets U and V, given by (1),we have Proof. Let #U#V min { #R,(#R) 2 #S 1/2}. W ={R S: R R, S S}. Following the idea of M. Garaev [4], we now denote by J the number of solutions (S 1,S 2,W,u) to the euation x(w S 1 ) + x(s 2 ) = u, S 1,S 2 S, W W, u U. (3)
I. Shparlinski / Finite Fields and Their Applications 14 (2008) 721 726 723 Since obviously the vectors (S 1,S 2,W,u)= ( S 1,S 2,R S 1,x(R)+ x(s 2 ) ), R R, S 1,S 2 S, are all pairwise distinct solutions to (3), we obtain J #R(#S) 2. (4) To obtain an upper bound on J we use Ψ to denote the set of all additive characters of F and write Ψ for the set of nontrivial characters. Using the identity 1 { 0, if z F ψ(z)=, 1, if z = 0, (5) we obtain J = 1 ψ ( x(w S 1 ) + x(s 2 ) u ) S 1 S S 2 S W W = 1 ψ ( x(w S 1 ) ) ψ ( x(s 2 ) ) ψ( u) S 1 S W W S 2 S = (#S)2 #U#W + 1 S 1 S W W ψ ( x(w S 1 ) ) ψ ( x(s 2 ) ) ψ( u). We now recall the main result of [10]. Let T ρ,ϑ (ψ, P, Q) = ρ(p)ϑ(q)ψ ( x(p Q) ), (6) P P Q Q where P, Q E(F ), ρ(p) and ϑ(q) are arbitrary complex functions supported on P and Q with ρ(p) 1, P P, and ϑ(q) 1, Q Q, and ψ is a nontrivial additive character of F. Note that we always assume that the values for which the corresponding summation term is not defined (that is, the terms with P = Q) are excluded from the summation. For the sum T ρ,ϑ (ψ, P, Q) given by (6) it is shown in [10] that for any fixed integer ν 1we have S 2 S T ρ,ϑ (ψ, P, Q) (#P) 1 1/(2ν) (#Q) 1/2 1/(2ν) + (#P) 1 1/(2ν) #Q 1/(4ν). (7) In our case (taking this bound with ν = 1, ρ(p) = ϑ(q)= 1, P = W, Q = S) it implies that ψ ( x(w S 1 ) ) (#W) 1/2 (#S) 1/2 1/2 + (#W) 1/2 #S 1/4. S 1 S W W
724 I. Shparlinski / Finite Fields and Their Applications 14 (2008) 721 726 Therefore, J (#S)2 #U#W ( (#W) 1/2 (#S) 1/2 1/2 + (#W) 1/2 #S 1/4) 1 ψ ( x(s) ) ψ(v). (8) S S Extending the summation over ψ to the full set Ψ and using the Cauchy ineuality, we obtain ψ ( x(s) ) ψ(u) ψ ( x(s) ) 2 2 ψ(u). (9) S S Recalling the orthogonality property (5), we derive ψ ( x(s) ) 2 = # { (S 1,S 2 ) S 2 : x(s 1 ) = x(s 2 ) } 2#S, S S since as it immediately follows from the Weierstraß euations, the curve E contains at most 2 points with a given value of the x-coordinate. Similarly, ψ(v) #U. S S Substituting these bounds in (9) we obtain ψ ( x(s) ) ψ(v) #S#U, S S which after inserting in (8), yields J (#S)2 #U#W (#W#U) 1/2( #S 1/2 + (#S) 3/2 1/4). (10) Thus, comparing (4) and (10), we derive (#S) 2 #U#W + (#W#U) 1/2( #S 1/2 + (#S) 3/2 1/4) #R(#S) 2. Since there are at most two points P E(F ) with the same value of x(p), we see that #V 0.5#W. Hence, #U#V min { #R,(#R#S) 2 1,(#R) 2 #S 1/2}.
I. Shparlinski / Finite Fields and Their Applications 14 (2008) 721 726 725 It remains to remark that since we obviously have min{#u, #V} #R the bound of the theorem is nontrivial only if #S 1/2, in which case (#R#S) 2 1 (#R) 2 #S 1/2, which concludes the proof. Corollary 2. Let R and S be arbitrary sets of E(F ) with 1 ε #R #S 1/2+ε. Then for the sets U and V, given by (1), we have #U#V (#R) 2 ε. Corollary 3. Let R = S be an arbitrary set of E(F ) with #R 3/4. Then for the sets U and V, given by (1), we have 3. Comments #U#V #R. It seems that the bound (7) is useful for our purpose only when it is taken with ν = 1. However, several other euations, besides (2), have been used to obtain various lower bounds in the sumproduct problem over finite fields. One can certainly try to use their analogues for its elliptic curve version which we have considered in the present paper. Possibly for some of them one can make use of (7) in its full generality. Finally, as in [4] we can justify that the uantity #R should be present in any lower bound on #U#V. Indeed, let = p be prime and let E be a curve over F p such that E(F p ) is a cyclic group (see [12] on statistics of cyclic elliptic curves) and let G be a generator. Let N min{#e(f p ), p} be any positive integer. We now take M = (pn) 1/2. Then the pigeonhole principle implies that for some integer L there is a set M {1,...,M} of cardinality #M M 2 /p N such that x(mg) {L + 1modp,...,L+ M mod p}, m M. We not take any subset N M of cardinality N and put R = S = ng, n N. Clearly, for the sets U and V, given by (1) we have Acknowledgment max{#u, #V} M (p#r) 1/2. This work was supported in part by ARC grant DP0556431.
726 I. Shparlinski / Finite Fields and Their Applications 14 (2008) 721 726 References [1] J. Bourgain, A.A. Glibichuk, S.V. Konyagin, Estimates for the number of sums and products and for exponential sums in fields of prime order, J. Lond. Math. Soc. 73 (2006) 380 398. [2] J. Bourgain, N. Katz, T. Tao, A sum product estimate in finite fields and applications, Geom. Funct. Anal. 14 (2004) 27 57. [3] M.Z. Garaev, An explicit sum-product estimate in F p, Int. Math. Res. Not. 2007 (2007), article ID rnm035, 1 11. [4] M.Z. Garaev, The sum-product estimate for large subsets of prime fields, preprint, available from http://arxiv.org/ abs/0706.0702, 2007. [5] D. Hart, A. Iosevich, D. Koh, M. Rudnev, Averages over hyperplanes, sum-product theory in finite fields, and the Erdös Falconer distance conjecture, preprint, available from http://arxiv.org/abs/0707.3473, 2007. [6] D. Hart, A. Iosevich, J. Solymosi, Sums and products in finite fields via Kloosterman sums, Int. Math. Res. Not. 2007 (2007), article ID rnm007, 1 14. [7] N.H. Katz, C.-Y. Shen, A slight improvement to Garaev s sum product estimate, Proc. Amer. Math. Soc., in press. [8] N.H. Katz, C.-Y. Shen, Garaev s ineuality in finite fields not of prime order, J. Analytic Combinatorics 3 (2008), Article #3. [9] A. Sárközy, On sums and products of residues modulo p, Acta Arith. 118 (2005) 403 409. [10] I.E. Shparlinski, Bilinear character sums over elliptic curves, Finite Fields Appl. 14 (1) (2008) 132 141. [11] J.H. Silverman, The Arithmetic of Elliptic Curves, Springer-Verlag, Berlin, 1995. [12] S.G. Vlăduţ, Cyclicity statistics for elliptic curves over finite fields, Finite Fields Appl. 5 (1999) 13 25.