Bull. Korean Math. Soc. 50 (2013), No. 1, pp. 233–240 http://dx.doi.org/10.4134/BKMS.2013.50.1.233 ON THE FOURTH POWER MEAN OF GENERALIZED TWO-TERM EXPONENTIAL SUMS Tingting Wang Abstract. In this paper, we use the elementary method and the theory of complex functions to study the computational problem of the fourth power mean of the generalized two-term exponential sums, and give two exact identities for them. 1. Introduction Let q ≥ 3 be a positive integer. For any integers m and n, the generalized two-term exponential sum C(m, n, k, χ; q) is defined by q X mak + na χ(a)e C(m, n, k, χ; q) = , q a=1 where χ denotes any Dirichlet character mod q, and e(y) = e2πiy . The various properties of C(m, n, k, χ; q) were investigated by many authors (see [2, 3, 4, 5] and [7]). For example, from the general result of Pigno and Pinner [7] we can deduce that 1 |C(m, 0, 2, χ; q)| ≤ 2ω(q) q 2 , for (m, q) = 1, where ω(q) denotes the number of all distinct prime divisors of q. In fact, |C(m, 0, 2, χ; q)| is a multiplicative function of q, and so if q = αs 1 α2 pα 1 p2 · · · ps denotes the prime power decomposition of q, χ = χ1 χ2 · · · χs , αi i mi = q/pi , where χi is a multiplicative character mod pα i , i = 1, 2, . . . , s. Then for (m, q) = 1, we have αi Y p s X s q 2 Y i √ mmi a i = 2ω(q) · q. χ (a)e |C(m, 0, 2, χ; q)| = 2 · ≤ pα i αi i p i=1 i i=1 a=1 Received July 22, 2011; Revised August 30, 2012. 2010 Mathematics Subject Classification. Primary 11L40, 11F20. Key words and phrases. generalized two-term exponential sums, fourth power mean, identity. This work is supported by the N.S.F. (11071194) of P. R. China and G.I.C.F. (YZZ12065) of NWU. c 2013 The Korean Mathematical Society 233 234 TINGTING WANG The case where q is a prime is due to Weil [8]. Zhang Wenpeng [9] proved that for any odd prime p and integer n with (n, p) = 1, we have X 4 |C(n, 0, 2, χ; p)| χ mod p = ( h √ i p , (p − 1) 3p2 − 6p − 1 + 4 np 2 (p − 1)(3p − 6p − 1), if p ≡ 1 mod 4; if p ≡ 3 mod 4; and X 6 χ mod p |G(n, 0, 2, χ; p)| = (p − 1)(10p3 − 25p2 − 4p − 1), if p ≡ 3 mod 4, where np is the Legendre symbol. The main purpose of this paper is to calculate exact formulae for the sums p X X 4 |C(m, n, 3, χ; p)| , | C(m, n, 2, χ; p) |4 and χ mod p n=1 and give two computational formulae for them. For the sake of convenience, we 2 p X a √ e let C(1, p) denotes the quadratic Gauss sums, i.e., C(1, p) = = p, p a=1 √ if p ≡ 1 mod 4; C(1, p) = i p, if p ≡ 3 mod 4, i2 = −1. Under these notations, we shall prove the following: Theorem 1. Let p be an odd prime. Then for any integers m and n with (mn, p) = 1, we have the identity X 4 |C(m, n, 2, χ; p)| χ mod p = 2p3 − 8p2 + 5p + 1 m −4m · n2 −1 4m · n2 + 2(p − 1)C(1, p) e + e , p p p p where m denotes the Legendre’s symbol, and m · m ≡ 1 mod p. p Theorem 2. Let p be an odd prime with p 6= 3a + 1. Then for any integer m with (m, p) = 1, we have the identity p X n=1 |C(m, n, 3, χ; p)|4 p(2p2 − 3p − 3), p2 (3p − 7) , = 2 p (2p − 6) , if χ is the principal character mod p, if χ is Legendre symbol mod p, otherwise. GENERALIZED TWO-TERM EXPONENTIAL SUMS 235 For general integer q ≥ 3, whether there exists a computational formula for q X X 2r 2r |C(m, n, k, χ; q)| |C(m, n, k, χ; q)| and χ mod q n=1 are two open problems, where k, r ≥ 3 are integers. 2. Several lemmas In this section, we shall give several lemmas, which are necessary in the proof of our theorems. First we have the following: Lemma 1. Let p be an odd prime. Then for any integers m and n with (mn, p) = 1, we have the identity 2 p X mb + nb m −4mn2 e = e C(1, p), p p p b=1 where xp denotes the Legendre’s symbol, and mm ≡ 1 mod p. Proof. Note that (mn, p) = 1, so from the properties of the complete residue system mod p we have 2 X p p X mb + nb m(mnb)2 + n(mnb) e e (1) = p p b=1 b=1 p X mn2 (b2 + b) e = p b=1 p X 4mn2 (4b2 + 4b) e = p b=1 2 p 4mn2 (2b + 1) − 1 X e = p b=1 ! p X 4mn2 b2 − 1 = e p b=1 p 4mn2 b2 −4mn2 X . e =e p p b=1 Note that for any integer n with (n, p) = 1, from Theorem 7.5.4 of [6] we 2 p X 2 nb n have C(n, p) = = m e = C(1, p), 4mn p p . Then from (1) p p b=1 we may immediately deduce the identity 2 p X mb + nb m −4mn2 e = e C(1, p). p p p b=1 236 TINGTING WANG This proves Lemma 1. Lemma 2. Let p be an odd prime with p 6= 3k + 1. Then for any character χ mod p and integer m with (m, p) = 1, we have the identity !2 p−1 p−1 p X X X mb3 (a3 − 1) + nb(a − 1) e χ(a) p n=1 a=2 b=1 if χ is the principal character mod p, p(p2 − p − 4), p(2p2 − 5p − 1), if χ is the Legendre symbol mod p, = p p2 − 4p − 1 , otherwise. Proof. Since p 6= 3k + 1, if b passes through a reduced residue system ( mod p), then b3 also passes through a reduced residue system (mod p). Now for all 2 ≤ a ≤ p − 1, it is clear that (a − 1, p) = 1. Thus, from the well known trigonometric sum 3 X q q X b n bn 0, if (n, q) = 1, = = e e q, if q | n, q q b=1 b=1 we have !2 p−1 X mb3 (a3 − 1) + nb(a − 1) e χ(a) (2) p n=1 a=2 b=1 !!2 p−1 p−1 p 3 X X X mb3 a − 1 (a3 − 1) + nb e χ(a) = p n=1 a=2 b=1 3 3 p−1 p−1 p−1 mb3 a − 1 (a3 − 1) − c − 1 (c3 − 1) X X X e χ(c) χ(a) =p p c=2 a=2 b=1 3 3 p−1 p−1 p−1 X mb a − 1 (a3 − 1) − c − 1 (c3 − 1) X X . e χ(ac) =p p a=2 c=2 p−1 p X X b=1 3 3 Note that a − 1 (a3 − 1) ≡ c − 1 (c3 − 1) mod p if and only if a = c or a = c. Also, p − 1 = p − 1. Thus if χ = χ0 is the principal character mod p, then from (2) we have !2 p−1 p p−1 X X X mb3 (a3 − 1) + nb(a − 1) (3) χ(a) e p n=1 a=2 b=1 p−1 X X p−1 2 1 = p (p − 1)(2p − 5) − = p(p − p − 4). a=2 c=2 a6=c, c GENERALIZED TWO-TERM EXPONENTIAL SUMS If χ = (4) ∗ p 237 is the Legendre’s symbol, then we obtain from (2) !2 p−1 p−1 p X a X X mb3 (a3 − 1) + nb(a − 1) e p p n=1 a=2 b=1 p−1 X a X p−1 c = p (p − 1)(2p − 5) − p p a=2 c=2 a6=c, c = p p(2p − 5) − p−1 X a=1 a p −1 !2 = p(2p2 − 5p − 1). If χ 6= χ0 and p∗ , then from (2) we have (5) p X p−1 X p−1 X mb3 (a3 − 1) + nb(a − 1) χ(a) e p n=1 a=2 b=1 p−1 p−1 p−1 p−1 XX XX χ(a)χ(c) χ(ac) − = p p = p p(p − 2) + p = p p(p − 2) + p = p p2 − 4p − 1 . !2 a=2 c=2 a=2 c=2 a=c or c p−2 X a=2 p−1 X a=1 χ(a2 ) − p−1 X a=2 χ2 (a) − 2p − !2 χ(a) p−1 X a=1 χ(a) − 1 !2 Combining (3), (4) and (5) we may immediately deduce Lemma 2. 3. Proof of the theorems In this section, we shall use the lemmas from Section 2 to complete the proof of our theorems. First we prove Theorem 1. For any character χ mod p, note that p−1 X X p−1 m(a2 − b2 ) + n(a − b) 2 χ (ab) e |C(m, n, 2, χ; p)| = p a=1 b=1 p−1 p−1 X X mb2 (a2 − 1) + nb(a − 1) e χ(a) (6) = . p a=1 b=1 238 TINGTING WANG Then from (6) and the orthogonality relation for character sums mod p we have X (7) |C(m, n, 2, χ; p)|4 χ mod p = = = = 2 p−1 X p−1 X mb2 (a2 − 1) + nb(a − 1) X e χ(a) p χ mod p a=1 b=1 p−1 2 p−1 X X mb2 (a2 − 1) + nb(a − 1) e (p − 1) p a=1 b=1 2 p−2 X p−1 X mb2 (a2 − 1) + nb(a − 1) 3 (p − 1) + (p − 1) + (p − 1) e p a=2 b=1 2 p−3 X p−1 X mb2 (a2 + 2a) + nba 3 (p − 1) + (p − 1) + (p − 1) e . p a=1 b=1 Then applying Lemma 1 and (7) we have X 4 (8) |C(m, n, 2, χ; p)| χ mod p = (p − 1)3 + (p − 1) p−3 X m(a2 + 2a) e + (p − 1) p a=1 = (p − 1)3 + (p − 1) p−3 X m(1 + 2a) + (p − 1) e p a=1 = (p − 1)3 + (p − 1) p−3 X m1 + 2a + (p − 1) e p a=1 −4m(a2 + 2a)(na)2 p −4m(1 + 2a)n2 p ! ! 2 C(1, p) − 1 2 C(1, p) − 1 2 ! −4m(1 + 2a)n2 C(1, p) − 1 p 2 p−2 X ma −4m · an2 3 = (p − 1) + (p − 1) + (p − 1) C(1, p) − 1 p e p a=2 3 = (p − 1) + (p − 1) + (p − 1)(p + 1)(p − 3) − (p − 1) R + R , where p−2 X ma −4m · an2 e C(1, p). R= p p a=2 GENERALIZED TWO-TERM EXPONENTIAL SUMS 239 It is clear that (9) R = C(1, p) " p−1 X a=1 ma p # 4mn2 −4man2 m −4mn2 −m e − e − e p p p p p 4m · n2 −1 m −4m · n2 −m C(1, p) − e − e p p p p p 2 2 −4m · n −1 4m · n m e + e , = p − C(1, p) p p p p = C(1, p) and m −4m · n2 −1 4m · n2 −1 (10) R = p − C(1, p) e + e . p p p p p Note that C 2 (1, p) = −1 p and C(1, p) · C(1, p) = p (see Theorem 9.16 of p [1]), so C(1, p) = −1 C(1, p). Then from (8), (9) and (10) we have p X χ mod p 4 |C(m, n, 2, χ; p)| = (p − 1)3 + (p − 1) + (p − 1)(p + 1)(p − 3) − 2(p − 1)R = (p − 1)3 + (p − 1) + (p − 1)(p + 1)(p − 3) − 2(p − 1)p −4m · n2 −1 m 4m · n2 e + e + 2(p − 1)C(1, p) p p p p = 2p3 − 8p2 + 5p + 1 4m · n2 m −4m · n2 −1 + 2(p − 1)C(1, p) e + e . p p p p This proves Theorem 1. Now we prove Theorem 2. Note that for any character χ mod p, we have (11) |C(m, n, 3, χ; p)|2 p−1 p−1 X X mb3 (a3 − 1) + nb(a − 1) e χ(a) = p a=1 b=1 p−1 p−1 X X mb3 (a3 − 1) + nb(a − 1) e χ(a) = (p − 1) + p a=2 b=1 ! p−1 p−1 3 X X mb3 a − 1 (a3 − 1) + nb e χ(a) = p−1+ . p a=2 b=1 240 TINGTING WANG So from Lemma 2 and (11) we have p X n=1 = p X n=1 4 |C(m, n, 3, χ; p)| (p − 1) + 2 = p(p − 1) + p X n=1 p−1 X χ(a) a=2 p−1 X a=2 p(2p2 − 3p − 3), p2 (3p − 7) , = 2 p (2p − 6) , p−1 X 3 mb3 a − 1 (a3 − 1) + nb p e b=1 χ(a) p−1 X 3 e b=1 !!2 mb3 a − 1 (a3 − 1) + nb p !!2 if χ is the principal character mod p, if χ is the Legendre symbol mod p, otherwise. This completes the proof of Theorem 2. Acknowledgment. The author would like to thank the referee for his very helpful and detailed comments which have significantly improved the presentation of this paper. References [1] Tom M. Apostol, Introduction to Analytic Number Theory, Springer-Verlag, New York, 1976. [2] T. Cochrane and C. Pinner, A further refinement of Mordell’s bound on exponential sums, Acta Arith. 116 (2005), no. 1, 35–41. , Using Stepanov’s method for exponential sums involving rational functions, J. [3] Number Theory 116 (2006), no. 2, 270–292. [4] T. Cochrane and Z. Y. Zheng, Pure and mixed exponential sums, Acta Arith. 91 (1999), no. 3, 249–278. [5] , Upper bounds on a two-term exponential sums, Sci. China Ser. A 44 (2001), no. 8, 1003–1015. [6] L. K. Hua, Introduction to Number Theory, Science Press, Beijing, 1979. [7] V. Pigno and C. Pinner, Twisted monomial Gauss sums modulo prime powers, http://www.math.ksu.edu/ pinner/Pubs/tgauss4a.pdf. [8] A. Weil, On some exponential sums, Proc. Nat. Acad. Sci. U.S.A. 34 (1948), 204–207. [9] W. Zhang, Moments of generalized quadratic Gauss sums weighted by L-functions, J. Number Theory 92 (2002), no. 2, 304–314. Department of Mathematics Northwest University Xi’an, Shaanxi, P. R. China E-mail address: [email protected]
© Copyright 2026 Paperzz