Journal of Inequalities in Pure and Applied Mathematics http://jipam.vu.edu.au/ Volume 6, Issue 3, Article 67, 2005 AN EXPLICIT MERTENS’ TYPE INEQUALITY FOR ARITHMETIC PROGRESSIONS OLIVIER BORDELLÈS 2 ALLÉE DE LA COMBE L A B ORIETTE 43000 AIGUILHE borde43@wanadoo.fr Received 27 September, 2004; accepted 04 May, 2005 Communicated by J. Sándor A BSTRACT. We give an explicit Mertens type formula for primes in arithmetic progressions using mean values of Dirichlet L-functions at s = 1. Key words and phrases: Mertens’ formula, Arithmetic progressions, Mean values of Dirichlet L−functions. 2000 Mathematics Subject Classification. 11N13, 11M20. 1. I NTRODUCTION AND M AIN R ESULT The very useful Mertens’ formula states that Y 1 e−γ 1 1− = 1+O p log x log x p≤x for any real number x ≥ 2, where γ ≈ 0.577215664 . . . denotes the Euler constant. Some explicit inequalities have been given in [4] where it is showed for example that −1 Y 1 (1.1) 1− < eγ δ (x) log x, p p≤x where (1.2) δ (x) := 1 + 1 . (log x)2 Let 1 ≤ l ≤ k be positive integers satisfying (k, l) = 1. The aim of this paper is to provide an explicit upper bound for the product −1 Y 1 (1.3) 1− . p p≤x p≡l (mod k) ISSN (electronic): 1443-5756 c 2005 Victoria University. All rights reserved. 172-04 2 O LIVIER B ORDELLÈS In [2, 5], the authors gave asymptotic formulas for (1.3) in the form Y 1 1− ∼ c (k, l) (log x)−1/ϕ(k) , p p≤x p≡l (mod k) where ϕ is the Euler totient function and c (k, l) is a constant depending on l and k. Nevertheless, because of the non-effectivity of the Siegel-Walfisz theorem, one cannot compute the implied constant in the error term. Moreover, the constant c (k, l) is given only for some particular cases in [2], whereas K.S. Williams established a quite complicated expression of c (k, l) involving a product of Dirichlet L−functions L (s; χ) and a function K (s; χ) at s = 1, where K (s; χ) is the generating Dirichlet series of the completely multiplicative function kχ defined by ( −χ(p) ) χ (p) 1 kχ (p) := p 1 − 1 − 1− p p for any prime number p and any Dirichlet character χ modulo k. The author then gave explicit expressions of c (k, l) in the case k = 24. It could be useful to have an explicit upper bound for (1.3) valid for a large range of k and x. Indeed, we shall see in a forthcoming paper that such a bound could be used to estimate class numbers of certain cyclic number fields. We prove the following result: Theorem 1.1. Let 1 ≤ l ≤ k be positive integers satisfying (k, l) = 1 and k ≥ 37, and x be a positive real number such that x > k. We have: v 1 −1 γ ϕ(k) u Y Y 1 e ϕ (k) 1 2(γ−B) u tζ (2) 1− <e log x · Φ (x, k) , 1− 2 · p p k p≤x p|k p≡l (mod k) where ( Φ (x, k) := exp !) √ √ 2 k log k X L 0 (1; χ) + 2 k log k + E − γ , ϕ (k) χ6=χ L 2 log x 0 B ≈ 0.261497212847643 . . . and E ≈ 1.332582275733221 . . . The restriction k ≥ 37 is given here just to use a simpler expression of the Polyá-Vinogradov inequality, but one can prove a similar result with k ≥ 9 only, the constants in Φ (x, k) being slightly larger. 2. N OTATION p denotes always a prime, 1 ≤ l ≤ k are positive integers satisfying (k, l) = 1 and k ≥ 37, x > k is a real number, ! n X 1 γ := lim − log n ≈ 0.5772156649015328 . . . n→∞ k k=1 is the Euler constant and γ1 := lim n→∞ n X log k k=1 k (log n)2 − 2 ! ≈ −0.07281584548367 . . . is the first Stieltjes constant. Similarly, E := lim n→∞ log n − J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 X log p p≤n p ! ≈ 1.332582275733221 . . . http://jipam.vu.edu.au/ A M ERTENS ’ INEQUALITY FOR A RITHMETIC and B := lim n→∞ X1 p≤n p 3 PROGRESSIONS ! − log log n ≈ 0.261497212847643 . . . χ denotes a Dirichlet character modulo k and χ0 is the principal character modulo k. For any P χ(n) Dirichlet character χ modulo k and any s ∈PC such that Re s > 1, L (s; χ) := ∞ n=1 ns is the Dirichlet L− function associated to χ. χ6=χ0 means that the sum is taken over all nonprincipal characters modulo k. Λ is the Von Mangoldt function and f ∗ g denotes the usual Dirichlet convolution product. 3. S UMS WITH P RIMES From [4] we get the following estimates: Lemma 3.1. X p ∞ X 1 =E−γ log p pα α=2 and ∞ XX 1 = γ − B. αpα p α=2 4. T HE P OLYÁ -V INOGRADOV I NEQUALITY AND C HARACTER S UMS WITH P RIMES Lemma 4.1. Let χ be any non-principal Dirichlet character modulo k ≥ 37. (i) For any real number x ≥ 1, X 9√ χ (n) < k log k. 10 n≤x (ii) Let F ∈ C 1 ([1; +∞[ , [0; +∞[) such that F (t) & 0. For any real number x ≥ 1, t→∞ X 9 √ χ (n) F (n) ≤ F (x) k log k. 5 n>x (iii) For any real number x > k, 0 X χ (p) √ L 2 2 k log k (1; χ) + 1 + E − γ . < log x p L p>x Proof. (i) The result follows from Qiu’s improvement of the Polyá-Vinogradov inequality (see [3, p. 392]). (ii) Abel summation and (i). (iii) Let χ 6= χ0 be a Dirichlet character modulo k ≥ 37 and x > k be any real number. 1, if n = 1 and 1 (n) = 1, we get: (a) Since χ (µ ∗ 1) = ε where ε (n) = 0, otherwise X µ (d) χ (d) X χ (m) =1 d m d≤x m≤x/d and hence, since χ 6= χ0 , X µ (d) χ (d) d≤x d X X 1 µ (d) χ (d) χ (m) = + 1 L (1; χ) d≤x d m J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 m>x/d http://jipam.vu.edu.au/ 4 O LIVIER B ORDELLÈS and thus, using (ii), √ √ X µ (d) χ (d) 9 k log k + 1 2 k log k 5 < . ≤ |L (1; χ)| |L (1; χ)| d d≤x (4.1) (b) Since log = Λ ∗ 1, we get: X χ (n) Λ (n) n≤x n X µ (d) χ (d) X χ (m) log m d m d≤x m≤x/d X X X χ (m) log m = + m = d≤x/e x/e<d≤x m≤x/d X µ (d) χ (d) X χ (m) log m χ (2) log 2 = + d m 2 m≤x/d d≤x/e X x/e<d≤x µ (d) χ (d) d X µ (d) χ (d) X µ (d) χ (d) X χ (m) log m = −L 0 (1; χ) − d d m d≤x/e + χ (2) log 2 2 d≤x/e X x/e<d≤x m>x/d µ (d) χ (d) d and, by using (ii), (4.1) and the trivial bound for the third sum, we get: X χ (n) Λ (n) √ L 0 9 X x log 2 e (4.2) log + 1+ < k log k 2 (1; χ) + L n 5x d 2 x n≤x d≤x/e 0 √ L 18 log 2 e ≤ k log k 2 (1; χ) + + 1+ L 5e 2 37 0 √ L < 2 k log k (1; χ) + 1 L since x > q ≥ 37. (c) By Abel summation, we get: X χ (p) X χ (p) log p 2 max . ≤ p>x p log x t≥x p p≤t Moreover, X χ (p) log p p≤t p = J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 X χ (n) Λ (n) n≤t n − ∞ XX χ (pα ) log p p α=2 pα ≤t pα http://jipam.vu.edu.au/ A M ERTENS ’ INEQUALITY FOR A RITHMETIC 5 PROGRESSIONS and then: ∞ X χ (p) log p X χ (n) Λ (n) X X 1 log p ≤ + p n pα p α=2 p≤t n≤t X χ (n) Λ (n) = +E−γ n n≤t 0 √ L < 2 k log k (1; χ) + 1 + E − γ L by (4.2) . This concludes the proof of Lemma 4.1. 5. M EAN VALUE E STIMATES OF D IRICHLET L−FUNCTIONS Lemma 5.1. (i) For any positive integers j, k, jk X 1 X log p c0 (j, k) 2ω(k) ϕ (k) = log (jk) + γ + + n k p − 1 jk n=1 p|k (n,k)=1 P where ω (k) := p|k 1 and |c0 (j, k)| ≤ 1. (ii) For any positive integer k ≥ 9, k ϕ (k) 2 ζ (2) Y 1− p|k 1 p2 2 X π log p + 2γ1 + γ + − log k + ≤ 0. 3 p−1 2 p|k (iii) For any positive integer k ≥ 9, Y 1/ϕ(k) |L (1; χ)| χ6=χ0 1 Y p 1 2 ≤ ζ (2) . 1− 2 p p|k Proof. (i) jk X 1 X µ (d) X 1 X µ (d) jk ε (d) d = = log +γ+ n d n d d jk n=1 (n,k)=1 d|k n≤jk/d d|k where |ε (d)| ≤ 1 and hence: jk X µ (d) X µ (d) log d X 1 1 X = {log (jk) + γ} − + ε (d) µ (d) n d d jk n=1 d|k (n,k)=1 d|k and we conclude by noting that X µ (d) d|k X µ (d) log d d|k J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 d d = =− d|k ϕ (k) , k ϕ (k) X log p k p−1 p|k http://jipam.vu.edu.au/ 6 O LIVIER B ORDELLÈS and X X ≤ ε (d) µ (d) µ2 (d) = 2ω(k) . d|k d|k (ii) Define A (k) := k ϕ (k) 2 ζ (2) Y 1− p|k 1 p2 2 X log p π + 2γ1 + γ + − log k + . 3 p−1 2 p|k Using [1] we check the inequality for 9 ≤ k ≤ 513 and then suppose k ≥ 514. Since Y p Y 1 k = ≤ p p−1 ϕ (k) p−1 p|k p|k we have taking logarithms X log p k k ≥ log ≥ log p−1 ϕ (k) k−1 p|k and from the inequality ([4]) k 2.50637 < eγ log log k + ϕ (k) log log k valid for any integer k ≥ 3, we obtain 2 2 2 2.50637 π2 k γ A (k) ≤ ζ (2) e log log k + + 2γ1 + γ + − log <0 log log k 3 k−1 if k ≥ 514. (iii) First, X |L (1; χ)|2 = lim S (N ) χ6=χ0 N →∞ where 2 S (N ) := Nk Nk X χ (n) χ (m) X 1 − . nm n n=1 m,n=1 (n,k)=1 Following a standard argument, we have using (i): 2 N k N k X 1 X 1 S (N ) = ϕ (k) − mn n n=1 m6=n=1 m≡n (mod k) (n,k)=(m,k)=1 = ϕ (k) Nk X n=1 (n,k)=1 J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 1 + ϕ (k) n2 (n,k)=1 Nk X m6=n=1 m≡n (mod k) (n,k)=(m,k)=1 1 − mn Nk X n=1 (n,k)=1 2 1 n http://jipam.vu.edu.au/ A M ERTENS ’ INEQUALITY FOR A RITHMETIC ≤ ϕ (k) ζ (2) Y 1− p|k 1 p2 1 p2 + 2ϕ (k) −j)k N (N X X j=1 n=1 (n,k)=1 = ϕ (k) ζ (2) Y 1− p|k + 7 PROGRESSIONS 1 − n (n + jk) Nk X n=1 (n,k)=1 ≤ ϕ (k) ζ (2) p|k = ϕ (k) ζ (2) Y 1 p2 + p|k (n,k)=1 (n,k)=1 2 jk N Nk 2ϕ (k) X 1 X 1 X 1 − k j=1 j n=1 n n n=1 (n,k)=1 1− 2 N Nk Nk X 2ϕ (k) X 1 X 1 1 X 1 − − k j=1 j n n=1+jk n n n=1 n=1 1 1− 2 p 1 n (N −j)k (n,k)=1 Y 2 (n,k)=1 ω(k) X log p c0 (j, k) 2 ϕ (k) log (jk) + γ + + + k j=1 j k p − 1 jk p|k 2 ω(k) X log p c0 (N, k) 2 ϕ (k) . − log (N k) + γ + + k p−1 Nk N 2ϕ (k) X 1 p|k We now neglect the dependance of c0 in k. Since M X 1 c1 (M ) = log M + γ + m M m=1 and M X log m (log M )2 c2 (M ) log M = + γ1 + , m 2 M m=1 where 0 < c1 (M ) ≤ 1 2 and |c2 (M )| ≤ 1, we get: 2 ϕ (k) 2c2 (N ) log N S (N ) ≤ ϕ (k) ζ (2) + (log N )2 + 2γ1 + k N p|k X log p c1 (N ) + 2 log k + γ + log N + γ + p−1 N p|k 2 X log p − log (N k) + γ + p−1 p|k N ω(k)+1 X X 2 ϕ (k) c0 (j) c0 (N ) log p + − log (N k) + γ + k2 j2 N p−1 j=1 Y 1 1− 2 p p|k − 22ω(k) c20 (N ) N 2k2 J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 http://jipam.vu.edu.au/ 8 O LIVIER B ORDELLÈS 2 2 X 1 log p ϕ (k) 1− 2 + 2γ1 + γ − log k + = ϕ (k) ζ (2) p k p−1 p|k p|k X log p 2c2 (N ) log N 2c1 (N ) + log k + γ + + N p−1 N p|k N ω(k)+1 X X ϕ (k) log p 2 c0 (j) c0 (N ) + − log (N k) + γ + k2 j2 N p−1 Y j=1 − p|k 22ω(k) c20 (N ) N 2k2 and then Y 1 lim S (N ) ≤ ϕ (k) ζ (2) 1− 2 N →∞ p p|k 2 2 X ϕ (k) log p 2ω(k) ϕ (k) π 2 + 2γ1 + γ − log k + + k p−1 3k 2 p|k and the inequality 2ω(k) ≤ ϕ (k) (valid for any integer k ≥ 3 and 6= 6) implies Y 1 lim S (N ) ≤ ϕ (k) ζ (2) 1− 2 N →∞ p p|k 2 2 2 X ϕ (k) π log p + 2γ1 + γ + − log k + k 3 p−1 p|k 2 2 Y Y 1 ϕ (k) k 1 = (ϕ (k) − 1) ζ (2) 1− 2 + ζ (2) 1− 2 ϕ (k) p k p p|k p|k 2 2 X π log p +2γ1 + γ + − log k + 3 p−1 p|k Y 1 ≤ (ϕ (k) − 1) ζ (2) 1− 2 p p|k if k ≥ 9 by (ii). Hence X Y 1 1 2 |L (1; χ)| ≤ ζ (2) 1− 2 . ϕ (k) − 1 χ6=χ p 0 J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 p|k http://jipam.vu.edu.au/ A M ERTENS ’ INEQUALITY FOR A RITHMETIC 9 PROGRESSIONS Now the IAG inequality implies: ) ( X Y 1 1 log |L (1; χ)|2 |L (1; χ)| ϕ(k) = exp 2ϕ (k) χ6=χ0 χ6=χ0 !) ( X 1 ϕ (k) − 1 2 |L (1; χ)| ≤ exp log 2ϕ (k) ϕ (k) − 1 χ6=χ 0 ≤ ζ (2) Y p|k ≤ ζ (2) Y ϕ(k)−1 2ϕ(k) 1 1− 2 p 12 1 . p2 1− p|k 6. P ROOF OF THE T HEOREM Lemma 6.1. If χ0 is the principal character modulo k and if x > k, then: −χ0 (p) Y 1 eγ ϕ (k) δ (x) 1− < · log x, p k p≤x where δ is the function defined in (1.2). Proof. Since x > k, Y p≤x p|k 1 1− p = Y p|k 1 1− p = ϕ (k) k and then Y p≤x 1 1− p −χ0 (p) = p≤x p-k = 1 1− p −1 1 1− p −1 Y Y Y p≤x p≤x p|k 1 1− p −1 ϕ (k) Y 1 1− = k p≤x p and we use (1.1) . Proof of the theorem. Let 1 ≤ l ≤ k be positive integers satisfying (k, l) = 1 and k ≥ 37, and x be a positive real number such that x > k. We have: −ϕ(k) Y −χ0 (p) Y Y −χ(p) !χ(l) Y 1 1 1 1− = 1− · 1− := Π1 × Π2 p p p p≤x p≤x p≤x χ6=χ p≡l (mod k) J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 0 http://jipam.vu.edu.au/ 10 O LIVIER B ORDELLÈS with Π1 < eγ ϕ (k) δ (x) · log x by Lemma 6.1. Moreover, k ( !) X X 1 Π2 = exp χ (l) − χ (p) log 1 − p p≤x χ6=χ0 ! ∞ X XX χ (p) = exp χ (l) αpα p≤x α=1 χ6=χ0 !) ( ∞ X χ (p) X X X χ (p) = exp χ (l) + p αpα p≤x p≤x α=2 χ6=χ 0 and if χ 6= χ0 , we have ! −1 ∞ Y X χ (p) X χ (p) X X χ (p) χ (pα ) 1− L (1; χ) = = exp + + p p p αpα p p>x p α=2 p≤x and thus ( !) ∞ ∞ α Y X X X X X X χ (p) χ (p ) χ (p) + − Π2 = L (1; χ)χ(l) · exp χ (l) − α p αp αpα p α=2 p>x p≤x α=2 χ6=χ χ6=χ 0 0 and hence ) ∞ X X χ (p) XX 1 |Π2 | ≤ |L (1; χ)| · exp + 2 (ϕ (k) − 1) p αpα p α=2 χ6=χ0 χ6=χ0 p>x !) ( Y X X χ (p) = e2(ϕ(k)−1)(γ−B) |L (1; χ)| · exp p>x p ( Y χ6=χ0 χ6=χ0 and we use Lemma 4.1 (iii) and Lemma 5.1 (iii). We conclude the proof by noting that, if δ(x) x > 37, e2(γ−B) < 1. R EFERENCES [1] PARI/GP, Available by anonymous ftp from the URL: ftp://megrez.math.u-bordeaux. fr/pub/pari. [2] E. GROSSWALD, Some number theoretical products, Rev. Colomb. Mat., 21 (1987), 231–242. [3] D.S. MITRINOVIĆ AND J. SÁNDOR (in cooperation with B. CRSTICI), Handbook of Number Theory, Kluwer Acad. Publishers, ISBN: 0-7923-3823-5. [4] J.B. ROSSER AND L. SCHŒNFELD, Approximate formulas for some functions of prime numbers, Illinois J. Math., 6 (1962), 64–94. [5] K.S. WILLIAMS, Mertens’ theorem for arithmetic progressions, J. Number Theory, 6 (1974), 353– 359. J. Inequal. Pure and Appl. Math., 6(3) Art. 67, 2005 http://jipam.vu.edu.au/