# On the number-theoretic functions ν(n) and Ω(n) - PDF

Description
ACTA ARITHMETICA LXXVIII.1 (1996) On the number-theoretic functions ν(n) and Ω(n) by Jiahai Kan (Nanjing) 1. Introduction. Let d(n) denote the divisor function, ν(n) the number of distinct prime factors,

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information
Category:

## Short Stories

Publish on:

Views: 41 | Pages: 7

Share
Transcript
ACTA ARITHMETICA LXXVIII.1 (1996) On the number-theoretic functions ν(n) and Ω(n) by Jiahai Kan (Nanjing) 1. Introduction. Let d(n) denote the divisor function, ν(n) the number of distinct prime factors, and Ω(n) the total number of prime factors of n, respectively. In 1984 Heath-Brown [4] proved the well-known Erdős Mirsky conjecture [1] (which seemed at one time as hard as the twin prime conjecture, cf. [4, p. 141]): (A) There exist infinitely many positive integers n for which d(n + 1) = d(n). The method of Heath-Brown [4] can also be used to prove the conjecture: (B) There exist infinitely many positive integers n for which Ω(n + 1) = Ω(n). Another conjecture of Erdős for d(n) is (cf. e.g. [5, p. 308]): (C) Every positive real number is a limit point of the sequence {d(n + 1)/d(n)}, and the similar conjecture for Ω(n) is (D) Every positive real number is a limit point of the sequence {Ω(n + 1)/Ω(n)}. It follows from the results of Heath-Brown that 1 is a limit point of the sequence {d(n + 1)/d(n)}, and also a limit point of the sequence {Ω(n + 1)/Ω(n)}. As for ν(n), Erdős has similar conjectures: (E) There exist infinitely many positive integers n for which ν(n + 1) = ν(n). (F) Every positive real number is a limit point of the sequence {ν(n + 1)/ν(n)}. [91] 92 J. Kan Compared with the status of conjectures (A), (B), (C), (D), much less is known about conjectures (E) and (F). The best result up to date for conjecture (E) is the following Theorem (Erdős Pomerance Sárközy) (cf. [2, p. 251, Theorem 1]). There exist infinitely many positive integers n for which where c denotes a positive constant. ν(n + 1) ν(n) c And for conjecture (F), no limit point of the sequence {ν(n + 1)/ν(n)} is known yet. The purpose of this paper is (i) to improve the result of Erdős Pomerance Sárközy about conjecture (E), and (ii) to prove conjectures (F) and (D). In fact, the following more general results will be proved here. Let b denote any given nonzero integer, and k denote any fixed integer greater than one. We have Theorem 1. There exist infinitely many positive integers n for which ν(n + b) ν(n) 1 and ν(n) = k. Theorem 2. Every positive real number is a limit point of the sequence {ν(n + b)/ν(n)}. Theorem 3. Every positive real number is a limit point of the sequence {Ω(n + b)/ω(n)}. 2. Lemmas. We deduce in this section some lemmas by the sieve method. Terminology and notations here have their customary meaning and coincide with those of [3] and [6]. Let A denote a finite set of integers, A X. Let and assume that, for squarefree d, A d = {a : a A, d a}, A d = ω(d) d X + r d, and ω(d) is multiplicative. Define P = {p : p a, a A} (i.e., P is the set of all primes dividing at least one a in A), and P the complement of P with respect to the set of all primes. In the following conditions the A i s denote positive constants. (Ω 1 ) 0 ω(p)/p 1 1/A 1. (Ω2(1)) A 2 ln ln 3X ω(p) p ln p ln z w A 2 if 2 w z. w p z (Ω 3 ) z p y, p P Number-theoretic functions ν(n) and Ω(n) 93 A p 2 A 3 ( X ln X z ) + y if 2 z y. (R (1, α)) There exists α (0 α 1) such that, for any given A 0, there is B = B(A) 0 such that µ 2 (d)3 ν(d) r d A 4 X ln A X. d x α ln B X, (d,p)=1 As a kind of exponential measure for the magnitude of the a s of A we introduce, for each positive integer r, the function (1) Λ r = r + 1 ln(4/(1 + 3 r )). ln 3 Clearly Λ r is increasing, Λ 1 = 1 and (2) r + 1 ln 4 ln 3 Λ ln 3.6 r r + 1 for r 2. ln 3 Lemma 1. Let (Ω 1 ), (Ω2(1)), (Ω 3 ) and (R (1, α)) hold. Suppose that (3) (a, P) = 1 for all a A. Let δ be a real number satisfying (4) 0 δ Λ 2, and let r 0 be the least integer of all r s (r 2) satisfying (5) a X α(λ r δ) Then we have, for X X 0, for all a A. (6) #{n : n A, n = p 1... p t+1 or p 1... p t+2 or... or p 1... p r, p 1 ... p t X 1/ ln ln X, X α/4 p t+1 p t+2 ... p r } c(r ( ( )) 0, δ) ln ln ln X t!α c(ω)x ln 1 X(ln ln X) t 1 O, ln ln X where p i s denote primes, t = r r 0, (7) c(r 0, δ) = 2(r (1 + 3 r 0 )(Λ r0 δ)) 1 δ(1 + 3 r 0 ) ln 3, and (8) c(ω) = p (1 ω(p)/p)(1 1/p) 1. P r o o f. This lemma follows from [6, Theorem 1 and p. 281, (39) of Remark 3]. Lemma 2. Let F (n) ( ±n) be an irreducible polynomial of degree g ( 1) with integer coefficients. Let ϱ(p) denote the number of solutions of the congruence F (m) 0 mod p. 94 J. Kan Suppose that (9) ϱ(p) p for all p, and also that (10) ϱ(p) p 1 if p F (0) and p g + 1. Then we have, for any fixed r r 0 = 2g + 1 and for x x 0 = x 0 (F ), (11) #{p : p x, F (p) = p 1... p r r0 +1 or p 1... p r r0 +2 or... or 3/2 (r r 0 )! p F (0) 1 ϱ(p)/(p 1) 1 1/p P r o o f. We consider the sequence p 1... p r, p 1 p 2 ... p r } 1 (ϱ(p) 1)/(p 1) 1 1/p p F (0) A = {F (p) : p x}, x ln 2 x(ln ln x) r r 0. and we take P to be the set of all primes. In [3, pp , Example 6] (with k = 1), in accordance with [3, p. 23 (3.48), p. 28 (4.15), p. 24 (3.51)], we choose (12) X = li x, ω(p) = ϱ 1(p) p for all p, p 1 where (cf. [3, p. 24 (3.53)]) (13) ϱ 1 (p) = { ϱ(p) if p F (0), ϱ(p) 1 if p F (0). From [3, p. 28 (4.15), p. 24 (3.52) and p. 24 (3.55)] we have (14) R d g ν(d) (E(x, d) + 1) if µ(d) 0, where (cf. [3, p. 22 (3.41)]) E(x, d) = max max π(y; d, a) li y/ϕ(d). 2 y x 1 a d (a,d)=1 It is now a matter of confirming the conditions under which Lemma 1 is valid. First consider (Ω 1 ). Here we see that, for p g + 1, (13), (10) and (9) imply that and hence that ϱ 1 (p) p 2, ω(p) p 2 ( p 1 p 1 1 ) p if p g + 1; g Number-theoretic functions ν(n) and Ω(n) 95 if, on the other hand, p g + 2, then, by [3, p. 24 (3.54)], and we find that ω(p) g p 1 p ϱ 1 (p) ϱ(p) g, g ( g + 1 p = 1 1 ) p, g + 1 thus verifying (Ω 1 ) with A 1 = g + 1. Condition (Ω2(1)) is a consequence of Nagel s result (cf. [3, p. 18 (3.17)] with k = 1) ϱ(p) p ln p = ln w + O F (1). Moreover, since p w #{p : p x, F (p ) 0 mod p 2 } #{n : n x, F (n) 0 mod p 2 } x p X ln X p 2 + 1, it is easy to see that (Ω 3 ) is satisfied. As for (R (1, α)), we see from (14) and Bombieri s theorem (cf. [3, p. 111, Lemma 3.3, p. 115, Lemmas 3.4 and 3.5]) that, for any given A 0, there is B = B(A) 0 such that d x 1/2 ln B X Thus (R (1, α)) holds with µ 2 (d)3 ν(d) r d x ln A+1 x (15) α = 1/2. X ln A X. Finally, because of our choice of P, (3) is trivially true (cf. [6, p. 285 (40)]). We may now apply Lemma 1. We take δ = 2/3 and r 0 = 2g + 1 and find that, by (15) and (2), for r r 0, α(λ r δ) 1 ( 2g ) 3 = g , so that (5) is satisfied if x x 1 = x 1 (F ). Hence, by Lemma 1, (12) and (15), we have (16) #{p : p x, F (p) = p 1... p r r0 +1 or p 1... p r r0 +2 or... or 2 (r r 0 )! c(r 0, δ) p p 1... p r, p 1 p 2 ... p r } 1 ϱ 1 (p)/(p 1) x 1 1/p ln 2 x (ln ln x)r r 0. 96 J. Kan It follows from (7), (2) and δ = 2/3 that (17) c(r 0, δ) 2(r Λ r0 + δ) 1 δ ln 3 ( ln 4 2 ln ) 1 2 ln 3 Combining (16), (17) and (13) we obtain (11), and the proof of Lemma 2 is complete. Lemma 3. Let a and b be integers satisfying (18) ab 0, (a, b) = 1 and 2 ab. Then, for any fixed integer r 3 and for x x 0 = x 0 (a, b), we have (19) #{p : p x, ap + b = p 1... p r 2 or p 1... p r 1 or p 1... p r, 3 (r 3)! (1 (p 1) 2 ) p 2 2 p ab p 1 p 2 ... p r } p 1 p 2 x ln 2 x (ln ln x)r 3. P r o o f. In Lemma 2 let F (n) = an + b. Since (18) implies (9), (10) and b 0, by Lemma 2 we have the assertion. 3. Proof of the Theorems. Let q i denote a prime. In Lemma 3 we take a = q 1 q 2... q r 2 with q 1 q 2 ... q r 2, and let n = ap. Then from (19) it is easy to see that there are infinitely many n for which and where so for such n, ν(n) = ν(ap) = ν(q 1 q 2... q r 2 p) = r 1 ν(n + b) = ν(ap + b) = t, t = r 2 or r 1 or r; ν(n + b) ν(n) 1 and ν(n) = r 1. This completes the proof of Theorem 1. If in Lemma 3 we take a = q 1 q 2... q s 1, q 1 q 2 ... q s 1, and let n = ap, then from (19) again we see that there are infinitely many n for which where ν(n) = Ω(n) = s and ν(n + b) = Ω(n + b) = t, t = r 2 or r 1 or r; Number-theoretic functions ν(n) and Ω(n) 97 so for such n, ν(n + b) Ω(n + b) = = t ν(n) Ω(n) s. Moreover, for any given positive real number α and for any small ε 0, the fraction t/s (where t, s are both variable) may be chosen to approximate α arbitrarily closely, i.e. α t/s ε. Thus α is a limit point of the sequence {ν(n + b)/ν(n)}, as well as a limit point of the sequence {Ω(n + b)/ω(n)}. We have thus completed the proof of Theorems 2 and 3. R e m a r k. The method here gives for the number of solutions of ν(n + b) ν(n) 1, n x, a lower bound x ln 2 x(ln ln x) t for t arbitrarily large but fixed. In view of the works of Heath-Brown, Hildebrand, and Erdős Pomerance Sárközy, it seems reasonable to conjecture that this lower bound is x/ ln ln x. References [1] P. Erdős and L. Mirsky, The distribution of the values of the divisor function d(n), Proc. London Math. Soc. (3) 2 (1952), [2] P. Erdős, C. Pomerance and A. Sárközy, On locally repeated values of certain arithmetic functions II, Acta Math. Hungar. 49 (1987), [3] H. Halberstam and H.-E. Richert, Sieve Methods, Academic Press, London, [4] D. R. Heath-Brown, The divisor function at consecutive integers, Mathematika 31 (1984), [5] A. Hildebrand, The divisor function at consecutive integers, Pacific J. Math. 129 (1987), [6] J. Kan, On the lower bound sieve, Mathematika 37 (1990), Nanjing Institute of Post and Telecommunications Nanjing Nanjing, China Received on and in revised form on (2928)
Related Search
Similar documents

View more...