MULTIPLE HARMONIC SUMS AND MULTIPLE HARMONIC STAR SUMS ARE (NEARLY) NEVER INTEGERS

Similar documents
New Multiple Harmonic Sum Identities

arxiv: v2 [math.nt] 4 Jun 2016

Proofs of the infinitude of primes

THE DENOMINATORS OF POWER SUMS OF ARITHMETIC PROGRESSIONS. Bernd C. Kellner Göppert Weg 5, Göttingen, Germany

An identity involving the least common multiple of binomial coefficients and its application

arxiv: v2 [math.nt] 9 Oct 2013

Chapter 8. Infinite Series

SOLUTION TO RUBEL S QUESTION ABOUT DIFFERENTIALLY ALGEBRAIC DEPENDENCE ON INITIAL VALUES GUY KATRIEL PREPRINT NO /4

An Asymptotic Formula for Goldbach s Conjecture with Monic Polynomials in Z[x]

EGYPTIAN FRACTIONS WITH EACH DENOMINATOR HAVING THREE DISTINCT PRIME DIVISORS

#A69 INTEGERS 13 (2013) OPTIMAL PRIMITIVE SETS WITH RESTRICTED PRIMES

Review (11.1) 1. A sequence is an infinite list of numbers {a n } n=1 = a 1, a 2, a 3, The sequence is said to converge if lim

ON `-TH ORDER GAP BALANCING NUMBERS

GENERAL QUARTIC-CUBIC-QUADRATIC FUNCTIONAL EQUATION IN NON-ARCHIMEDEAN NORMED SPACES

Integral points of a modular curve of level 11. by René Schoof and Nikos Tzanakis

= 1 2x. x 2 a ) 0 (mod p n ), (x 2 + 2a + a2. x a ) 2

Fermat numbers and integers of the form a k + a l + p α

A NOTE ON AN ADDITIVE PROPERTY OF PRIMES

NOTES Edited by William Adkins. On Goldbach s Conjecture for Integer Polynomials

8.1 Sequences. Example: A sequence is a function f(n) whose domain is a subset of the integers. Notation: *Note: n = 0 vs. n = 1.

Representing integers as linear combinations of powers

arxiv:math/ v2 [math.nt] 3 Dec 2003

Chapter 8. P-adic numbers. 8.1 Absolute values

Math 324 Summer 2012 Elementary Number Theory Notes on Mathematical Induction

#A45 INTEGERS 12 (2012) SUPERCONGRUENCES FOR A TRUNCATED HYPERGEOMETRIC SERIES

AP Calculus Chapter 9: Infinite Series

Convergence of sequences and series

arxiv:math/ v1 [math.nt] 27 Feb 2004

ON MATCHINGS IN GROUPS

On some inequalities between prime numbers

ON A PARTITION PROBLEM OF CANFIELD AND WILF. Departamento de Matemáticas, Universidad de los Andes, Bogotá, Colombia

SUMSETS MOD p ØYSTEIN J. RØDSETH

A Geometric Proof that e is Irrational and a New Measure of its Irrationality

Cullen Numbers in Binary Recurrent Sequences

Taylor and Maclaurin Series. Copyright Cengage Learning. All rights reserved.

MATH 324 Summer 2011 Elementary Number Theory. Notes on Mathematical Induction. Recall the following axiom for the set of integers.

Automata and Number Theory

Zhi-Wei Sun Department of Mathematics, Nanjing University Nanjing , People s Republic of China. Received 8 July 2005; accepted 2 February 2006

A Subrecursive Refinement of FTA. of the Fundamental Theorem of Algebra

Math 200 University of Connecticut

Similar to sequence, note that a series converges if and only if its tail converges, that is, r 1 r ( 1 < r < 1), ( 1) k k. r k =

On Systems of Diagonal Forms II

EVERY NATURAL NUMBER IS THE SUM OF FORTY-NINE PALINDROMES

PRACTICE PROBLEMS: SET 1

Section 11.1 Sequences

MATH 117 LECTURE NOTES

Math 229: Introduction to Analytic Number Theory Čebyšev (and von Mangoldt and Stirling)

Introduction to Series and Sequences Math 121 Calculus II Spring 2015

Zhi-Wei Sun Department of Mathematics, Nanjing University Nanjing , People s Republic of China

On the irreducibility of certain polynomials with coefficients as products of terms in an arithmetic progression

be a sequence of positive integers with a n+1 a n lim inf n > 2. [α a n] α a n

Solutions to Assignment-11

ON PERMUTATION POLYNOMIALS OF PRESCRIBED SHAPE

Factorization of integer-valued polynomials with square-free denominator

GENERALIZATIONS OF SOME ZERO-SUM THEOREMS. Sukumar Das Adhikari Harish-Chandra Research Institute, Chhatnag Road, Jhusi, Allahabad , INDIA

MA257: INTRODUCTION TO NUMBER THEORY LECTURE NOTES

arxiv:math/ v1 [math.nt] 6 May 2003

Product Representations of Polynomials

RESEARCH PROBLEMS IN NUMBER THEORY

#A5 INTEGERS 18A (2018) EXPLICIT EXAMPLES OF p-adic NUMBERS WITH PRESCRIBED IRRATIONALITY EXPONENT

Irrationality exponent and rational approximations with prescribed growth

PACKING POLYNOMIALS ON SECTORS OF R 2. Caitlin Stanton Department of Mathematics, Stanford University, California

Journal of Inequalities in Pure and Applied Mathematics

Pade Approximations and the Transcendence

MATH 1231 MATHEMATICS 1B CALCULUS. Section 4: - Convergence of Series.

Solutions to Problem Set 4 - Fall 2008 Due Tuesday, Oct. 7 at 1:00

SOME REMARKS ON NUMBER THEORY BY P. ERDŐS 1. Let ABSTRACT This note contains some disconnected minor remarks on number theory. (1) Iz j I=1, 1<j<co be

Chapter 1. Introduction to prime number theory. 1.1 The Prime Number Theorem

Primes in arithmetic progressions

1 x i. i=1 EVEN NUMBERS RAFAEL ARCE-NAZARIO, FRANCIS N. CASTRO, AND RAÚL FIGUEROA

On two-point configurations in subsets of pseudo-random sets

THE HALF-FACTORIAL PROPERTY IN INTEGRAL EXTENSIONS. Jim Coykendall Department of Mathematics North Dakota State University Fargo, ND.

On intervals containing full sets of conjugates of algebraic integers

Almost fifth powers in arithmetic progression

An Exploration of the Arithmetic Derivative

A CONSTRUCTION OF ARITHMETIC PROGRESSION-FREE SEQUENCES AND ITS ANALYSIS

SUMS OF ENTIRE FUNCTIONS HAVING ONLY REAL ZEROS

ON A FAMILY OF ELLIPTIC CURVES

Another Proof of Nathanson s Theorems

ON A THEOREM OF TARTAKOWSKY

Splitting Fields for Characteristic Polynomials of Matrices with Entries in a Finite Field.

Incomplete exponential sums over finite fields and their applications to new inversive pseudorandom number generators

Local Fields. Chapter Absolute Values and Discrete Valuations Definitions and Comments

Part II. Number Theory. Year

Analytic Number Theory Solutions

arxiv:math/ v3 [math.co] 15 Oct 2006

INFINITE SEQUENCES AND SERIES

INTEGERS DIVISIBLE BY THE SUM OF THEIR PRIME FACTORS

103 Some problems and results in elementary number theory. By P. ERDÖS in Aberdeen (Scotland). Throughout this paper c, c,... denote absolute constant

Jonathan Sondow 209 West 97th Street Apt 6F New York, NY USA

The Distribution of Generalized Sum-of-Digits Functions in Residue Classes

On polynomially integrable convex bodies

DISTRIBUTION OF FIBONACCI AND LUCAS NUMBERS MODULO 3 k

1 i<j k (g ih j g j h i ) 0.

Markov s Inequality for Polynomials on Normed Linear Spaces Lawrence A. Harris

On Polya's Orchard Problem

Mei-Chu Chang Department of Mathematics University of California Riverside, CA 92521

Je Lagarias, University of Michigan. Workshop on Discovery and Experimentation in Number Theory Fields Institute, Toronto (September 25, 2009)

1 Adeles over Q. 1.1 Absolute values

Transcription:

#A0 INTEGERS 7 (207) MULTIPLE HARMONIC SUMS AND MULTIPLE HARMONIC STAR SUMS ARE (NEARLY) NEVER INTEGERS Khodabakhsh Hessami Pilehrood The Fields Institute for Research in Mathematical Sciences, Toronto, Canada hessamik@gmail.com Tatiana Hessami Pilehrood The Fields Institute for Research in Mathematical Sciences, Toronto, Canada hessamit@gmail.com Roberto Tauraso Dipartimento di Matematica, Università di Roma Tor Vergata, Roma, Italy tauraso@mat.uniroma2.it Received: /4/6, Accepted: 3/5/7, Published: 4/4/7 Abstract It is well known that the harmonic sum H n () = P n k= k is never an integer for n >. In 946, Erdős and Niven proved that the nested multiple harmonic sum H n ({} r ) = P applek < <k rapplen k k r can take integer values only for a finite number of positive integers n. In 202, Chen and Tang refined this result by showing that H n ({} r ) is an integer only for (n, r) = (, ) and (n, r) = (3, 2). In 204, Hong and Wang extended these results to the general arithmetic progression case. In this paper, we consider the integrality problem for arbitrary multiple harmonic and multiple harmonic star sums and show that none of these sums is an integer with some natural exceptions like those mentioned above.. Introduction A classic result of elementary number theory is that even though the partial sum of the harmonic series P n k= /k increases to infinity, it is never an integer for n >. Apparently the first published proof goes back to Theisinger [9] in 95, and, since then, it has been proposed as a challenging problem in several textbooks; among all we mention [5, p.6, Problem 30], [7, p.33, Problem 37], [8, p.53, Problem 250]. In 946, Erdős and Niven [3] proved a stronger statement: there is only a finite number of integers n for which there is a positive integer r apple n such that the r-th

INTEGERS: 7 (207) 2 elementary symmetric function of, /2,..., /n, that is applek < <k rapplen k k r, is an integer. In 202, Chen and Tang [] refined this result and succeeded to show that the above sum is not an integer with the only two exceptions: either n = r = or n = 3 and r = 2. Recently, this theme has been further developed by investigating the case when the variables of the elementary symmetric functions are /f(), /f(2),..., /f(n) with f(x) being a polynomial of nonnegative integer coe cients: see [4] and [0] for f(x) = ax + b and see [6] for f of degree at least two. In this paper, we consider the integrality problem for sums which are not necessarily symmetric with respect to their variables. For an r-tuple of positive integers s = (s,..., s r ) and an integer n r, we define two classes of multiple harmonic sums: the ordinary multiple harmonic sum (MHS) H n (s,..., s r ) = k s applek < <k rapplen and the star version (MHS-star also denoted by S in the literature) H? n(s,..., s r ) = k s applek apple applek rapplen, () ksr r. (2) ksr r The number l(s) := r is called the length and s := P r j= s j is the weight of the multiple harmonic sum. Note that H n ({m} r ) is the r-th elementary symmetric function of /f(), /f(2),..., /f(n) with f(x) = x m. The multiple sums () and (2) are of a certain interest because by taking the limit as n goes to when s r > (otherwise the infinite sums diverge) we get the so-called multiple zeta value and the multiple zeta star value, lim H n(s,..., s r ) = (s,..., s r ) and lim n! n! H? n(s,..., s r ) =? (s,..., s r ), respectively. Note that the integrality of the MHS-star is quite simple to study. Theorem. Let n of H? (s ) =. r. Then H? n(s,..., s r ) is never an integer with the exception Proof. If n >, then by Bertrand s postulate, there is at least a prime p such that n/2 < p apple n. Then p apple n < 2p and H? (s) = applek apple applek rapplen 9i : k i6=p k s ksr r + applek apple applek rapplen 8i, k i=p k s ksr r = a bp t + p s

INTEGERS: 7 (207) 3 where gcd(b, p) = and t < s. Assume that H? (s) = m 2 N +. Then bmp s = ap s t + b, which is a contradiction because p divides the left-hand side and p does not divide the right-hand side. On the other hand, the case of the ordinary MHS is much more intricate. Our result is given below whereas the entire next section is dedicated to its proof. Theorem 2. Let n r. Then H n (s,..., s r ) is never an integer with the exceptions of H (s ) = and H 3 (, ) =. Summarizing and generalizing the results of Theorem, Theorem 2 and papers [, 4, 6, 0], we propose the following problem. Problem. Let n and r be positive integers such that r apple n. Let s = (s,..., s r ) be an r-tuple of positive integers and f(x) be a nonzero polynomial of nonnegative integer coe cients, f(x) 6= x. Let H n,f (s,..., s r ) = f(k ) s f(k r ) sr and H? n,f(s,..., s r ) = applek < <k rapplen applek apple applek rapplen f(k ) s f(k r ) sr. Find all values of n and s for which H n,f (s,..., s r ) and H? n,f (s,..., s r ) are integers. Note that when f(x) = x, the multiple sums H n,f (s,..., s r ) and H? n,f (s,..., s r ) coincide with multiple harmonic sums () and (2) and therefore, the integrality problem in this case is addressed in Theorem and Theorem 2, respectively. Throughout the paper, all the numerical computations were performed by using Maple TM. 2. Proof of Theorem 2 The main tool in the proof of Theorem is Bertrand s postulate. Unfortunately, such result is not strong enough to imply Theorem 2. The statement described in the following remark will replace it. Notice that the same argument had been used by Erdős and Niven in [3] and successively taken up in [, 4] and [0]. Remark. Let apple r apple n. If there is a prime p 2 ( n r+, n r ] and r < p, then H n(s) is not an integer when l(s) = r. Indeed, since < p < 2p < < rp apple n < (r +)p, it follows that H n (s) = applek < <k rapplen 9i : p-k i k s ksr r + applek < <k rapplen 8i, p k i k s ksr r = a bp t + cp s

INTEGERS: 7 (207) 4 where gcd(b, p) = gcd(c, p) = and t < s. Assume that H n (s) = m 2 N +. Then bcmp s = cap s t + b, which is a contradiction because p divides the left-hand side and p does not divide the right-hand side. For any integer r, let A r = [ p2p[rp, (r + )p) where P is the set of primes. Note that by Bertrand s postulate, A = [2, +). The crucial property of the set A r is that n 2 A r if and only if there exists a prime p such that p 2 ( n r+, n r ]. The next lemma is a variation of [4, Lemma 2.4]. Lemma. For any positive integer r, A r is cofinite, i.e., N \ A r is finite. Let m r = max(n \ A r ) +. Then the first few values of m r are as follows: r 2-3 4-5 6-7 8 9-3 4-6 m r /r 2 29 37 53 27 49 r 7 8-20 2-22 23-29 30-39 40-69 m r /r 2 223 307 33 54 36 and for any positive integer r 24, m r apple (r + ) exp( p.4 r). Proof. Let r and let n be such that x := n/(r + ) max(exp( p.4 r), 3275). Then ln 2 (x).4 r implies n r = x + x r x +.4 x ln 2 (x) x + x 2 ln 2 (x). Moreover, by [2, Theorem ], there is a prime p such that x n p 2 x, x + 2 ln 2 (x) r +, n r which implies that n 2 A r. This proves that N \ A r is finite. Note that if r 47, then exp( p.4 r) 3275. On the other hand, if 24 apple r apple 46 and n 2 [(r + ) exp( p.4 r), (r + )3275), one can verify directly that n 2 A r. In order to compare values of multiple harmonic sums of the same length, the following definition and lemma will be useful. Definition. Let s = (s,..., s r ) and t = (t,..., t r ) be two r-tuples of positive integers. We say that s t, if w(s) w(t), i.e., if s + + s r t + + t r, and s apple t,..., s l apple t l, s l+ t l+,..., s r t r for some 0 apple l apple r. In particular, s t if s i t i for i =,..., r.

INTEGERS: 7 (207) 5 Lemma 2. Let s = (s,..., s r ) and t = (t,..., t r ) be two r-tuples of positive integers, and s t. Then for any positive integer n, H n (s,..., s r ) apple H n (t,..., t r ). Proof. Let 0 apple l apple r be such that s apple t,..., s l apple t l and s l+ t l+,..., s r t r. Now we compare corresponding terms k s ksr r and, where k < < k r, ktr r k t of multiple sums H n (s) and H n (t). Since s + + s r k t s t2 s2 k and therefore, 2 k t l s l l apple k t+ +t l (s + +s l ) l = k (s l+ t l+ )+ +(s r t r) l k apple k s l+ t l+ l+ k k s apple ksr r which implies H n (s) apple H n (t), as required. Remark 2. Notice that H n () = + n k=2 k t sr tr r, ktr r Hence, if apple r apple n then, by the previous lemma, Z n k apple + dx = ln(n) +. x H n (s,..., s r ) apple H n ({} r ) apple (H n()) r r! apple t + + t r, we have t+ +tr (s+ +sr) l (ln(n) + )r r! where the second inequality holds because each term of H n ({} r ) is contained r! times in the expansion of (H n ()) r. Let p be a prime and let p (q) be the p-adic order of the rational number q, that is, if a, b are coprime with p and n 2 Z, then p (ap n /b) = n. It is known that the p-adic order satisfies the inequality p (a + b) min( p (a), p (b)) where the equality holds if p (a) 6= p (b). The following lemma will be our basic tool to find an upper bound for the index s. Lemma 3. Given 2 apple r apple n and (s 2,..., s r ), there exists an integer M (which depends on (s 2,..., s r ) and n) such that H n (s, s 2,..., s r ) is never an integer for any positive integer s > M.

INTEGERS: 7 (207) 6 Proof. If r = n, then H n (s, s 2,..., s r ) is trivially not an integer for all s > 0. If r < n, then we have that H n (s,..., s r ) = nr+ k= c k k s where c k := k<k 2< <k rapplen k s2 2 > 0. ksr r Let p be the largest prime in [2, n r + ]. Then 2p > n r + (otherwise by Bertrand s postulate there is a prime q such that p < q < 2p apple n r + ). Let 0 B M := max @ p (c p ), p (c p ) min applekapplen r+ C ( p (c k )) A. We will show that p (H n (s)) < 0, which implies that H n (s) is not an integer. Assume that s > M, then Now i) s > p (c p ) and ii) s > p (c p ) min By ii), we have that 0 p B @ applekapplen r+ and by i), Therefore c k k s C A H n (s) = c p p s + min applekapplen r+ 0 p (H n (s)) = min B @ B p @ applekapplen r+ c k ck p = min k s k s. applekapplen r+ cp p = p (c p ) s < 0. p s 0 applekapplen r+ c k k s C A, p cp p s applekapplen r+ p (c k ) > p (c p ) C A = p cp p s ( p (c k )). < 0. cp s = p, p s Remark 3. If, in addition to the inequalities 2 apple p apple n r +, the prime p satisfies n/2 < p, then ( p (c k ) min max 2appleiappler s i, if apple k < p, p k<k 2< <k rapplen k s2 2 = ksr r 0, if p apple k apple n r +.

INTEGERS: 7 (207) 7 Therefore This means that min applekapplen r+ ( p (c k )) max s i. 2appleiappler M 0 := p (c p ) + max 2appleiappler s i M. When the assumption n/2 < p is met, we will prefer to use the bound M 0 to M because the bound M 0 is computationally easier to determine. Proof of Theorem 2. Since H n (s ) = H n(s? ), by Theorem, the statement holds for r =. If e(ln(n) + ) apple r, then by Remark 2 and the fact that e r > r r /r!, we have r (ln(n) + )r ln(n) + H n (s) apple < apple, r! r/e which implies that H n (s) can not be an integer. Assume that 2 apple r < e(ln(n) + ). Then exp(r/e ) < n and, since it can be verified that (r + ) exp( p.4 r) apple exp(r/e ) for r 30, it follows that n 2 A r and we are done as soon as we use Remark. Remark can be applied successfully even when 2 apple r apple 29 and n 9599 because m r apple 33r apple 33 29 = 9599 apple n. Hence the cases remained to consider are 2 apple r apple 29 and r apple n < m r (with n 62 A r ). For r = 25, 26, 27, 28, 29, by Remark 2, H n (s) < H mr (s) apple H mr ({} r ) apple (ln(m r) + ) r where the last inequality can be easily verified numerically. For r = 9, 20, 2, 22, 23, 24, by Lemma 2, r! < H n (s) < H mr (s) apple H mr ({} r ) and the non-integrality of H n (s) is implied by the following evaluations: H m24 ({} 24 ) < 0.025084028 <, H m23 ({} 23 ) < 0.068740285 <, H m22 ({} 22 ) < 0.45564965 <, H m2 ({} 2 ) < 0.369820580 <, H m20 ({} 20 ) < 0.379560254 <, H m9 ({} 9 ) < 0.96202538 <.

INTEGERS: 7 (207) 8 The strategy to handle the cases where 2 apple r apple 8 is fairly more complicated because H mr ({} r ) >. The analysis is based on the numerical values presented in Tables, 2, and 3. Here we give a detailed explanation of how the data in such tables are calculated and used for r = 5. The other cases can be treated in a similar way. What turns out at the end is that the only exception for r 2 is H 3 (, ) =. We first determine the optimal set of length 5, that is, a set of 5-tuples such that the multiple harmonic sums with n = m 5 = 45 are the largest sums less than with small weights (columns 2 and 3): H m5 ({} 4, 2) < 0.502399297 <, H m5 (, 2, 2,, ) < 0.8508767 <, H m5 (, 4, {} 3 ) < 0.88376754 <. Then, thanks to Lemma 2, the size of the set of multiple harmonic sums less than (and therefore not integral) can be extended. Let 5 apple n < m 5. If s 2 4, then H n (s, s 2, s 3, s 4, s 5 ) apple H n (, 4, {} 3 ) < H m5 (, 4, {} 3 ) <. If s 2 2 {2, 3} and there is s j 2 with 3 apple j apple 5, then H n (s, s 2, s 3, s 4, s 5 ) apple H n (, 2, s 3, s 4, s 5 ) apple H n (, 2, 2,, ) < H m5 (, 2, 2,, ) <. If s 2 = and s 3 3, then H n (s,, s 3, s 4, s 5 ) apple H n (,, 3,, ) apple H n (, 2, 2,, ) < H m5 (, 2, 2,, ) <. If s 2 =, s 3 = 2, and s 4 2 or s 5 2, then H n (s,, 2, s 4, s 5 ) apple H n (,, 2, s 4, s 5 ) apple H n (,, 2, 2, ) apple H n (, 2, 2,, ) < H m5 (, 2, 2,, ) <. If s 2 =, s 3 =, and s 4 3, then H n (s,,, s 4, s 5 ) apple H n ({} 3, 3, ) apple H n (, 2, 2,, ) < H m5 (, 2, 2,, ) <. If s 2 =, s 3 =, s 4 = 2, and s 5 2, then H n (s,,, 2, s 5 ) apple H n ({} 3, 2, s 5 ) apple H n ({} 3, 2, 2) apple H n (, 2, 2, {} 2 ) < H m5 (, 2, 2, {} 2 ) <. If s 2 = s 3 = s 4 = and s 5 2, then H n (s, {} 3, s 5 ) apple H n ({} 4, s 5 ) apple H n ({} 4, 2) < H m5 ({} 4, 2) <.

INTEGERS: 7 (207) 9 The set of 5-tuples of positive integers which are excluded by the analysis above is what we call the exclusion set (column 4): (s, 3, {} 3 ), (s, 2, {} 3 ), (s,, 2,, ), (s,,, 2, ), (s, {} 4 ) with s. By using Lemma 3 and Remark 3, we are able to give an upper bound for s (column 5) and therefore to reduce the size of the exclusion set to a finite number. Finally, it su ces to compute the rational number H n (s) for 5 apple n < m 5 and for s in the exclusion set with the upper bound for s established in the last column. For r = 5, none of them is an integer. Acknowledgement. Kh. Hessami Pilehrood and T. Hessami Pilehrood gratefully acknowledge support from the Fields Institute Research Immersion Fellowships. References [] Y. Chen and M. Tang, On the elementary symmetric functions of, /2,..., /n, Amer. Math. Monthly 9 (202), 862-867. [2] P. Dusart, Inégalités explicites pour (x), (x), (x) et les nombres premiers, C. R. Math. Acad. Sci. Soc. R. Can. 2 (999), 53-59. [3] P. Erdős and I. Niven, Some properties of partial sums of the harmonic series, Bull. Amer. Math. Soc. 52 (946), 248-25. [4] S. Hong and C. Wang, The elementary symmetric functions of reciprocals of elements of arithmetic progressions, Acta Math. Hungar. 44 (204), 96-2. [5] K. Ireland and M. Rosen, A Classical Introduction to Modern Number Theory, Springer- Verlag, 990. [6] Y. Luo, S. Hong, G. Qian, C. Wang The elementary symmetric functions of a reciprocal polynomial sequence, C. R. Math. Acad. Sci. Paris 352 (204), 269-272. [7] M. B. Nathanson, Elementary Methods in Number Theory, Springer-Verlag, 2000. [8] G. Pólya and G. Szegő, Problems and Theorems in Analysis II, Springer-Verlag, 976. [9] L. Theisinger, Bemerkung über die harmonische Reihe, Monatsh. Math. 26 (95), 32-34. [0] C. Wang and S. Hong, On the integrality of the elementary symmetric functions of, /3,..., /(2n ), Math. Slovaca 65 (205), 957-962.

INTEGERS: 7 (207) 0 Table : Optimal sets, exclusion sets and upper bounds for 2 apple r apple 8. Length Optimal set of Upper bound for Exclusion set Upper bound r (s, s 2,..., s r ) H mr (s, s 2,..., s r ) with s for s 2 (, 2) < 0.99409932 (s, ) apple 3 3 (,, 2) < 0.70626068 (s, 2, ) apple 3 (, 3, ) < 0.589038 (s,, ) apple 3 (,,, 2) < 0.6846275 (s, 3,, ) apple 4 4 (, 2, 2, ) < 0.537884954 (s, 2,, ) apple 3 (, 4,, ) < 0.627965284 (s,, 2, ) apple 3 (s,,, ) apple 3 (s, 3, {} 3 ) apple 4 ({} 4, 2) < 0.502399297 (s, 2, {} 3 ) apple 3 5 (, 2, 2,, ) < 0.8508767 (s,, 2,, ) apple 3 (, 4, {} 3 ) < 0.88376754 (s,,, 2, ) apple 3 (s, {} 4 ) apple 3 (s, 4, {} 4 ) apple 5 ({} 4, 2, ) < 0.8606229 (s, 3, {} 4 ) apple 4 (,, 3, {} 3 ) < 0.6757688 (s, 2, 2, {} 3 ) apple 3 6 (, 2,, 2,, ) < 0.5729690 (s, 2, {} 4 ) apple 3 (, 3, 2, {} 3 ) < 0.50634468 (s,, 2, {} 3 ) apple 3 (, 5, {} 4 ) < 0.56379826 (s,,, 2,, ) apple 3 (s, {} 5 ) apple 2 (s, 4, {} 5 ) apple 5 ({} 4, 2,, ) < 0.998935309 (s, 3, {} 5 ) apple 4 (,, 3, {} 4 ) < 0.7246509 (s, 2, 2, {} 4 ) apple 3 7 (, 2,, 2, {} 3 ) < 0.637523826 (s, 2, {} 5 ) apple 3 (, 3, 2, {} 4 ) < 0.53072055 (s,, 2, {} 4 ) apple 3 (, 5, {} 5 ) < 0.55328509 (s,,, 2, {} 3 ) apple 4 (s, {} 6 ) apple 2 (s, 4, {} 6 ) apple 5 ({} 5, 2,, ) < 0.7564380 (s, 3, {} 6 ) apple 4 (,, 3, {} 5 ) < 0.985465806 (s, 2, 2, {} 5 ) apple 5 8 (, 2,, 2, {} 4 ) < 0.896426675 (s, 2, {} 6 ) apple 3 (, 3, 2, {} 5 ) < 0.733626794 (s,, 2, {} 5 ) apple 4 (, 5, {} 6 ) < 0.750552945 (s,,, 2, {} 4 ) apple 4 (s, {} 3, 2, {} 3 ) apple 2 (s, {} 6 ) apple 2

INTEGERS: 7 (207) Table 2: Optimal sets, exclusion sets and upper bounds for 9 apple r apple. Length Optimal set of Upper bound for Exclusion set Upper bound r (s, s 2,..., s r ) H mr (s, s 2,..., s r ) with s for s (s, 5, {} 7 ) apple 7 (s, 4, {} 7 ) apple 5 (s, 3, 2, {} 6 ) apple 6 ({} 6, 2,, ) < 0.92597697 (s, 3, {} 7 ) apple 4 (, 2, {} 3, 2, {} 3 ) < 0.57075733 (s, 2, 2, {} 6 ) apple 3 (,, 2,, 2, {} 4 ) < 0.59649570 (s, 2,, 2, {} 5 ) apple 4 9 (,,, 3, {} 5 ) < 0.70790000 (s, 2,,, 2, {} 4 ) apple 3 (, 3,, 2, {} 5 ) < 0.88465084 (s, 2, {} 7 ) apple 4 (, 2, 3, {} 6 ) < 0.9797067 (s,, 3, {} 6 ) apple 4 (, 4, 2, {} 6 ) < 0.8498342 (s,, 2, 2, {} 5 ) apple 3 (, 6, {} 7 ) < 0.89773579 (s,, 2, {} 6 ) apple 3 (s,,, 2, {} 5 ) apple 3 (s, {} 3, 2, {} 4 ) apple 3 (s, {} 4, 2, {} 3 ) apple 3 (s, {} 8 ) apple 2 (s, 5, {} 8 ) apple 6 (s, 4, {} 8 ) apple 6 (s, 3, 2, {} 7 ) apple 4 ({} 6, 2, {} 3 ) < 0.93845772 (s, 3, {} 8 ) apple 4 (, 2, {} 3, 2, {} 4 ) < 0.56773422 (s, 2, 2, {} 7 ) apple 4 (,, 2,, 2, {} 5 ) < 0.558322794 (s, 2,, 2, {} 6 ) apple 4 0 (,,, 3, {} 6 ) < 0.644468502 (s, 2,,, 2, {} 5 ) apple 3 (, 3,, 2, {} 6 ) < 0.787882830 (s, 2, {} 8 ) apple 3 (, 2, 3, {} 7 ) < 0.847077826 (s,, 3, {} 7 ) apple 4 (, 4, 2, {} 7 ) < 0.697660582 (s,, 2, 2, {} 6 ) apple 3 (, 6, {} 8 ) < 0.7393695 (s,, 2, {} 7 ) apple 4 (s,,, 2, {} 6 ) apple 3 (s, {} 3, 2, {} 5 ) apple 4 (s, {} 4, 2, {} 4 ) apple 3 (s, {} 9 ) apple 3 (s, 5, {} 9 ) apple 8 (s, 4, {} 9 ) apple 5 (s, 3, 2, {} 8 ) apple 6 ({} 6, 2, {} 4 ) < 0.84836649 (s, 3,, {} 8 ) apple 5 (, 2,,, 2, {} 6 ) < 0.84877884 (s, 2, 2, {} 8 ) apple 4 (,, 2, 2, {} 7 ) < 0.836565448 (s, 2,, 2, {} 7 ) apple 3 (, 3,, 2, {} 7 ) < 0.62949675 (s, 2,,, {} 7 ) apple 3 (, 2, 3, {} 8 ) < 0.653475 (s,, 3, {} 8 ) apple 5 (, 4, 2, {} 8 ) < 0.533753072 (s,, 2, {} 8 ) apple 3 (, 6, {} 9 ) < 0.548020075 (s,,, 2, {} 7 ) apple 3 (s, {} 3, 2, {} 6 ) apple 3 (s, {} 4, 2, {} 5 ) apple 3 (s, {} 0 ) apple 3

INTEGERS: 7 (207) 2 Table 3: Optimal sets, exclusion sets and upper bounds for 2 apple r apple 8. Length Optimal set of Upper bound for Exclusion set Upper bound r (s, s 2,..., s r ) H mr (s, s 2,..., s r ) with s for s (s, 4, {} 0 ) apple 5 (s, 3, {} 0 ) apple 4 ({} 6, 2, {} 5 ) < 0.622525355 (s, 2, 2, {} 9 ) apple 3 (, 2,,, 2, {} 7 ) < 0.6346842 (s, 2,, 2, {} 8 ) apple 4 2 (,, 2, 2, {} 8 ) < 0.59533583 (s, 2, {} 0 ) apple 3 (, 3, 2, {} 9 ) < 0.86806820 (s,, 3, {} 9 ) apple 4 (, 5, {} 0 ) < 0.779899998 (s,, 2, {} 9 ) apple 4 (s,,, 2, {} 8 ) apple 3 (s, {} 3, 2, {} 7 ) apple 3 (s, {} 4, 2, {} 6 ) apple 3 (s, {} ) apple 3 (s, 4, {} ) apple 5 ({} 5, 2, {} 7 ) < 0.68409082 (s, 3, {} ) apple 4 (, 2,, 2, {} 9 ) < 0.678679596 (s, 2, 2, {} 0 ) apple 4 3 (,, 3, {} 0 ) < 0.694738378 (s, 2, {} ) apple 3 (, 3, 2, {} 0 ) < 0.54442958 (s,, 2, {} 0 ) apple 3 (, 5, {} ) < 0.48478855 (s,,, 2, {} 9 ) apple 3 (s, {} 3, 2, {} 8 ) apple 3 (s, {} 2 ) apple 2 (s, 3, {} 2 ) apple 4 ({} 4, 2, {} 9 ) < 0.887408969 (s, 2, {} 2 ) apple 3 4 (, 2, 2, {} ) < 0.9578432 (s,, 2, {} ) apple 3 (, 4, {} 2 ) < 0.8823535 (s,,, 2, {} 0 ) apple 3 (s, {} 3 ) apple 3 ({} 3, 2, {} ) < 0.79976532 (s, 3, {} 3 ) apple 4 5 (, 2, 2, {} 2 ) < 0.52690836 (s, 2, {} 3 ) apple 3 (, 4, {} 3 ) < 0.443394022 (s,, 2, {} 2 ) apple 3 (s, {} 4 ) apple 2 6 (,, 2, {} 3 ) < 0.68305336 (s, 2, {} 4 ) apple 3 (, 3, {} 4 ) < 0.50957243 (s, {} 5 ) apple 3 7 (,, 2, {} 4 ) < 0.732770497 (s, 2, {} 5 ) apple 3 (, 3, {} 5 ) < 0.54674269 (s, {} 6 ) apple 3 8 (, 2, {} 6 ) < 0.722868056 (s, {} 7 ) apple 3