login

Year-end appeal: Please make a donation to the OEIS Foundation to support ongoing development and maintenance of the OEIS. We are now in our 61st year, we have over 378,000 sequences, and we’ve reached 11,000 citations (which often say “discovered thanks to the OEIS”).

Denominator of the fraction (Sum_{prime p | n} 1/p - 1/n).
15

%I #64 Dec 31 2023 10:24:21

%S 1,1,1,4,1,3,1,8,9,5,1,4,1,7,15,16,1,9,1,20,7,11,1,24,25,13,27,28,1,1,

%T 1,32,33,17,35,36,1,19,13,40,1,21,1,44,45,23,1,16,49,25,51,52,1,27,11,

%U 8,19,29,1,60,1,31,63,64,65,11,1,68,69,35,1,72

%N Denominator of the fraction (Sum_{prime p | n} 1/p - 1/n).

%C Theorem. If n is a prime or a Carmichael number, then a(n) = A309132(n) = denominator of (N(n-1)/n + D(n-1)/n^2), where B(k) = N(k)/D(k) is the k-th Bernoulli number. This is a generalization of Theorem 1 in A309132 that A309132(p) = 1 if p is a prime. The proof generalizes that in A309132. As an application of Theorem, for n a prime or a Carmichael number one can compute A309132(n) without calculating Bernoulli numbers; see A309268.

%C A composite number n is a Giuga number A007850 if and only if a(n) = 1. (In fact, Sum_{prime p | n} 1/p - 1/n = 1 for all known Giuga numbers n.)

%C Semiprimes m = pq such that 1/p + 1/q - 1/m = p/q are exactly A190275. - _Amiram Eldar_ and _Thomas Ordowski_, Jul 22 2019

%C The preceding comment may be rephrased as "Semiprimes m = pq such that A326689(m) = p and a(m) = q are exactly A190275." - _Jonathan Sondow_, Jul 22 2019

%C More generally, semiprimes m = pq such that 1/p + 1/q - 1/m = P/Q are exactly A190273, where P <> Q are primes. In other words, semiprimes m such that A326689(m) is prime and a(m) is prime are exactly A190273. - _Amiram Eldar_ and _Thomas Ordowski_, Jul 25 2019

%H Antti Karttunen, <a href="/A326690/b326690.txt">Table of n, a(n) for n = 1..16384</a>

%H Antti Karttunen, <a href="/A326690/a326690.txt">Data supplement: n, a(n) computed for n = 1..65537</a>

%H Wikipedia, <a href="https://en.wikipedia.org/wiki/Bernoulli_number">Bernoulli number</a>

%H Wikipedia, <a href="https://en.wikipedia.org/wiki/Carmichael_number">Carmichael number</a>

%H Wikipedia, <a href="https://en.wikipedia.org/wiki/Giuga_number">Giuga number</a>

%F a(n) = 1 if n is a prime or a Giuga number A007850.

%F a(n) = denominator of (N(n-1)/n + D(n-1)/n^2) if n is a Carmichael number A002997.

%F a(n) = denominator((A069359(n) - 1)/n). - _Peter Luschny_, Jul 22 2019

%e -1/1, 0/1, 0/1, 1/4, 0/1, 2/3, 0/1, 3/8, 2/9, 3/5, 0/1, 3/4, 0/1, 4/7, 7/15, 7/16, 0/1, 7/9, 0/1, 13/20, 3/7, 6/11, 0/1, 19/24, 4/25, 7/13, 8/27, 17/28, 0/1, 1/1

%e a(12) = denominator of (Sum_{prime p | 12} 1/p - 1/12) = denominator of (1/2 + 1/3 - 1/12) = denominator of 3/4 = 4.

%e Computing A309132(561) involves numerator(B(560)) which has 865 digits. But 561 is a Carmichael number, so Theorem implies A309132(561) = a(561) = denominator(1/3 + 1/11 + 1/17 - 1/561) = denominator(90/187) = 187.

%p A326690 := n -> denom((A069359(n)-1)/n):

%p seq(A326690(n), n=1..72); # _Peter Luschny_, Jul 22 2019

%t PrimeFactors[n_] := Select[Divisors[n], PrimeQ];

%t f[n_] := Denominator[Sum[1/p, {p, PrimeFactors[n]}] - 1/n];

%t Table[ f[n], {n, 100}]

%o (PARI) a(n) = denominator(sumdiv(n, d, isprime(d)/d) - 1/n); \\ _Michel Marcus_, Jul 19 2019

%o (SageMath)

%o p = lambda n: [n//f[0] for f in factor(n)]

%o A326690 = lambda n: ((sum(p(n)) - 1)/n).denominator()

%o [A326690(n) for n in (1..72)] # _Peter Luschny_, Jul 22 2019

%o (Magma) [1] cat [Denominator(&+[1/p:p in PrimeDivisors(k)]-1/k):k in [2..72]]; // _Marius A. Burtea_, Jul 27 2019

%Y Numerators are A326689. Quotients n/a(n) are A326691.

%Y Cf. A069359, A007947 (denominator of Sum_{prime p | n} 1/p).

%Y Cf. A002997, A007850, A190275, A309132, A309235, A309268, A309378, A326692.

%K nonn,frac

%O 1,4

%A _Jonathan Sondow_, Jul 18 2019