|
|
A233466
|
|
Numbers k such that phi(k) = (k-5)/2.
|
|
0
|
|
|
|
OFFSET
|
1,1
|
|
COMMENTS
|
According to the following theorem I discovered recently, the 20-digit number 18446744047939747845 is in the sequence.
Theorem: If k and m are integers, k < 6, and p = 2^2^k + m is a prime such that p does not divide 2^2^k-1, then p*(2^2^k-1) is a solution to the equation phi(x) = (x+m)/2.
Note that m cannot be -1 and for k < 6, 2^2^k-1 is the product of the first k Fermat primes.
Take m=-5; since 2^2^2-5, 2^2^3-5 and 2^2^5-5 are prime we get three terms of the sequence.
Take m=1; since 2^2^0+1, 2^2^1+1, 2^2^2+1, 2^2^3+1 and 2^2^4+1 are prime (Fermat primes) we get five terms of the sequence A050474.
Conjecture (i): There is no solution to the equation phi(x) = (x-1)/2.
Conjecture (ii): The sequence has only three terms and a(3) = (2^2^5-5) * (2^2^5-1) = 18446744047939747845.
Conjecture (i) is a part of Lehmer's totient problem. Conjecture (ii) is disproved with the term a(3) = 6992962170388485 = 3 * 5 * 17 * 353 * 929 * 83623931. - Max Alekseyev, Oct 28 2023
a(5) <= 202317618492499837497376768005 = 3 * 5 * 17 * 257 * 65951 * 10414721 * 4494603392933. - Max Alekseyev, Oct 30 2023
|
|
LINKS
|
|
|
EXAMPLE
|
phi(165) = 80 = (165-5)/2.
|
|
MATHEMATICA
|
Do[If[EulerPhi[n]==1/2(n-5), Print[n]], {n, 1, 70001, 4}]
|
|
PROG
|
(Python)
from itertools import islice, count
from sympy import totient
def A233466gen(): return filter(lambda n:2*totient(n) == n-5, count(1, 2))
|
|
CROSSREFS
|
|
|
KEYWORD
|
nonn,more,hard,nice
|
|
AUTHOR
|
|
|
EXTENSIONS
|
Missing term a(3) inserted, a(4) confirmed by Max Alekseyev, Oct 28 2023
|
|
STATUS
|
approved
|
|
|
|