login
A322165
Numbers k that give record values for s(k)*phi(k)/k^2, where s(k) is the sum of squares of the differences between consecutive totatives of k (A322144).
0
1, 3, 4, 6, 10, 12, 15, 18, 20, 21, 30, 42, 60, 70, 105, 210, 385, 770, 1155, 2310, 4620, 5005, 10010, 15015, 30030, 60060, 90090, 120120, 150150, 180180, 210210, 240240, 255255, 510510
OFFSET
1,2
COMMENTS
Erdős conjectured that this ratio is bounded and offered $500 for a proof. The conjecture was proved by Montgomery and Vaughan, who won the prize.
Is this sequence infinite? If yes, what is lim_{n->oo} s(a(n))*phi(a(n))/a(n)^2?
REFERENCES
Richard K. Guy, Unsolved Problems in Number Theory, 3rd ed., Springer, 2004, Chapter B40, Gaps between totatives, p. 146.
LINKS
Paul Erdős, The difference of consecutive primes, Duke Mathematical Journal, Vol. 6, No. 2 (1940), pp. 438-441, alternative link.
H. L. Montgomery & R. C. Vaughan, On the distribution of reduced residues, Annals of Mathematics, Vol. 123, No. 2 (1986), pp. 311-333.
EXAMPLE
The values of the ratio at the first terms of the sequence are 0, 0.222..., 0.5, 0.888..., 0.96, 1, 1.031..., ...
MATHEMATICA
ratio[n_] := Module[{v=Differences[Select[Range[n], GCD[n, #] == 1 &]]^2}, Total[v] * (Length[v]+1) / n^2]; seq={}; rm=-1; Do[r=ratio[n]; If[r>rm, rm=r; AppendTo[seq, n]], {n, 1, 1000}]; seq
PROG
(PARI) s(n) = {v = select(x->gcd(x, n)==1, vector(n, k, k)); sum(i=1, #v-1, (v[i+1] - v[i])^2); } \\ A322144
lista(nn) = {my(m = -1); for (n=1, nn, newm = s(n)*eulerphi(n)/n^2; if (newm > m, print1(n, ", "); m = newm); ); } \\ Michel Marcus, Nov 29 2018
CROSSREFS
KEYWORD
nonn,more
AUTHOR
Amiram Eldar, Nov 29 2018
STATUS
approved