On a problem of Arnold: the average multiplicative order of a given integer (Q365900)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On a problem of Arnold: the average multiplicative order of a given integer
scientific article

    Statements

    On a problem of Arnold: the average multiplicative order of a given integer (English)
    0 references
    0 references
    0 references
    9 September 2013
    0 references
    For coprime integers \(g\) and \(n\), let \(l_g(n)\) denote the multiplicative order of \(g\) modulo \(n\), that is, the smallest integer \(k\geq 1\) such that \(g^k\equiv 1(\text{mod~}n)\). The Carmichael function \(\lambda(n)\) is the exponent of the group of invertible residue classes modulo \(n\). We have \(l_g(n)|\lambda(n)\). Let \(T_g(x)\) be the average of \(l_g(n)\) as \(n\leq x\) ranges over integers coprime to \(g\). \textit{V. Arnold} [J. Math. Fluid Mech. 7, Suppl. 1, S4--S50 (2005; Zbl 1134.11344)] conjectured that if \(|g|>1\), then \(T_g(x)\sim c(g)x/\log x\) as \(x\) tends to infinity and with \(c(g)>0\) some constant. Assuming GRH, this conjecture was disproved by \textit{I. E. Shparlinski} [Discrete Contin. Dyn. Syst. 17, No. 4, 901--917 (2007; Zbl 1127.11052)]. In the present paper, also assuming GRH, the true asymptotic behaviour of \(T_g(x)\) is determined (sharpening the work of Shparlinski) and it is shown that it is the same as that of \(x^{-1}\sum_{n\leq x}\lambda(n)\). Since the main contribution to average of \(\lambda(n)\) comes from a density-zero subset of the integers, the authors have to show that \(l_g(n)\) is large for many \(n\) for which \(\lambda(n)\) is large. A second main theorem, also assuming GRH, gives the asymptotics of the ratio \[ \sum_{p\leq x}l_g(p)/\sum_{p\leq x}1, \] where the sums are over the primes \(p\leq x\). The latter ratio is asymptotically equal to \(x\) times a complicated constant that equals a rational number depending on \(g\) times the universal constant \[ S:=\prod_p\big(1-{p\over p^3-1}\big)=0.57595996889\ldots \] where the product is over all primes \(p\). A weaker result in this direction had already been obtained by \textit{P. J. Stephens} [J. Number Theory 8, 313--332 (1976; Zbl 0334.10018]. The reviewer and \textit{P. Stevenhagen} [J. Number Theory 85, No. 2, 291--304 (2000; Zbl 0966.11042)] provided some corrections to the results of Stephens.
    0 references
    0 references
    0 references
    average multiplicative order
    0 references
    0 references
    0 references