Distribution of the number of factors in random ordered factorizations of integers (Q1970618)

For \(n\geq 2\), let \(a(n)\) denote the number of different ordered factorizations of \(n\) with each factor exceeding 1, and put \(a(1)=1\). Define \(\rho\in (1,2)\) by \(\zeta(\rho)= 2\). It is known that \(A(x):= \sum_{n\leq x}a(n)= -(\rho\zeta'(\rho))^{-1} x^\rho+ E(x)\), where \(E(x)\) is an error term for which estimates were derived in the 1930's, first by L. Kalmár and then by S. Ikehara. By strengthening the upper bounds established by these authors for \(|\zeta(\sigma+it)|\) for \(\sigma>1\), \(|t|\geq t_0\), the present author shows that \(E(x)\ll x^\rho\exp (-c(\log\log x)^{\frac 32- \varepsilon})\). The main objective of this paper is to investigate the number \(A(x,m)\) of ordered factorizations of \(n\leq x\) into exactly \(m\) factors exceeding 1, where \(m\) lies in a range depending on \(x\). An asymptotic formula for \(A(x,m)\) is established for \(1\leq m= o((\log x)^{3/5})\) in Theorem 2 and for \((\log x)^{1/2}\leq m\leq \frac{\log x}{\log 2} (1-c\exp(- (\log\log x)^{\frac 13- \varepsilon}))\) in Theorem 3 by different analytic arguments, and an upper bound is obtained for the remaining \(m\) using Rankin's method. The first theorem establishes that (assuming a uniform probability measure) the distribution of the number of factors in a random ordered factorization is asymptotically normal. The expected value of the number of distinct factors in a random ordered factorization is derived in Theorem 4, thus correcting, as stated by the author, a result mentioned in [Adv. Appl. Math. 19, 378-414 (1997; Zbl 0943.05009)] by \textit{H.-K. Hwang} and \textit{Y.-N. Yeh}. As this value is much smaller than the corresponding value when the factors are not necessarily distinct, the author concludes that ``there are many small factors occurring with high frequencies in a random ordered factorization''.