On the prime power factorization of \(n\)! (Q5902111)

Let \(q_1,~\ldots,~q_k\) be distinct primes. For a prime \(q\) and a positive integer \(n\) let \(e_q(n)\) be the exponent of \(q\) in the factorization of \(n!\). Recently, there has been a lot of interest in studying the distribution of the numbers \(e_{q_1}(n),\ldots,e_{q_k}(n)\) in arithmetic progressions modulo \(m_1,\ldots,m_k\), respectively. Inspired by a problem of Erdős and Graham, \textit{J. W. Sander} conjectured in 2001 [J. Number Theory 90, No. 2, 316--328 (2001; Zbl 1024.11010)] that the above distribution should be uniform. This was proved to be so under some technical assumption by \textit{F. Luca} and \textit{P. Stănică} in [J. Number Theory 102, No. 2, 298--305 (2003; Zbl 1049.11092)], and in its full generality by \textit{D. Berend} and \textit{G. Kolesnik} in 2007 [J. Number Theory 124, No. 1, 181--192 (2007; Zbl 1124.11036)]. In the paper under review, the author sharpens the above results by providing a good upper bound on the error term \[ E(x;{\mathbf m},{\mathbf q},{\mathbf r})=A(x;{\mathbf m}, {\mathbf q},{\mathbf r})-{{x}\over {m_1\cdots m_k}}, \] where \({\mathbf q}=(q_1,\ldots,q_k)\),~\({\mathbf m}=(m_1,\ldots,m_k)\), \({\mathbf r}=(r_1,\ldots,r_k)\) is a vector of residue classes modulo \(m_1,\ldots,m_k\), respectively, and \(A(x;{\mathbf m},{\mathbf q},{\mathbf r})\) counts the number of positive integers \(n\leq x\) satisfying \(e_{q_i}\equiv r_i\pmod {m_i}\) for all \(i=1,\ldots,k\). The proof of this result is based on a bound for an exponential sum with the numbers \(e_q(n)\). The author also shows that if \(m\) is fixed and \(q\) is sufficiently large with respect to \(m\), then the numbers \(e_q(n^2)\) cover all residue classes modulo \(m\) as \(n\) ranges over all the positive integers. The paper concludes with a few conjectures regarding the distribution of the numbers \(e_{q_i}(n^{l_i})\) modulo \(m_i\) for \(i=1,\ldots,k\), where \(l_1,\ldots,l_k\) are fixed positive integers.