The distribution of square-full integers. II (Q746937): Difference between revisions

A square-full positive integer \(n\) satisfies \(p^2|n\) whenever the prime \(p|n\). Let \(Q(x)\) denote the number of square-full integers \(n\leq x\). The best known unconditional result is of the form \(Q(x)= Ax^{1/2}+ Bx^{1/3}+ o(x^{1/6})\), where \(A\), \(B\) are specific constants. Various authors have obtained a smaller error term of the shape \(O(x^{\theta+\varepsilon})\) under the assumption of the Riemann hypothesis for some \(\theta\) and \(\varepsilon>0\). \textit{J. Wu} [Arch. Math. 77, No. 3, 233--240 (2001; Zbl 1003.11043)] reduced the value of \(\theta\) under the Riemann hypothesis to \(\theta={12\over 85}= 0.1411\dots\). The aim of the current paper is to prove that under the R. H. \(\theta= {121\over 860}= 0.1406\dots\) is allowable. The proof depends on deriving two key results. The first in Theorem 3 concerns a new estimate for the number of lattice points satisfying multiple conditions. This is then used in Theorem 2 to bound a triple exponential sum by a sum of 12 terms. In Wu's paper [loc. cit.] the error term is expressed in terms of a sum involving \(\psi(t) =\{t\}-{1\over 2}\), and in the current paper this sum is estimated by using standard methods together with the key result in Theorem 2, which leads to the smaller value of \(\theta\). Part I, see Ark. Mat. 32, No.2, 449--454 (1994; Zbl 0827.11055); see also ``On the distribution of squarefull numbers'', (in English), J. Harbin Inst. Technol. 30, No. 5, 4--6 (1998).