LATTICE POINTS CLOSE TO A SMOOTH CURVE AND SQUAREFULL NUMBERS IN SHORT INTERVALS (Q3151102)

If \(Q(x)\) denotes the number of square-full integers up to \(x\) then it is known that \[ Q(x+x^{1/2+\theta})- Q(x)\sim \frac{\zeta(3/2)} {2\zeta(3)} x^\theta \] for suitable exponents \(\theta< 1/2\). Various authors have examined this problem, and the best result to date allows any \(\theta> 1/8\). This is due to \textit{M. N. Huxley} and the author [Math. Proc. Camb. Philos. Soc. 119, 201-208 (1996; Zbl 0852.11046)]. The present paper extends the admissible range further by permitting any \(\theta> 19/154= 0.1233\ldots\). The proof of this result reduces to counting integer points close to the curve \(a^2b^3= x\). Several approaches to problems of this type have been used in the past, and here the author develops a method of \textit{H. P. F. Swinnerton-Dyer} [J. Number Theory 6, 128-135 (1974; Zbl 0285.10020)]. As an illustration of the author's results in this direction, let \(f\) be a real valued function on \([M,2M]\) with a continuous third derivative, and satisfying \(M^{3/2}\ll|x^j f^{(j)}(x)|\ll M^{3/2}\) for \(j=2\) and 3. Then for any \(\varepsilon> 0\) there are \(O(M^{13/18+ \varepsilon})\) integers \(m\in [M, 2M]\) for which \(\|f(m)\|< M^{-2/3}\). Previously one knew only that there were \(O(M^{3/4})\) values of \(m\) for which \(f(m)\) is integral. Divided differences play a key part in the proof of this latter result, as does the concept of `quadratic major arcs' developed by \textit{M. N. Huxley} [Number Theory in Progress. Vol. 2: Zakopane-Kośielisko 1997, de Gruyter, Berlin, 911-940 (1999; Zbl 0930.11073)]. Curiously, although the original problem concerns integral points near an analytic curve, issues arise concerning integral points on algebraic curves. In particular, a result on the number of zeros of ternary quadratic forms due to the reviewer [Ann. Math. (2) 155, 553-598 (2002; Zbl 1039.11044)] is used.