On the distribution of powered numbers
Extracted source text · arXiv 2306.05981
\documentclass{amsart}
\usepackage{mathrsfs}
\usepackage{amsfonts,amssymb,amsmath,amsthm}
\usepackage{color}
\usepackage{enumerate}
\usepackage{time}
%\usepackage{showlabels}
\newcommand{\rge}[1]{\color{red} #1\color{black} }
\newcommand{\blu}[1]{\color{blue} #1\color{black} }
\def\fk{\mathfrak}
\def\NN{{\mathbb N}}
\def\ZZ{{\mathbb Z}}
\def\RR{{\mathbb R}}
\def\QQ{{\mathbb Q}}
\def\CC{{\mathbb C}}
\def\FF{{\mathbb F}}
\def\GG{{\mathbb G}}
\def\KK{{\mathbb K}}
\def\eps{\varepsilon}
\def\al{\alpha}
\def\la{\lambda}
\def\lm{\lambda}
\def\Lm{\Lambda}
\def\bs{\backslash}
\def\del{\Tht}
\def\inf{\infty}
%\def\rho{\varrho}
\def\phi{\varphi}
\def\gm{\gamma}
\def\sgm{\sigma}
\def\Gm{\Gamma}
\def\Ups{\Upsilon}
\def\kp{\kappa}
\def\ra{\rightarrow}
\def\Ra{\Rightarrow}
\def\la{\leftarrow}
\def\La{\Leftarrow}
\def\LR{\Leftrightarrow}
\def\bal{\mbox{\boldmath$\alpha$}}
\newcommand\multindic[2]{\genfrac{}{}{0pt}{}{#1}{#2}} % replaces atop
\renewcommand{\Re}{\mbox{Re}\,}
\renewcommand{\Im}{\mbox{Im}\,}
\newcommand{\f}[2]{\frac{#1}{#2}}
\newcommand{\be}[1]{\begin{equation}\label{#1}}
\newcommand{\ee}{\end{equation}}
\newcommand{\multsum}[2]{\sum_{{\scriptstyle #1}\atop {\scriptstyle #2}}}
\newcommand{\trisum}[3]{\sum_{{{\scriptstyle #1}\atop {\scriptstyle #2}}\atop{\scriptstyle #3}}}
\newcommand{\multprod}[2]{\prod\limits_{{\scriptstyle #1}\atop {\scriptstyle
#2}}}
\newcommand{\multmax}[2]{\max_{{\scriptstyle #1}\atop {\scriptstyle #2}}}
\newcommand{\rf}[1]{{\rm (\ref{#1})}}
\newcommand{\Pe}[1]{P_{#1,{\sf e}}}
\newcommand{\Po}[1]{P_{#1,{\sf o}}}
\newcommand{\Se}[1]{S_{#1,{\sf e}}}
\newcommand{\So}[1]{S_{#1,{\sf o}}}
\newcommand{\hatPe}[1]{\widehat{P}_{#1,{\sf e}}}
\newcommand{\hatPo}[1]{\widehat{P}_{#1,{\sf o}}}
%\def\cal{\mathcal}
\def\cal{\mathscr}
\def\tht{\vartheta}\def\Tht{\Theta}
\def\th{\theta}
\def\dd{{\mathrm d}}
\def\r{\rho}
\def\LL{{\cal L}}
\def\vv{^{(v)}}
\def\rme{{\mathrm{e}}}
\def\PP{{\mathrm{P}}}
\def\Ups{\Upsilon}
\def\re{{\mathrm{Re}}}
\def\ts{\textstyle}
\def\ov{\overline}
\def\fk{\mathfrak}
\def\Ups{\Upsilon}
\def\NN{{\mathbb N}}
\def\QQ{{\mathbb Q}}\def\CC{{\mathbb C}}\def\ZZ{{\mathbb Z}}
\def\bfX{{\bf X}}
\def\bfx{{\bf x}}\def\bfy{{\bf y}}
\def\bfi{{\bf i}}
\def\bfj{{\bf j}}\def\bfc{{\bf c}}
\def\bfa{{\bf a}}
\def\bfb{{\bf b}}
\def\ups {{\upsilon}}
\def\ep{{\varepsilon}}
\newtheorem{thm}{Theorem}
\newtheorem*{Thm}{Theorem}
\newtheorem*{prop}{Proposition}
\newtheorem{lemma}{Lemma}
\makeatletter
\renewcommand\theequation{\thesection.\arabic{equation}}
\@addtoreset{equation}{section}
\makeatother
\makeatletter
\def\blfootnote{\xdef\@thefnmark{}\@footnotetext}
\makeatother
%Include your preferred graphics and mathematics packages here,
%using the command \usepackage{}
%The \newtheorem command is used to define theorem-like environments
%that normally REQUIRE A PROOF, for example:
%\newtheorem{theorem}{Theorem}[section] % 1st argument is your name for it
%\newtheorem{lemma}[theorem]{Lemma} % 2nd argument is what is printed
%\newtheorem{corollary}[theorem]{Corollary}
%\newtheorem{proposition}[theorem]{Proposition}
%To control the numbering sequence of these environments, see
%Lamport's book on LaTeX [2, p. 193].
% This is usually unnumbered
% The numbering sequence of these environments can be controlled in the
% same way as for \newtheorem; see Lamport's book on LaTeX, p. 193.
% The default LMS numbering of equations in long papers is (1.1), (1.2), (2.1), etc.
% In short papers, to change the numbering to (1), (2), etc., 'uncomment' the next line.
% \simpleequations
% Otherwise, use the AMS \numberwithin command.
% TOP MATTER
\title[Powered numbers]% end with percent
{On the distribution of powered numbers} % This is the full title of the paper
% Use lowercase letters in title except for proper names
% Avoid equations in title if possible
% Do not use the \thanks{} command; use \extraline{} instead (see below).
\author{J. Brüdern and O. Robert}
%\dedication{A dedication can be included here}
%Insert `2000 Mathematics Subject Classification' numbers here:
%\classno{11D72 (primary), 11L15, 11P05 (secondary)}
% Please refer to {\tt http://www.ams.org/msc/} for a list of codes}
%\extraline{Acknowledgements of grants and financial support should
%be included here; more general \textsl{Acknowledgements} are better
%placed either immediately before the bibliography (see
%page~\pageref{ackref}) or at the end of the introduction. Since
%author names should not carry footnote marks, instead refer to `The
%first author', etc. No `keywords' should be supplied. This guide was
%last revised on \filedate\ and documents {\tt lms.cls} version
%\fileversion.}
\begin{document}
%(\today, \now)
\begin{abstract} Asymptotic formulae are established for the number of natural numbers $m$ with
largest square-free divisor not exceeding $m^\tht$, for any fixed positive parameter $\tht$. Related counting functions are also considered.
\end{abstract}
\maketitle
\blfootnote{Keywords: powered numbers, nuclear numbers}
\blfootnote{MSC(2020): 11N25, 11N37}
\section{Introduction}
Motivated by questions about diophantine equations and the $abc$-conjecture, Mazur \cite{M} proposed to smooth out the set of positive $l$-th powers in a multiplicative way, by what he named {\em powered} numbers. To introduce the latter,
let $k(m)$ denote the largest square-free divisor of the natural number $m$, and let $ i(m) = {\log m}/{\log k(m)}$.
For all $l\in\mathbb N$ one has $i(m^l)\ge l$. Mazur's powered numbers (relative to $l$) are the numbers $m\in\mathbb N$ with $i(m)\ge l$. Note here that $l$ need not be integral in this definition, but for $l\in\mathbb N$ the powered numbers (relative to $l$) contain the $l$-th powers. It is proposed in \cite{M} to replace, within a given diophantine equation, an $l$-th power by the corresponding powered numbers, and to consider the resulting equation between powered numbers as the associated ``rounded'' diophantine equation.
In this note we analyse the distribution of powered numbers.
We find it is more appropriate to work with the real number $\tht=1/l$. The condition $i(m)\ge l$ is expressed equivalently as $k(m)\le m^\tht$, and for any $\tht>0$ we define the set
\[
\cal{A}(\tht)=\{m\in \NN\colon k(m)\le m^{\tht}\}.
\]
Thus, Mazur's powered numbers (relative to $l$) are exactly the elements of
$\cal{A}(1/l)$. For analytic approaches to rounded diophantine problems it is indispensable to determine the density of the set $\cal{A}(\tht)$. Our principal goal are asymptotic formulae for the number $S_\tht(x)$ of elements in $\cal{A}(\tht)$ that do not exceed $x$, and for related counting functions.
It is not difficult to see that for any $0<\tht<1$ the number $S_\tht(x)$ obeys the inequalities
\be{simple} x^\tht \ll S_\tht(x) \ll x^{\tht+\varepsilon} \ee
whenever $\varepsilon$ is a given positive real number and $x$ is large in terms of $\varepsilon$. It now transpires that for $\tht=1/l$ the powered numbers are not much denser than the $l$-th powers. A weaker version of \eqref{simple} occurs in Mazur \cite{M} who
refers to Granville, showing him an ``easy'' argument supposedly confirming the inequalities $x^{\tht-\varepsilon} \ll S_\tht(x) \ll x^{\tht+\varepsilon}$. In Mazur's article there is no indication how this would go but the simplest argument we know allows one to take $\varepsilon=0$ in the lower bound. To substantiate this claim, fix a number $\tht\in (0,1)$. Define $l\in \mathbb N$ by $1/l<\tht\le 1/(l-1)$, and $t\in\mathbb R$
by $l-t=1/\tht$. Then $t\in(0,1]$ and
\be{1/theta}
{1}/{\tht}=t(l-1)+(1-t)l.
\ee
Let $W(x)$ be the number of natural numbers $w$ that have a representation
\be{rep-w}
w=8^{l}n^{l-1}m^{l}
\ee
with $n,m$ square-free and constrained to the intervals
\be{range-n-m}
\frac{1}{2}\,\Big(\frac{x}{8^l}\Big)^{\tht t}<n\le \Big(\frac{x}{8^l}\Big)^{\tht t},\quad \frac{1}{2}\,\Big(\frac{x}{8^l}\Big)^{\tht (1-t)}<m\le \Big(\frac{x}{8^l}\Big)^{\tht (1-t)}.
\ee
The numbers $w$ counted by $W(x)$ uniquely determine the
square-free numbers $n,m$ with \eqref{rep-w} and \eqref{range-n-m}. This follows from unique factorisation. Hence $W(x)$ equals the number of pairs $n,m$ of square-free numbers satisfying \eqref{range-n-m}. In particular, we see that $W(x)\gg x^{\tht}$.
Next, we observe that \eqref{rep-w} gives $k(w)\le 2nm$, and therefore $k(w)\le 2(8^{-l}x)^{\tht}$. By \eqref{1/theta} and \eqref{range-n-m}, we also have $w\ge 2^{-(2l-1)}x$. It follows that $k(w)\le w^{\tht}$. Hence, numbers counted by $W(x)$ are also counted by $S_{\tht}(x)$. This yields $S_{\tht}(x)\ge W(x)$, and the left inequality in \eqref{simple} follows at once.
For the upper bound in \eqref{simple} one may refer to authorities like Tenenbaum \cite[Theorem II.1.15]{T}. However, the shortest argument available to us uses Rankin's trick within the following chain of obvious inequalities:
\[ S_\tht(x) \le \multsum{m\le x}{k(m)\le x^\tht} 1\le \sum_{m=1}^\infty
\Big(\frac{x^\tht}{k(m)}\Big) \Big(\frac{x}{m}\Big)^\varepsilon \ll x^{\tht
+\varepsilon}. \]
A finer analysis of the counting function $S_\tht(x)$ is currently an undesired lacuna in the literature. As a partial remedy we establish an asymptotic formula that should be sufficient for many applications. Let $1\le y\le x$. Following Robert and Tenenbaum \cite{RT2013} we write $N(x,y)$ for the number of $m\in\mathbb N$ with $m\le x$ and $k(m)\le y$. The numbers counted here are their {\em nuclear} numbers. It seems natural to expect that the trivial upper bound
$S_{\tht}(x)\le N(x,x^{\tht})$ should not be very wasteful. However, it was conjectured by Erd\"os (1962) and proved by De Bruijn and van Lint (1963) that
\be{erdos62}
\sum_{m\le x}\frac{m}{k(m)}=o\Big( \sum_{m\le x}\frac{x}{k(m)} \Big)\qquad (x\to+\infty).
\ee
For a quantitative version of this estimate, see \cite[Th\'eor\`eme 4.4]{RT2013}. These results suggest that the order of magnitude of $S_{\tht}(x)$ is somewhat smaller than that of $N(x,x^{\tht})$, and this is indeed the case.
Before we make this precise,
we recall an estimate for $N(x,y)$. We are primarily interested in the case $y=x^{\tht}$ with $\tht>0$, but we work in the wider range $y>\exp \big((\log x)^{2/3}\big)$. The multiplicative function
\be{psi}
\psi(m)=\prod_{p \mid m}(p+1)\qquad (m\ge 1)
\ee
and the function $F:[0,\infty)\to [0,\infty)$ defined by
\[
F(v)=\frac{6}{\pi^2}\sum_{m\ge 1}\frac{1}{\psi(m)}\min\Big(1,\frac{e^v}{m}\Big)
\]
are featured in the uniform asymptotic formula
\be{nuc2}
N(x,y)=\big(1+o(1)\big)yF\big(\log(x/y)\big)\qquad \big(y> \exp \big((\log x)^{2/3}\big),\thinspace x\to+\infty\big)
\ee
that is contained in \cite[Proposition 10.1]{RT2013}. As we shall see momentarily,
for each pair $\tht, x$ with $0<\tht<1$ and $x\ge 2$, there is exactly one real number $\alpha=
\alpha_{\tht}(x)>0$ with
\be{seltsam}
\sum_{p}\frac{p^{\alpha}\log p}{(p^{\alpha}-1)\big(1+ (p+1)(p^{\alpha}-1) \big)}
=(1-\tht)\log x. \ee
We are now in a position to state our first result.
\begin{thm}\label{S-theta} Let $0<\tht<1$ fixed. Then, for $x\ge 27$, one has
\be{S7}
S_{\tht}(x)= x^{\tht} F\big( (1-\tht)\log x\big)\frac{ \alpha_{\tht}(x) }{\tht} \Bigg(1+ O\Big( \sqrt{\frac{\log\log x}{\log x}} \Big)\Bigg) .
\ee
As $x\to \infty$, one also has
\be{S8}
S_{\tht}(x)=(1+o(1)) x^{\tht} F\big( (1-\tht)\log x\big)\frac{1}{\tht} \Big(\frac{2}{1-\tht}\Big)^{1/2} \big((\log x) \log\log x\big)^{-1/2}.
\ee
\end{thm}
This result calls for several comments. First we take $y=x^\tht$ in \eqref{nuc2} and substitute the resulting equation
\be{Ntheta}
N(x,x^{\tht})=(1+o(1)) x^{\tht} F\big( (1-\tht)\log x\big)\qquad (x\to+\infty)
\ee
within \eqref{S8} to infer that
\[ S_{\tht}(x)=(1+o(1)) N(x,x^{\tht}) \frac{1}{\tht} \Big(\frac{2}{1-\tht}\Big)^{1/2} \big((\log x) \log\log x\big)^{-1/2} \quad (x\to +\infty).
\]
This is an analogue of \eqref{erdos62}, in quantitative form that is of strength comparable to \cite[Th\'eor\`eme 4.4]{RT2013}.
Our second comment concerns the implicitly defined function $\alpha_\tht(x)$.
It originates in the Dirichlet series
\be{defG}
\cal{G}(s)= \frac{6}{\pi^2} \sum_{m\ge 1}\frac{1}{\psi(m)m^s}=\frac{6}{\pi^2}\prod_{p}\left(1+\frac{1}{(p+1)(p^s-1)}\right)
\ee
that converges absolutely in $\text{Re}\, s > 0$, and therefore has no zeros in this half plane. For real numbers $\sigma>0$ we have $\cal G(\sigma)>0$. We may then define
\[
g(\sigma)=\log \cal{G}(\sigma).
\]
Note that $g$ extends to a holomorphic function on the right half plane, and one computes the logarithmic derivative of $\cal G(s)$ from the
Euler product representation \eqref{defG} to
\[ g'(s) = - \sum_{p}\frac{p^{s}\log p}{(p^{s}-1)\big(1+ (p+1)(p^{s}-1) \big)}. \]
Note that the sum on the right hand side here coincides with the sum in \eqref{seltsam}. On differentiating again, it transpires that the real function $g':(0,\infty)\to \mathbb R$ is increasing. Considering $\sigma\to 0$ and $\sigma\to\infty$ one finds that its range is the open interval $(-\infty,0)$. We conclude that for a given $v>0$ there is a unique positive number $\sigma_v$ with
\[
g'(\sigma_v)+v=0.
\]
In \cite[Lemme 6.6]{RT2013} it is shown that
\be{estim-sigma-v}
\sigma_v=(1+o(1)) \sqrt{\frac{2}{v\log v}}\qquad (v\to+\infty).
\ee
Here we choose $v=(1-\tht)\log x$ and then have
$\alpha_\tht(x)=\sigma_{v}$. In particular, we see that \eqref{S7} and \eqref{estim-sigma-v} imply \eqref{S8}. Thus, it only remains to prove \eqref{S7}.
Our last comment concerns the actual size of $S_\tht(x)$. This requires some more information on the function $F$.
From \cite[(2.12)]{RT2013} we have the asymptotic relation
\be{logF}
\log F(v)=\big(1+o(1)\big)\sqrt{\frac{8v}{\log v}}\qquad (v\to+\infty) .
\ee
By inserting \eqref{logF} into \eqref{S8}, we deduce that
there exists some positive number $\beta(x;\tht)>0$ with
\be{helpme}
S_{\tht}(x)=x^{\tht + \beta(x;\tht) }\qquad (0<\tht<1,\thinspace x\ge x_0(\tht))
\ee
and the property that for any fixed $\tht\in(0,1)$ one has
\[
\beta(x;\tht)=(1+o(1))\sqrt{\frac{8(1-\tht)}{(\log x)(\log\log x)}}\qquad (x\to+\infty).
\]
Note that \eqref{helpme} yields another proof of \eqref{simple}.
\bigskip
We now turn to local estimates for $S_{\tht}(x)$. In our case, this amounts to comparing the respective behaviour of $S_{\tht}(zx)$ and $S_{\tht}(x)$ uniformly for large $x$, when $z$ is in some sense sufficiently close to $1$. Such estimates are often obtained with the saddle-point method, and we follow this route here, too. In a suitable range for $z$, the fraction $S_{\tht}(zx)/S_{\tht}(x)$ may be approximated by a simple function of $z$.
\begin{thm}\label{local-estim}
Let $0<\tht<1$. Then for $x$ large, we have
\[
S_{\tht}(zx)=z^{\tht}S_{\tht}(x)\Bigg(1+ O\Big( \sqrt{\frac{\log\log x}{\log x}} \Big)\Bigg)
\]
uniformly for $z>0$ with $|\log z|\ll \log\log x$.
\end{thm}
Finally, we consider the counting function for a variation of the powered numbers. For given $\tht\in (0,1)$ and $\Tht\in \RR$, we consider
$$
S_{\tht,\Tht}(x)=\#\{n\le x\colon k(n)\le n^{\tht}(\log n)^{\Tht}\}.
$$
Note that $S_{\tht}(x)=S_{\tht,0}(x)$. The set of integers such that $k(n)\le n^{\tht}(\log n)^{\Tht}$ plays a prominent role in a forthcoming paper, and therefore, we provide an estimate for $S_{\tht,\Tht}(x)$. It turns out that the conditions $k(n)\le n^{\tht}$ and $k(n)\le n^{\tht}(\log n)^{\Tht}$ are relatively close, and that the ratio $S_{\tht,\Tht}(x)/S_{\tht}(x)$ is roughly of size $(\log x)^{\Tht}$.
\begin{thm}\label{k-log}
Let $0<\tht<1$ and $\Tht\in \RR$ be fixed. Then for $x$ large, one has
\[
S_{\tht,\Tht}(x)=(\log x)^{\Tht}S_{\tht}(x)\Bigg(1+ O\Big( \sqrt{\frac{\log\log x}{\log x}} \Big)\Bigg).
\]
\end{thm}
\section{Proof of Theorem 1}
In this section we derive Theorem 1. Before we embark on the main argument, we fix some notation and recall a pivotal result concerned with the distribution of square-free numbers. This involves the function $\psi(m)$ as defined in \eqref{psi}, the M\"obius function $\mu(m)$, and for a parameter $0\le \gamma\le \frac{1}{2}$ at our disposal, the product
\[
r_\gamma(m)=\prod_{p\mid m}\big(1+4\gamma p^{-1/2}\big).\]
One then has the estimate (\cite[(10.1)]{RT2013})
\be{sum-mu}
\sum_{l\le z} \mu^2(lk)=\frac{6kz}{\pi^2\psi(k)} +O\big( r_\gamma(k)z^{1-\gamma} \big) \ee
that holds uniformly relative to the square-free number $k$ and the real parameters $z, \gamma$ in the ranges
$z\ge 1$, $0\le \gamma \le \frac12$.
The first steps of our argument follow the pattern laid out in \cite[Sect.\ 10]{RT2013}. Unique factorisation shows that for all natural numbers $n$ there exists exactly one pair of coprime natural numbers $l, m$ with $\mu(l)^2=1$ and $n=lmk(m)$. Note that the two conditions $(l,m)=1$ and $\mu(l)^2=1$ are equivalent to the single condition $\mu(lk(m))^2=1$. Further, one has $k(n)=lk(m)$. With $\tht\in(0,1)$ now fixed, it follows that $S_\tht(x)$ equals the number of $(l,m)\in\mathbb N^2$ satisfying the conditions
\[ \mu(lk(m))^2=1, \quad lmk(m)\le x, \quad lk(m) \le (lmk(m))^\tht. \]
These last three conditions we recast more compactly as
\be{32}
\mu(lk(m))^2=1, \quad lk(m) \le \min (x/m, m^{\tht/(1-\tht)}).
\ee
From now on, the number $\kappa=\tht/(1-\tht)$ features prominently, and we also put $y=x^\tht$. Note that
\be{32a} \min (x/m, m^\kappa) = m^\kappa \;\; \text{ if and only if }\,\, m\le x/y. \ee
Hence, we consider the ranges $m\le x/y$ and $x/y<m\le x$ separately. By \eqref{32} and \eqref{32a}, this leads to the decomposition
\be{33} S_{\tht}(x)=S_{\tht}^{(1)}(x)+S_{\tht}^{(2)}(x) \ee
in which
\[
S_{\tht}^{(1)}(x) =\!\multsum{m\le x/y}{k(m)\le m^{\kappa}} \sum_{l\le m^{\kappa}/{k(m)}}\!\mu(lk(m))^2,\quad
S_{\tht}^{(2)}(x)= \! \multsum{x/y<m\le x}{mk(m)\le x} \sum_{l\le {x}/mk(m)}\!\mu(lk(m))^2.
\]
We apply \eqref{sum-mu} with $k=k(m)$ to both inner sums and obtain
\begin{align}
S_{\tht}^{(1)}(x) & = \frac{6}{\pi^2}\multsum{m\le x/y}{k(m)\le m^{\kappa}}
\frac{ m^{\kappa} }{\psi(m)}+O(R_1), \label{S1} \\
S_{\tht}^{(2)}(x) & = \frac{6}{\pi^2}\multsum{x/y<m\le x}{mk(m)\le x} \frac{ x }{m\psi(m)}+O(R_2) .\label{S2}
\end{align}
where
\[ R_1 = \multsum{m\le x/y}{k(m)\le m^{\kappa}}r_\gamma(m)\Big( \frac{ m^{\kappa }}{k(m)} \Big)^{1-\gamma}, \qquad
R_2 = \multsum{x/y<m\le x}{mk(m)\le x} r_\gamma(m)\Big( \frac{ x }{mk(m)} \Big)^{1-\gamma}.\]
It turns out that $R_1$ and $R_2$ are small. In order to couch their estimation, as well as the analysis of other error terms that arise later, under the umbrella of a single treatment, we choose parameters $\gamma$ and $\sigma$ with $0<\gamma<\sigma\le \frac12$ and introduce the series
\be{E} E = \sum_{m=1}^\infty r_\gamma(m)\Big(\frac{y}{k(m)}\Big)^{1-\gamma} \Big(\frac{x/y}{m}\Big)^\sigma. \ee
The conditions $\sigma>\gamma>0$ ensure convergence. It is routine to show that
\[ R_1+R_2 \ll E. \]
In fact, by Rankin's trick,
$$ R_1 \le \multsum{m\le x/y}{k(m)\le m^{\kappa}}r_\gamma(m)\Big( \frac{ m^{\kappa }}{k(m)}\Big)^{1-\gamma} \Big(\frac{x/y}{m}\Big)^\sigma. $$
For $m\le x/y$ one has $m^\kappa \le y$, and it follows that $R_1\le E$.
Likewise, one confirms $R_2\le E$ by observing that $1-\gamma>
\frac12 \ge \sigma$ so that
\be{help} \Big(\frac{x}{mk(m)}\Big)^{1-\gamma} = \Big(\frac{y}{k(m)}\Big)^{1-\gamma}
\Big(\frac{x/y}{m}\Big)^{1-\gamma} \le \Big(\frac{y}{k(m)}\Big)^{1-\gamma}
\Big(\frac{x/y}{m}\Big)^\sigma. \ee
The appearance of $k(m)$ in the summation conditions on the right hand sides of \eqref{S1} and \eqref{S2} is a nuisance, and we proceed by removing these. If the condition $k(m)\le m^\kappa$ is removed from the sum in \eqref{S1} one imports an error no larger than
\[ \multsum{m\le x/y}{k(m)> m^{\kappa}} \frac{ m^{\kappa} }{\psi(m)}
\le \multsum{m\le x/y}{k(m)> m^{\kappa}}
\frac{ m^{\kappa} }{k(m)}
\le \sum_{m\le x/y} \Big(\frac{m^{\kappa}}{k(m)}\Big)^{1-\gamma} \Big(\frac{x/y}{m}\Big)^\sigma\le E.
\]
Here, the last inequality is obtained by the argument that completed the estimation of $R_1$.
Similarly, if the condition $mk(m)\le x$ is removed from the summation condition in \eqref{S2}, then the resulting error does not exceed
\[ \multsum{x/y<m\le x}{mk(m)> x} \frac{ x }{m\psi(m)}
\le \multsum{x/y<m\le x}{mk(m)> x} \frac{ x }{mk(m)}
\le \multsum{m>x/y}{mk(m)> x} \frac{ x }{mk(m)} = R, \]
say. By Rankin's trick and \eqref{help},
\[ R\le \sum_{m>x/y} \Big(\frac{ x }{mk(m)}\Big)^{1-\gamma}
\le \sum_{m>x/y} \Big(\frac{ y }{k(m)}\Big)^{1-\gamma}\Big(\frac{x/y}{m}\Big)^\sigma \le E. \]
Collecting together, we deduce from \eqref{S1} and \eqref{S2} the asymptotic relations
\[
S_{\tht}^{(1)}(x) = \frac{6}{\pi^2}\sum_{m\le x/y}
\frac{ m^{\kappa} }{\psi(m)}+O(E), \qquad
S_{\tht}^{(2)}(x) = \frac{6}{\pi^2}\sum_{x/y<m\le x} \frac{ x }{m\psi(m)}+O(E) ,
\]
and by \eqref{33}, we infer that
\[ S_\tht(x) = \frac{6}{\pi^2}\sum_{m\le x} \frac{ m^{\kappa} }{\psi(m)}\min(1, xm^{-\kappa-1} ) + O(E). \]
Note that the sum on the right is a partial sum of a convergent series. If one completes the sum, then it is immediate that the error thus imported is bounded by
$R$, and hence by $E$. We have now reached the provisional expansion
\be{S3}S_\tht(x) = \frac{6}{\pi^2}\sum_{m=1}^\infty \frac{ m^{\kappa} }{\psi(m)}\min(1, xm^{-\kappa-1} ) + O(E).
\ee
It remains to estimate $E$. In its definition \eqref{E}, we encounter a sum over a multiplicative function, and so
\begin{align*} E =& y^{1-\gamma} (x/y)^\sigma \prod_{p}\Big(1+\frac{p^{\gamma}r(p;\gamma)}{p(p^{\sigma}-1)}\Big)\\ \le& y^{1-\gamma} (x/y)^\sigma\prod_{p}\Big(1+\frac{p^{\gamma}}{p(p^{\sigma}-1)}\Big)\prod_{p}\Big(1+\frac{4\gamma p^{\gamma-\frac{1}{2}}}{p(p^{\sigma}-1)}\Big).
\end{align*}
Now, since $p^{\gamma-\frac{1}{2}}\le 1$, $\gamma< \sigma$, $p^{\sigma}-1\ge \sigma\log p$, and since the sum $\sum_{p}\frac{1}{p\log p}$ converges,
\[
\prod_{p}\Big(1+\frac{4\gamma p^{\gamma-\frac{1}{2}}}{p(p^{\sigma}-1)}\Big)\le \prod_{p}\Big(1+\frac{4}{p\log p}\Big)\ll 1.
\]
As on an earlier occasion, we write $v= (1-\tht)\log x = \log \frac{x}{y}$, and then choose $\sigma=\sigma_v$ and $\gamma=\sigma_v-\frac{1}{\log y}$. By \eqref{defG} one has
\[
\prod_{p}\Big(1+\frac{p^{\gamma}}{p(p^{\sigma}-1)}\Big)=\frac{\pi^2}{6}\cal{G}(\sigma)\prod_p\Big(1+\frac{(p^{\gamma}-1)(p+1)+1}{p+p(p+1)(p^{\sigma}-1)} \Big).
\]
Now, on recalling \eqref{estim-sigma-v},
\[
\prod_{p\le e^{1/\sigma}}\Big(1+\frac{(p^{\gamma}-1)(p+1)+1}{p+p(p+1)(p^{\sigma}-1)} \Big)\le \prod_{p\le e^{1/\sigma}}\Big(1+\frac{1}{p}\Big)\ll \frac{1}{\sigma}\ll (v\log v)^{1/2}
\]
while one also has
\[
\prod_{p> e^{1/\sigma}}\Big(1+\frac{(p^{\gamma}-1)(p+1)+1}{p+p(p+1)(p^{\sigma}-1)} \Big)\le \prod_{p> e^{1/\sigma}}\left(1+\frac{1}{p^{1+\sigma-\gamma}}\right)\ll \zeta\Big(1+\frac{1}{\log y}\Big)\ll \log y.
\]
On collecting together, this shows that
\[
E\ll y^{1-\sigma_v}e^{ v\sigma_v}\cal{G}(\sigma_v)(v\log v)^{1/2}\log y.
\]
From \cite[(2.11)]{RT2013} we deduce that
\[
F(v)\asymp \Big(\frac{\log v}{v}\Big)^{1/4} e^{v\sigma_v} \cal{G}(\sigma_v) \qquad (v\ge 2),
\]
and hence
\[
E \ll y^{1-\sigma_v}F(v) v^{3/4} (\log v)^{1/4} \log y.
\]
With the choice of $y$ and $v$, one has $\sigma_v=\alpha_{\tht}(x)$. Moreover, $\log y$ and $v$ have the order of magnitude $\log x$ so that the last inequality now reads
\be{Enew}
E\ll x^{\tht} F((1-\tht)\log x)x^{-\tht \alpha_{\tht}(x)}(\log x)^{7/4}(\log\log x)^{1/4}.
\ee
\medskip
Our final task is to compare our estimate for $E$ with the size of the sum on the right of \eqref{S3}.
Recall that in view of \eqref{simple} and \eqref{Ntheta}, $S_{\tht}(x)$ and $N(x,x^{\tht})$ are of comparable size. In
order to mimick the estimate \eqref{Ntheta},
we introduce the function
\[
H_{\tht}(x)=\frac{6}{\pi^2x^{\tht}}\sum_{m=1}^\infty \frac{ m^{\kappa} }{\psi(m)}\min(1, xm^{-\kappa-1} )
\]
so that \eqref{S3} now reads
\[
S_\tht(x) =x^{\tht}H_{\tht}(x)+O(E).
\]
Our aim is to give an estimate of $H_\tht(x)$ by using the saddle-point method, and to describe more precisely $ H_\tht(x) /F((1-\tht)\log x)$ as $x\to+\infty$.
\begin{lemma}\label{lemmaH}
Fix $\tht$ with $0<\tht<1$. Then for $x\ge 27$ one has
\[
H_\tht(x)=F((1-\tht)\log x)\frac{\alpha_{\tht}(x)}{\tht}\Bigg(1+O\Big(\sqrt{\frac{\log\log x}{\log x}}\Big)\Bigg).
\]
\end{lemma} \noindent
\textit{Proof.} The argument is modelled on \cite[Section 8]{RT2013}. Recall the identity
\[
\frac{1}{2\pi \mathrm{i}}\int_{\sigma+\mathrm{i}\RR}\frac{y^{s}}{s(1-s)}\,\dd s=\min(1,y),\qquad (0<\sigma<1,\thinspace y>0).
\]
We then have the integral representation
\begin{align*}
H_\tht(x)&=\frac{1}{x^{\tht}}\frac{6}{\pi^2}\sum_{m\ge 1} \frac{ m^{\kappa} }{\psi(m)}\frac{1}{2\pi \mathrm{i}}\int_{\sigma+\mathrm{i}\RR}\frac{1}{s(1-s)} \left(\frac{x}{m^{1/(1-\tht)}}\right)^s\dd s
\\
&= \frac{1}{2\pi \mathrm{i}}\int_{\sigma+\mathrm{i}\RR} \cal{G}\Big(\frac{s-\tht}{1-\tht}\Big)x^{s-\tht} \frac{\dd s}{s(1-s)}
\\
&= \frac{1}{2\pi }\int_{\RR} \cal{G}\Big(\frac{\sigma+\mathrm{i}t -\tht}{1-\tht}\Big)x^{\sigma+\mathrm{i}t-\tht} \frac{\dd t}{(\sigma+\mathrm{i}t)(1-\sigma-\mathrm{i}t)} .
\end{align*}
After a linear change of variable in $t$, we arrive at
\[H_\tht(x)=
\frac{1}{2\pi }\int_{\RR} \cal{G}\Big(\frac{\sigma-\tht}{1-\tht}+\mathrm{i}t \Big)x^{\sigma+\mathrm{i}(1-\tht)t-\tht} \frac{(1-\tht)\,\dd t}{(\sigma+\mathrm{i}(1-\tht)t)(1-\sigma-\mathrm{i}(1-\tht)t)} .
\]
Recall again that $v=(1-\tht)\log x$, and that $\alpha_{\tht}(x)=\sigma_v$. We take $\sigma=\tht+(1-\tht)\sigma_v$. For large $x$ one then has $0<\sigma<1$, and the previous formula for $H_\tht(x)$ becomes
\[
H_\tht(x)= \frac{1}{2\pi }\int_{\RR} \frac{ \cal{G}(\sigma_v+\mathrm{i}t) e^{(\sigma_v+\mathrm{i}t)v} }
{\big(\tht+(1-\tht)(\sigma_v+\mathrm{i}t)\big)\big( 1-\sigma_v-\mathrm{i}t \big) } \, {\dd t}.
\]
After truncation, we have
\[
H_\tht(x)=\frac{1}{2\pi } \int_{-v^2}^{v^2}
\frac{ \cal{G}(\sigma_v+\mathrm{i}t) e^{(\sigma_v+\mathrm{i}t)v} }
{\big(\tht+(1-\tht)(\sigma_v+\mathrm{i}t)\big)\big( 1-\sigma_v-\mathrm{i}t \big) } \, \dd t
+
O\left( \frac{\cal{G}(\sigma_v) e^{v\sigma_v} }{v^2} \right).
\]
Moreover, following the proof of \cite[Th\'eor\`eme 8.6]{RT2013}, we set
\[
\eta_v=(\log v)/\sqrt{g''(\sigma_v)}\asymp (\log v)^{3/4}/v^{3/4}
\]
and recall \cite[Lemme 8.5]{RT2013}, asserting that for some $c>0$ we have
\[
|\cal{G}(\sigma_v+\mathrm{i}t)|\ll \cal{G}(\sigma_v)e^{-c (\log v)^2}\qquad (\eta_v\le |t|\le \exp((\log v)^{38/37}).
\]
It now follows that
\[
H_\tht(x)=\frac{1}{2\pi }
\int_{-\eta_v}^{\eta_v}
\frac{ \cal{G}(\sigma_v+\mathrm{i}t) e^{(\sigma_v+\mathrm{i}t)v} }
{\big(\tht+(1-\tht)(\sigma_v+\mathrm{i}t)\big)\big( 1-\sigma_v-\mathrm{i}t \big) } \, {\dd t}
+
O\left( \frac{\cal{G}(\sigma_v) e^{v\sigma_v} }{v^2} \right).
\]
Setting
\[
D_{m}=v^{(m+1)/2}(\log v)^{(m-1)/2}\qquad (m\ge 0),
\]
we have
\[
H_\tht(x)=\frac{ \cal{G}(\sigma_v) e^{v\sigma_v} }{2\pi } \int_{-\eta_v}^{\eta_v} \Upsilon (t)e^{-g''(\sigma_v)t^2/2} \dd t+ O\left( \frac{\cal{G}(\sigma_v) e^{v\sigma_v} }{v^2} \right)
\]
where
\[
\Upsilon(t)= \frac{e^{-\mathrm{i}g'''(\sigma_v)t^3/6+O(t^4D_4)}}{ \big(\tht+(1-\tht)(\sigma_v+\mathrm{i}t)\big)\big( 1-\sigma_v-\mathrm{i}t \big) }.
\]
By Taylor expansion, we infer
\[ \Upsilon(t) = \frac{\mathrm Z(t)}{(\tht+(1-\tht)\sigma_v)(1-\sigma_v) }\]
where
\[ \mathrm Z(t)=
1+\mathrm{i}t \Big(\frac1{1-\sigma_v}-\frac{1-\tht}{\tht+(1-\tht)\sigma_v} \Big) -{\mathrm{i}}g'''(\sigma_v)\frac{t^3}6 +O(t^2+D_4t^4+D_3^2t^6) .
\]
Now, still following the pattern of the proof of \cite[Th\'eor\`eme 8.6]{RT2013}, one is lead to
\[
H_\tht(x)=\frac{x^{(1-\tht) \alpha_{\tht}(x) } \cal{G}\big(\alpha_{\tht}(x)\big) }{\tht \sqrt{2\pi g''\big( \alpha_{\tht}(x) \big) }}\Bigg(1+O\Big(\sqrt{\frac{\log\log x}{\log x}}\Big)\Bigg).
\]
We omit the details. From \cite[Th\'eor\`eme 8.6]{RT2013} we import the relation
\[
F(v)=\frac{e^{v\sigma_v } \cal{G}(\sigma_v) }{\sigma_v\sqrt{2\pi g''(\sigma_v)}}\left(1+O\left( \sqrt{\frac{\log v}{v}} \right)\right)\qquad (v\ge 2),
\]
and the lemma follows.
\qed
\smallskip
We may now complete the proof of Theorem 1. Using the lemma and \eqref{estim-sigma-v}, we obtain
\[
H_{\tht}(x)\asymp F((1-\tht)\log x) \big((\log x)\log\log x\big)^{-1/2},
\]
so that the estimate \eqref{Enew} implies
\[
E\ll x^{\tht}H_{\tht}(x)x^{-\tht\alpha_{\tht}(x)}(\log x)^{9/4}(\log\log x)^{3/4}.
\]
We then have
\[
S_{\tht}(x)=x^{\tht}H_{\tht}(x)\big(1+O\big( x^{-\tht\alpha_{\tht}(x)}(\log x)^{9/4}(\log\log x)^{3/4} \big)\big).
\]
We may now replace $H_{\tht}(x)$ by the estimate from Lemma \ref{lemmaH}. Since
\[
x^{-\tht\alpha_{\tht}(x)}(\log x)^{9/4}(\log\log x)^{3/4}\ll \sqrt{\frac{\log\log x}{\log x}},
\]
this yields \eqref{S7}. As remarked earlier, \eqref{S8} follows from \eqref{estim-sigma-v} and \eqref{S7}. The proof of Theorem 1 is complete.
\section{Proof of Theorem \ref{local-estim} }
Subject to the hypotheses of Theorem \ref{local-estim}, when $x$ is large, one has $\log zx\asymp \log x$. Hence, Theorem 1 implies that
\be{Szx}
S_{\tht}(zx)=(zx)^{\tht} F\big( (1-\tht)\log zx\big)\frac{ \alpha_{\tht}(zx) }{\tht} \Bigg(1+ O\Big( \sqrt{\frac{\log\log x}{\log x}} \Big)\Bigg)
\ee
holds uniformly for $|\log z|\ll \log\log x$. We recall \cite[Proposition 8.7]{RT2013}. This asserts that uniformly for $|t|\le v^{3/4}(\log v)^{1/4}$, one has
$$
F(v+t)=e^{t\sigma_{v}}F(v)\Big(1+O\Big(\frac{\log v + t^2/v}{\sqrt{v\log v}}\Big) \Big).
$$
Using this estimate with $v=(1-\tht)\log x$ and $t=(1-\tht)\log z$, one finds
\[
F\big( (1-\tht)\log zx\big)=F\big( (1-\tht)\log x\big)z^{(1-\tht)\alpha_{\tht}(x)} \Bigg(1+ O\Big( \sqrt{\frac{\log\log x}{\log x}} \Big)\Bigg).
\]
Moreover, in the ranges for $x$ and $z$ considered here, one has
\[
z^{(1-\tht)\alpha_{\tht}(x)}=1+O\big( \alpha_{\tht}(x)\log z \big),
\]
which yields an admissible error term.
Finally, \cite[(8.9)]{RT2013} implies that uniformly for $|t|\le v/2$ one has
$$
\sigma_{v+t}=\sigma_v\big(1+ O(|t|/v)\big).
$$
Hence
\[
\alpha_{\tht}(zx) =\alpha_{\tht}(x) \Big(1+ O\Big( \frac{\log z}{\log x} \Big)\Big)
\]
which again leads to an admissible error term. Inserting these estimates in \eqref{Szx} completes the proof of Theorem \ref{local-estim}.
\medskip
It may be worth pointing out that the above argument actually proves a little more. A close inspection of the proof of Theorem \ref{local-estim} shows that the estimate
\[
S_{\tht}(zx)=z^{\tht+(1-\tht)\alpha_0(x)}S_{\tht}(x)\Bigg(1+ O\Big( \sqrt{\frac{\log\log x}{\log x}} +\frac{(\log z)^2}{(\log x)^{3/2}(\log\log x)^{1/2}} \Big)\Bigg)
\]
holds uniformly in the range $z>0$, $x>27$, $|\log z|\le (\log x)^{3/4}(\log\log x)^{1/4}$.
\section{Proof of Theorem \ref{k-log}}
Before proving Theorem \ref{k-log}, we briefly sketch the main steps. We first choose a suitable real number $U=U(x)$ such that $\log U=(\log x)(1+o(1))$, and count the integers $n$ not exceeding $x$ sucht that $k(n)\le n^{\tht}(\log U)^{\Tht}$. The first step is to show that the number of these integers is essentially $S_{\tht}(x)$ multiplied by $(\log x)^{\Tht}$. The second step is to prove that the number of these integers is close to $S_{\tht,\Tht}(x)$.
In light of this description, for any $x\ge 1$ and any $z>0$, we set
$$
B(x,z)=\#\{n\le x\colon k(n)\le z n^{\tht}\}.
$$
\begin{thm}\label{logx}Let $0<\tht<1$ be fixed. Then for $x$ large, one has
\[
B(x,z)=z S_{\tht}(x)\Bigg(1+ O\Big( \sqrt{\frac{\log\log x}{\log x}} \Big)\Bigg)
\]
uniformly for $z>0$ with $|\log z|\ll \log\log x$.
\end{thm}
\textit{Proof. }We follow very closely the proof of Theorem 1 which corresponds to the case $z=1$.
We redefine the meaning of $y$, now set to $y=x^{\tht}z$, and keep the notation $\kappa$ and $E$. Note that in the sequel of this proof, the error term $E$ is to be interpreted with the current specific choices for $x$ and $y$.
For any $n\ge 1$, recall that there is a unique pair $l,m$ with $n=lmk(m)$ and $\mu^2\big(lk(m)\big)=1$. The conditions $n\le x$ and $k(n)\le z^{\tht}$ become
\[
lmk(n)\le x\quad\mbox{ and }\quad lk(m)\le m^{\kappa}z^{1+\kappa}.
\]
By keeping the same dichotomy $m\le x/y$ and $m>x/y$ it is straightforward that
\[
B(x,z)=\frac{6}{\pi^2}\sum_{m=1}^{\infty}\frac{1}{\psi(m)}\min\big(m^{\kappa}z^{1+\kappa},x/m \big)+O(E).
\]
Moreover, \eqref{S3} with the parameter $xz^{-\kappa-1}$ reads
\[
S_{\tht}(xz^{-\kappa-1})=\frac{6}{\pi^2}\sum_{m=1}^{\infty}\frac{1}{\psi(m)}\min\big(m^{\kappa},xz^{-\kappa-1}/m \big)+O(z^{-(1+\kappa)(1-\gamma)}E).
\]
Hence, one has
\[
B(x,z)=z^{1+\kappa}S_{\tht}(xz^{-\kappa-1})+O(z^{(1+\kappa)\gamma}E).
\]
Still choosing $\gamma=\sigma_v-\frac{1}{\log y}$ and $\sigma=\sigma_v$, we have $z^{(1+\kappa)\gamma}\ll 1$. Moreover, \eqref{Enew} implies that for some $c>0$ one has
\[
E\ll x^{\tht} F\big( (1-\tht)\log x \big)\exp\Big(-c\sqrt{\frac{\log x}{\log\log x}}\Big).
\]
This is sufficient to ensure that
\[
E\ll z S_{\tht}(x) \sqrt{\frac{\log\log x}{\log x}}.
\]
Finally we estimate the term $S_{\tht}(xz^{-\kappa-1})$ by Theorem \ref{local-estim}. Inserting this in the estimate for $B(x,z)$ and noticing that in the main term the exponent $1+\kappa-(1+\kappa)\tht $ of $z$ is equal to $1$ gives the expected result. \qed
\smallskip
We may now complete the proof of Theorem \ref{k-log}. It is sufficient to prove the result for $\Tht\neq 0$, since in the case $\Tht=0$ one has $S_{\tht,\Tht}(x)=S_{\tht}(x)$. We put
\[C=1+(1+|\Tht|)/\tht \quad\text{and}\quad
U=U(x)=x(\log x)^{-C}.
\]
Note that we have $C>0$, $C\tht \ge 1$ and $\Tht+ C\tht \ge 1$.
\smallskip
First consider the case $\Tht>0$. Any integer $n$ counted by $S_{\tht,\Tht}(x)$ satisfies $k(n)\le n^{\tht}(\log x)^{\Tht}$, whence $S_{\tht,\Tht}(x)\le B\big(x,(\log x)^{\Tht}\big)$. Now, a lower bound is obtained by noticing that the set of integers $U<n\le x$ counted in $S_{\tht,\Tht}(x)$ contains the integers such that $k(n)\le n^{\tht}(\log U)^{\Tht}$. These deliberations yield the inequalities
\[
B\big(x,(\log U)^{\Tht}\big)-B\big(U,(\log U)^{\Tht}\big) \le S_{\tht,\Tht}(x)\le B\big(x,(\log x)^{\Tht}\big).
\]
Now, using Theorem \ref{logx} to estimate $ B\big(x,(\log U)^{\Tht}\big)$ and $ B\big(x,(\log x)^{\Tht}\big)$, and then replacing $(\log U)^{\Tht}$ by $(\log x)^{\Tht}$ at the price of an admissible error term, one obtains the main term to estimate $ S_{\tht,\Tht}(x)$. For the remaining term, Theorem \ref{logx} and the definition of $U$ imply that
\[
B\big(U,(\log U)^{\Tht}\big)\ll (\log U)^{\Tht}S_{\tht}(U)\ll (\log x)^{\Tht}S_{\tht}(U).
\]
Finally, Theorem \ref{local-estim} with the choice $z=(\log x)^{-C}$ implies that
\[
S_{\tht}(U)\ll z^{\tht}S_{\tht}(x)\ll (\log x)^{-\tht C}S_{\tht}(x) \ll (\log x)^{-1}S_{\tht}(x).
\]
Gathering these estimates, we obtain an upper bound for $B\big(U,(\log U)^{\Tht}\big)$ that provides an admissible error term, and therefore proves the result for $\Tht>0$.
\smallskip
The case $\Tht<0$ is very similar. Any integer counted by $S_{\tht,\Tht}(x)$ either satisfies $n\le U$ with $k(n)\le n^{\tht}$, or $U<n\le x$ with $k(n)\le n^{\tht}(\log U)^{\Tht}$. Moreover, any integer $n\le x$ such that $k(n)\le n^{\tht}(\log x)^{\Tht}$ is counted in $S_{\tht,\Tht}(x)$. We conclude that
\[
B\big(x,(\log x)^{\Tht}\big) \le S_{\tht,\Tht}(x)\le B\big(x,(\log U)^{\Tht}\big)+ S_{\tht}(U).
\]
As before, one uses Theorem \ref{logx} to estimate $B\big(x,(\log U)^{\Tht}\big)$ and $B\big(x,(\log x)^{\Tht}\big)$. This provides the main term. Finally, Theorem \ref{local-estim} with $z=(\log x)^{-C}$ provides
\[
S_{\tht}(U)\ll z^{\tht}S_{\tht}(x)\ll (\log x)^{-\tht C}S_{\tht}(x) \ll (\log x)^{\Tht -1}S_{\tht}(x),
\]
which yields an admissible error term.
\begin{thebibliography}{99}
\bibitem{M} B. Mazur. Questions about powers of numbers. Notices Amer. Math. Soc. 47 (2000), no. 2, 195--202.
\bibitem{RT2013} O. Robert and G. Tenenbaum. Sur la répartition du noyau d'un entier. Indag. Math. (N.S.) 24 (2013), no. 4, 802--914.
\bibitem{T} G. Tenenbaum. Introduction to analytic and probabilistic number theory, Graduate Studies in Mathematics, vol 163. American Mathematical Society, Providence, RI, $3^{\mathrm{rd}}$ edn., 2015.
\end{thebibliography}
\vspace{2ex}\noindent
J\"org Br\"udern \\
Universit\"at G\"ottingen\\
Mathematisches Institut\\
Bunsenstrasse 3--5\\
D 37073 G\"ottingen \\
Germany\\
jbruede@gwdg.de\\
[2ex]
Olivier Robert\\
Universit\'e de Lyon, Universit\'e de Saint-\'Etienne\\
Institut Camille Jordan CNRS UMR 5208\\
25, rue Dr R\'emy Annino\\
F-42000, Saint-\'Etienne\\
France\\
olivier.robert@univ-st-etienne.fr
\end{document}