I am reading the article "A Direct Approach to the Mellin Transform".
To prove Lemma 2 (see below), the authors used the following inequality
\begin{equation} (I): \hspace{1cm}
|u^h-1|\leq |h| |\log u| u^h,
\end{equation}
where $h\in\mathbb{C}$ $\left(|h|<\delta\right)$ and $u\in\mathbb{R}_+$. Then they wrote
\begin{align}(II): \hspace{1cm}
|f_{\mathcal{M}}^{\wedge}\left(s+h\right)-f_{\mathcal{M}}^{\wedge}\left(s\right)|
&\leq
\int_{0}^{\infty}|f\left(u\right)\left(u^{s+h-1}-u^{s-1}\right)|\mathrm{d}u\\
&\leq |h|
\int_{0}^{\infty}|f\left(u\right)\left(\log u\right)u^{s+h-1}|\mathrm{d}u,
\end{align}
where $f_{\mathcal{M}}^{\wedge}\left(s\right)$ is the Mellin transform of $f$. However, I can not prove the first inequality, and I think it is wrong.
I want to use the elementary inequality
\begin{equation}
|e^z-1|\leq |z|e^{|z|}.
\end{equation}
But what I can get is
\begin{equation}
|u^h-1|=|e^{h\log u}-1|\leq |h| |\log u| e^{|h \log u|}.
\end{equation}
My question: Can we find a corrected version of inequality (I) s.t. the second inequality (II) can be simply proved by using the corrected version ?

2026-03-25 17:38:00.1774460280
Find a correct form for $|u^h-1|\leq |h| |\log u| u^h$
59 Views Asked by Bumbble Comm https://math.techqa.club/user/bumbble-comm/detail At
1
Ok, there seems to be a misprint in (2.5). I suppose it should be $\log u$ times the given power of $u$ (as they use below). Supposing this, if the authors have not made similar mistakes in proving (the correct version of) (2.5) then the last inequality is correct. Let $|h|<\delta$. My the MVT (using a path $th, 0\leq t\leq 1$ from 0 to $h$) we get $$ |u^h -1| = |e^{h \log u} - 1| \leq |h||\log u| \sup_{0\leq t\leq 1} |u^{th}| \leq |h||\log u| \sup_{|k|<\delta} |u^k| $$
For $0<u\leq 1$, $|k|<\delta$ you have $|u^{k}|\leq u^{-\delta}$ and for $u\geq 1$ you have $|u^k|\leq u^{\delta}$ which fits with the (correct version of) inequality (2.5) to give the stated bound. The authors (or referees) should have been more vigilant.