0
$\begingroup$

Related to this question.

For $x_+ \in (0,\infty)$, $a \in \mathbb{R}$ let $F\colon[0,x_+] \to [a,\infty)$ be a twice continuous differentiable (in $(0,x_+)$) function with $f := F'$, $f(x) > 0$, and $f'(x) < 0$ for all $x \in (0,x_+]$. Moreover, we assume that $$\lim_{x \to 0} f(x) = \infty$$ and $F(0) = a$ holds.

The question: Does this implies that there exists a $\beta \in (0,\infty)$ such that $f(x)f(y) \ge \beta f(xy)$ for all $x,y \in (0,x_+]$.

This special version came in my mind after i analyzed that the counterexample here relies on the fact that the function $f$ is so steep that the primitive integral $F$ has limit $\lim_{x \to 0}F(x) = -\infty$.

$\endgroup$

1 Answer 1

2
$\begingroup$

Define $f$ on $]0,e^{-2}]$ by $$f(x) = \frac{1}{-\sqrt{x}\ln(x)}.$$ Then $f$ is positive, decreasing since $$\frac{\mathrm{d}}{\mathrm{d}x}\big(-\sqrt{x}\ln(x)\big) = \frac{-\ln(x)-2}{2\sqrt{x}} \ge 0 \text{ for all } x \in (0,e^{-2}].$$ Moreover $f(x) \to +\infty$ as $x \to 0$. But since $f(x) = o(x^{-2/3})$ as $x \to 0$, we can define $F$ on $[0,e^{-2}]$ by $$F(x) = \int_0^x f(t) \mathrm{d}t.$$ The assumptions hold with $a=0$. Yet, for all $x \in (0,e^{-2}]$, $$\frac{f(x)^2}{f(x^2)} = \frac{-x \ln(x^2)}{x \ln^2(x)} = \frac{-2}{\ln(x)} \to 0 \textrm{ as } x \to 0,$$ so the ratio $f(x)f(y)/f(xy)$ is not bounded away from $0$.

$\endgroup$
2
  • $\begingroup$ Looks very good the example. I still have to go through the whole thing step by step. In the last equation you have to fix f(x)^2/(f(x)^2) to f(x)^2/f(x^2). $\endgroup$ Jun 25 at 19:47
  • $\begingroup$ @maximilian43 I have just corrected the typo. $\endgroup$ Jun 25 at 20:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.