Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm looking for the example of a continuous function $f$, which is:
1. positive and strictly increasing in $(0,a)$ for some $a>0$,
2. $f(0)=0$,
3. $\lim\limits_{x\rightarrow0^{+}}\frac{x}{f(x)}=0$
and for which there doesn't exist an interval $(0,b)$, $b>0$, such that the map $x\rightarrow\frac{x}{f(x)}$ is strictly increasing for all $x\in(0,b)$.
Thank you in advance for your help!

share|cite|improve this question

Choose a positive exponent $u$ and a positive scalar $z$ and consider the function $f$ defined by $$ \color{red}{f(x)=x^u+zx\cos(x^{u-1})}. $$ Then $f(x)>0$ for every $x$ in a neighborhood of $0^+$ if $u<1$ because $f(x)\geqslant x^u-zx$, $f(0)=0$ because $u>0$, and $f$ is increasing near $0^+$ for suitable values of $u$ because $f'(x)$ has the sign of $$ u+z(1-u)\sin(x^{u-1})+zx^{1-u}\cos(x^{u-1}), $$ whose liminf when $x\to0^+$ is $u-z(1-u)$. Thus, every $u>z/(z+1)$ makes $f$ increasing in a neighborhood of $0^+$. Furthermore, $f(x)/x$ is equivalent to $x^{u-1}$ because $u<1$ hence $x/f(x)\to0$ when $x\to0^+$. And $f(x)/x=g(x^{u-1})$ with $g(t)=t+z\cos(t)$ hence the sense of variation of $f(x)/x$ at $x$ when $x\to0^+$ is the opposite of the sense of variation of $g(t)$ at $t=x^{u-1}$, that is, when $t\to+\infty$. Since the derivative of $g$ is $g'(t)=1-z\sin(t)$ and the sine oscillates between $-1$ and $+1$, the sense of variation of $f(x)/x$ at $x$ when $x\to0^+$ changes infinitely many times if $z>1$.

Finally, any $u$ and $z$ do as soon as $$ z/(z+1)<u<1<z. $$

share|cite|improve this answer
Another fine example. I'm very grateful for all your help. – John Nov 5 '11 at 18:49

For any positive integer $n$, let $$f(x)=x\sqrt{2n}\qquad\text{when}\qquad \frac{1}{2n}\le x\le \frac{1}{2n-1}.$$ For $\dfrac{1}{2n+1} \le x \le \dfrac{1}{2n}$, define $f(x)$ by interpolating linearly between the values of $f$ at $\dfrac{1}{2n+1}$ and at $\dfrac{1}{2n}$. Finally, let $f(0)=0$.

It is clear that $f$ is continuous on the interval $[0,1]$. We verify that $f$ is strictly increasing. The only thing that needs to be checked is that $f\left(\dfrac{1}{2n+1}\right)< f\left(\dfrac{1}{2n}\right)$. Note that $$f\left(\dfrac{1}{2n+1}\right)=\frac{\sqrt{2n+2}}{2n+1}\qquad\text{and}\qquad f\left(\dfrac{1}{2n}\right)=\frac{\sqrt{2n}}{2n}.$$ Square both sides. A little manipulation yields the desired inequality.

Now we examine $\dfrac{x}{f(x)}$. Between $\dfrac{1}{2n}$ and $\dfrac{1}{2n-1}$, this has constant value $\dfrac{1}{\sqrt{2n}}$. Thus any interval $(0,b)$ contains infinitely many intervals on which $\dfrac{x}{f(x)}$ is constant.

For the final condition, we need to verify that $\displaystyle\lim_{x\to 0^+}\dfrac{x}{f(x)}=0$. This is easy: $\dfrac{x}{f(x)}$ approaches $0$ roughly like $\sqrt{x}$.

Comment: We arranged for $\dfrac{x}{f(x)}$ to be constant on the intervals $\dfrac{1}{2n} \le x \le \dfrac{1}{2n-1}$. By a small modification of the definition of $f(x)$, we can even arrange for $\dfrac{x}{f(x)}$ to go down a little on these intervals.

share|cite|improve this answer
Thanks a lot, André! Really nice example. I noticed only one typo: in the fourth line it should be $f(0)=0$. Could you, please, explain in details what do you mean by a "small modification" mentioned in Comment? – John Nov 5 '11 at 14:10
@John: Thanks for the typo fix. Will try to expand comment, probably not today (full). – André Nicolas Nov 5 '11 at 15:24
@John: I think if you make the first line $f(x)=(x+x^2/n^2)$ you get there. You want $f$ to grow just a bit faster than linearly. – Ross Millikan Nov 5 '11 at 15:29

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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