1
$\begingroup$

Say $f$ is a real valued function that is differentiable (and continuous) whose domain is $[a,b]$. If \begin{equation} f'(x) > 0, \forall x\in(a,b] \text{ and } f'(a) = 0 \end{equation} Can I say that $f$ is strictly increasing in $[a,b]$ or the only thing that I can derive from the above is that it's strictly increasing in $(a,b]$?

Example: \begin{equation} g(x) = \frac{2 + x^5}{8} \end{equation} and \begin{equation} g'(x) = \frac{5x^4}{8} \end{equation} now it's obvious that $ g'(x) >0, \forall x \in (0,1]$, therefore $g$ is strictly increasing in $(0,1]$. Is there anyway (if it's true) I can say that $g$ is strictly increasing in $[0,1]$?

My effort: Let $x\in(a,b]$, then $f$ is: \begin{equation} \left.\begin{array}{c} \text{continuous in $[a,x]$} \\ \text{differentiable in $(a,x)$} \end{array}\right\} \overset{\text{M.V.T}}{\implies} \text{There exists a $\xi \in (a,x)$ such that} \end{equation} \begin{equation} f'(\xi) = \frac{f(x) - f(a)}{x - a} \end{equation} But we know that $f'(y) > 0, \forall y\in(a,b] \supset (a,b)$ and $x-a > 0$ since $x\in (a,b]$. Therefore it must be that: \begin{equation} f(x) - f(a) >0, \text{ where $x\in(a,b]$} \end{equation} Thus, $f$ is strictly increasing in $[a,b]$. Is my proof correct?

$\endgroup$
5
  • 2
    $\begingroup$ $f(x)-f(a)>0$ for $x >a$ by Mean Value Theorem and this makes $f$ strictly increasing in $[a,b]$. $\endgroup$ Commented Jun 16, 2023 at 9:48
  • $\begingroup$ g(x)>g(0) for all x>0 so g is strictly increasing $\endgroup$
    – Tim
    Commented Jun 16, 2023 at 9:48
  • $\begingroup$ Yes, you can so so @Nyquist-er $\endgroup$ Commented Jun 16, 2023 at 9:49
  • $\begingroup$ It should be $g'(x) > 0$ for all $x\in (0,1]$, not $<0$. $\endgroup$
    – stange
    Commented Jun 16, 2023 at 9:53
  • $\begingroup$ @geetha290krm, I edited my question providing an effort of a proof, is that what you had in mind? $\endgroup$
    – Nyquist-er
    Commented Jun 16, 2023 at 10:20

1 Answer 1

0
$\begingroup$

Yes you absolutely can, you just need to have $f \in C^0([a,b])$ and differentiable on $(a,b)$ with $f'>0$. I'm writing an answer but I suggest you try on your own first

Obviously it is strictly increasing on $(a,b)$, so we just need to ensure that $a < x \implies f(a) < f(x)$. If you suppose that $f(a) \geqslant f(x)$, you'll necessarily have $\forall y \in (a,x), \, f(y) < f(a)$, and in particular $$\forall z \in \left(a, a + \frac{x-a}{2}\right],\quad f(z) \leqslant f\left(a + \frac{x-a}{2}\right) < f(a) $$ Then you take the limit for $z \rightarrow a$, and you get $f(a) < f(a)$ which is absurd.

Note that with my proof, I don't even need $f$ to be differentiable, only strictly increasing on $(a,b)$.

$\endgroup$
3
  • $\begingroup$ The condition $f'>0$ can be relaxed. For example, if $f' \ge 0$ and $f'=0$ only at isolated points, the conclusion is the same. $\endgroup$ Commented Jun 16, 2023 at 9:54
  • $\begingroup$ Can still apply mean value theorem? $\endgroup$
    – Tim
    Commented Jun 16, 2023 at 9:57
  • $\begingroup$ as @PierreCarre said there are multiple cases where it is still true and $f' =0$ on some points, but there is no necessary and sufficient condition of $f'$ for the strict increase of $f$. $\endgroup$ Commented Jun 16, 2023 at 10:00

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .