4
$\begingroup$

Suppose we have a differentiable function $ g $ that maps from a real interval $ I $ to the real numbers and suppose $ g'(r)>0$ for all $ r$ in $ I $. Then I want to show that $ g^{-1}$ is differentiable on $g(I). $

Intuitively this makes sense but I can't come up with a neat proof. I was thinking to use the mean value theorem but I'm not sure if that would get me anywhere.

$\endgroup$
0

2 Answers 2

8
$\begingroup$

By definition of derivative we have

$$g'(r)=\lim_{x\to r}\frac{g(x)-g(r)}{x-r}=\lim_{x\to r}\frac{g(x)-g(r)}{g^{-1}(g(x))-g^{-1}(g(r))}$$

Since $g'(r)>0$, this means

$$\lim_{x\to r}\frac{g^{-1}(g(x))-g^{-1}(g(r))}{g(x)-g(r)}=\frac{1}{g'(r)}$$

Let $y\in g(I)$, say $g(x)=y$ for some $x\in I$, and let $y_n\to y$. Then for each $n$ there is a unique $x_n\in I$ with $g(x_n)=y_n$. Further, one can show that if $f$ is continuous and strictly monotone on an interval $I$, then $f^{-1}$ is continuous on $f(I)$, hence we have $x_n\to x$, it follows that

$$\lim_{n\to\infty}\frac{g^{-1}(y_n)-g^{-1}(y)}{y_n-y}=\lim_{n\to\infty}\frac{x_n-x}{g(x_n)-g(x)}=\frac{1}{g'(x)}$$

Hence $g^{-1}$ is differentiable.

$\endgroup$
7
  • $\begingroup$ "Hence" is not fully justified; the last limit you wrote is not exactly the same as the definition of derivative. $\endgroup$
    – user147263
    Commented Feb 9, 2015 at 1:34
  • $\begingroup$ Why can we say that $ x-r $ is the same thing as $ g^{-1}(g(x))-g^{-1}(g(r)) $? does injectivity have anything to do with this $\endgroup$
    – Rina
    Commented Feb 9, 2015 at 1:47
  • $\begingroup$ @FamousBlueRaincoat Good point, I've added more details $\endgroup$
    – Frank Lu
    Commented Feb 9, 2015 at 2:00
  • $\begingroup$ @Rina I've added more details in my answer $\endgroup$
    – Frank Lu
    Commented Feb 9, 2015 at 2:00
  • $\begingroup$ thanks Frank! this clears up a lot $\endgroup$
    – Rina
    Commented Feb 9, 2015 at 2:41
-1
$\begingroup$

If $g'(r)>0$ then $g(r)$ in injective.

Exist $g^{-1}(r)$.

Then use the chain rule for derivate $g^{-1}(g(r))=r$.

$\endgroup$
7
  • $\begingroup$ This is true; but how would that help prove the claim? $\endgroup$
    – Rina
    Commented Feb 9, 2015 at 1:09
  • $\begingroup$ I see; what role does injectivity play? Why does $ g^{-1}(r) $ exist $\endgroup$
    – Rina
    Commented Feb 9, 2015 at 1:16
  • 1
    $\begingroup$ @Rina: A function has a (set-theoretic) inverse precisely when it's injective and surjective. Since a function $f : A\to B$ always surjects onto its image $im(f)$, there will be an inverse $f^{-1} : im(f)\to A$ precisely when $f$ is injective. $\endgroup$
    – Stahl
    Commented Feb 9, 2015 at 1:19
  • $\begingroup$ oh okay got it. thanks a lot $\endgroup$
    – Rina
    Commented Feb 9, 2015 at 1:23
  • 4
    $\begingroup$ To apply the chain rule to the composition $g^{-1}\circ g$, one needs to know that both functions involved are differentiable. Which is to be proved. $\endgroup$
    – user147263
    Commented Feb 9, 2015 at 1:31

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