1

Let $f:[a,b] \to \mathbb{R}$ be continuous on $[a,b]$ and differentiable on $(a,b)$. Show that $\lim_{x \to a} f'(x) = A > \Rightarrow f'(a)$ exists and equals $A$

I am unable to think of any way to solve this problem

I tried using mean value theorem on $[a,x ]$ for $a < x < b$

$\exists \; c \in ]a,x[ \; \Rightarrow f'(c) = \frac{f(x)-f(a)}{x-a} \Rightarrow \lim_{x \to a} f'(c) = \lim_{x \to a} \frac{f(x)-f(a)}{x-a} $

Now the RHS represents $f'(a)$ but how does it equals $A$? because $f'(c) $ is a constant

Paramanand Singh
  • 87,309
So Lo
  • 1,567
  • The LHS equals $A$ and therefore the RHS also equals $A$ and thus $f'(a) =A$. Note that $f'(c) $ is not a constant but by definition it depends on $x$. – Paramanand Singh Dec 02 '17 at 06:53
  • @ParamanandSingh How does LHS Equal A ? – So Lo Dec 02 '17 at 07:12
  • It is given that $f'(x) \to A$ as $x\to a$. Now $a<c<x$ so as $x\to a$ the variable $c\to a$ and therefore $f'(c) \to A$. – Paramanand Singh Dec 02 '17 at 07:45
  • @ParamanandSingh but $c$ is any fixed value in $]a,x[$ by mean value theorem. How does $c \to b$ if $x \to b$. $c$ can also stay close to $a$ itself. Or my understanding of mean value theorem is wrong ? – So Lo Dec 02 '17 at 10:41
  • 1
    $c$ depends on the interval $[a, x]$. If $x$ changes the interval changes and thus $c$ also changes. To make thing concrete if $a=0$ then the value of $c$ for interval $[0,0.1]$ and for interval $[0,0.2]$ are not necessarily same. Don't you think so? And $c\to a$ as $x\to b$ and not $c\to b$ as $x\to b$. This happens because $a<c<x$ and when you take limit of this inequality as $x\to a$, $c$ must tend to $a$. – Paramanand Singh Dec 02 '17 at 11:59
  • 1
    If you want to deal with $b$ then the situation has to be different like $x<c<b$ and then let $x\to b$. In that case $c\to b$. – Paramanand Singh Dec 02 '17 at 12:02
  • @ParamanandSingh thank you so much for clearing that up. – So Lo Dec 02 '17 at 12:04
  • 1
    There is a minor typo in my comment which I am fixing here. And $c\to a$ as $x\to\color{red} {a} $ and not $c\to b$ as $x\to b$. – Paramanand Singh Dec 02 '17 at 12:13

1 Answers1

1

You are very close. A slight rewrite will give you the result.

So take $x_n = a + \epsilon_n$, where $\epsilon_n > 0$ are so small that $a+\epsilon_n < b$ and $\epsilon_n \to 0$ as $n \to \infty$.

Apply the mean value theorem to $[a,x]$ to conclude that there is $c_n \in (a,x_n)$ such that $f'(c_n) = \frac{f(x)-f(a)}{x-a}$. Now, take limits on both sides : the right side is the derivative of $a$, and the right hand side is $\lim_{n \to \infty} f'(c_n)$, which exists and equals $A$. Hence, the result follows.

Sarvesh Ravichandran Iyer
  • 76,105