I have to solve this limit:

$$\lim\limits_{x\to 0} \left( \dfrac{e^x - 1 }{x} \right) $$

I can easily do it by using hospital's rule, numerator becomes $e^0$ and denominator becomes $1$.

How should I solve that limit by using definition of $e$ as $e= \lim\limits_{n \to \infty } {\left( {1 + \frac 1n } \right)}^n$

  • $\begingroup$ Let n= 1/x. Then "$x\to 0$" becomes "$n\to\infty$". $\endgroup$ – user247327 Mar 7 '17 at 14:58
  • $\begingroup$ Can you use the series expansion of $e^{x}$? $\endgroup$ – mrnovice Mar 7 '17 at 14:58
  • $\begingroup$ You would use L'Hopital (involving computing the derivative of $\exp$ at $0$, and invoking a rather non-trivial theorem to conclude) to compute what is by definition the derivative of $\exp$ at $0$? $\endgroup$ – Clement C. Mar 7 '17 at 15:01
  • $\begingroup$ See this one $\endgroup$ – Juniven Mar 7 '17 at 15:16
  • $\begingroup$ @ΘΣΦGenSan yes,I wanted to solve by using definition of x given in my question $\endgroup$ – Noddy Mar 7 '17 at 15:20

Method 1: Defintion of derivative Hint:

$$\lim_{x\to 0}\frac{e^x-e^0}{x-0}.$$

This is nothing but $\frac{d}{dx}e^x|_{x=0}$.

Method 2:Taylor Series

You could also plug in the definition of $e^x$ as a Taylor series and collect all higher order terms as $O(x^2)$: $e^x=1+x+O(x^2)$.

Method 3: Euler

The last method was the method Euler used, which is not rigorous but still very nice.

Look at $$\lim_{x\to 0}\frac{a^x-1}{x}$$, Euler first showed convergence (not sure if he even cared about it :D) and then said that we want to choose $a$ such that the limit is $1$. So approximately $$\frac{a^x-1}{x}\approx 1\implies a^x\approx 1+x \implies a \approx (1+x)^{1/x}.$$ Now, substitute $n=1/x$, hence $n\to \infty$ and we get $$a\approx \lim_{n\to \infty}\left(1+\frac{1}{n} \right)^n=e$$


You can see that:


So since $x\mapsto e^x$ is differentiable,

$$\lim_{x\to 0}\frac{e^x-1}{x}=\lim_{x\to 0}\frac{e^x-e^0}{x-0}=\lim_{x\to 0}\frac{d}{dx} (e^x)=\lim_{x\to 0}e^x=1.$$

  • 1
    $\begingroup$ This is basically a proof of l'Hospital rule. It does not answer OP's question... $\endgroup$ – Roberto Rastapopoulos Mar 7 '17 at 15:01
  • $\begingroup$ Also, this is basically wrong. You are using continuity of the derivative of $\exp$, for which reason? (1) you do not need it, as this is the definition of $\exp'(0)$, (2) the same argument ought to work for functions whose derivative at $0$ exists, but who are not continuously differentiable. $\endgroup$ – Clement C. Mar 7 '17 at 15:04
  • $\begingroup$ No. This shows the limit it f'(x) while LH says that f/g has the same limit as f'/g'. $\endgroup$ – marty cohen Mar 7 '17 at 15:06
  • 1
    $\begingroup$ @martycohen Roberto's comment is wrong indeed, but so is the step which writes $\lim_{x\to 0} \frac{d}{dx} e^x$ (it is only true here because $\exp$ is nice, but replacing $\exp$ by some $f$ differentiable at $0$ (and nowhere else) should still be OK in the OP's question). It should be $\frac{d}{dx} e^x \mid_{x=0}$, or even better $\exp'(0)$. $\endgroup$ – Clement C. Mar 7 '17 at 15:09

Here is an attempt to use only integers.

To show $\lim_{x \to 0} \dfrac{e^x-1}{x} = 0 $ using $e = \lim_{n \to \infty} \left(1+\frac1{n}\right)^n $.

From this, $e^x = \lim_{n \to \infty} \left(1+\frac1{n}\right)^{nx} $.

This is a double limit, which is often problematical.

For general $x$, this seems to require the generalized binomial theorem. To avoid this, I will assume $x = \frac1{m}$ as integer $m \to \infty$, and let $n=km$ as integer $k \to \infty$.

Then $e^{1/m} = \lim_{k \to \infty} \left(1+\frac1{mk}\right)^{k} $. Now we can use the binomial theorem.

$\begin{array}\\ \left(1+\frac1{mk}\right)^{k} &=\sum_{j=0}^k \binom{k}{j}\frac1{(mk)^j}\\ &=\sum_{j=0}^k \frac{\prod_{i=0}^{j-1}(k-i)}{j!}\frac1{(mk)^j}\\ \text{so}\\ f(m, k) &=\dfrac{\left(1+\frac1{mk}\right)^{k}-1}{1/m}\\ &=m\sum_{j=1}^k \frac{\prod_{i=0}^{j-1}(k-i)}{j!}\frac1{(mk)^j}\\ &\le m\sum_{j=1}^k \frac{1}{j!}\frac1{m^j}\\ &= \sum_{j=1}^k \frac{1}{j!}\frac1{m^{j-1}}\\ &=1+ \sum_{j=2}^k \frac{1}{j!}\frac1{m^{j-1}}\\ &\le 1+ \frac1{m}\sum_{j=2}^k \frac{1}{j!}\frac1{m^{j-2}}\\ &= 1+ \frac1{m}\sum_{j=0}^{k-2} \frac{1}{(j+2)!}\frac1{m^j}\\ \end{array} $

Since $\sum_{j=0}^{k-2} \frac{1}{(j+2)!}\frac1{m^j}$ is easily shown to be bounded by $1$ (actually, by $\sum_{j=0}^{k-2} \frac{1}{(j+2)!} = e-2$), $f(m, k) \le 1+\frac1{m}$.

Since $f(m, k) \ge 1$ (from its first term), $\lim_{m \to \infty} f(m, k) = 1$.


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.