Take the 2-minute tour ×
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.

I knew that $e^x=\lim \limits_{n\to+\infty }{\left(1+\frac{x}{n}\right)^n}$. But I've never seen its proof. So I tried to prove it using $\exp(\ln x)=\ln(\exp(x))=x$. Here is what I've tried so far :

$$ \left(1+\frac{x}{n}\right) ^n=e^{n\ln(1+\frac{x}{n})}$$ $$\text{I'll now study just } {n\ln\left(1+\frac{x}{n}\right)}.$$$$ \text{If this function has the line }y=x \text{ as oblique asymptote, then the equality is proven.}$$

$$ n\ln\left(1+\frac{x}{n}\right) = n\ln\left(\frac{n+x}{n}\right)$$

$$=n[\ln(n+x)-ln(n)]$$ $$=n\left[\int_1^{n}\frac{dt}{t}+\int_{n}^{x}\frac{dt}{t}-\int_1^{n}\frac{dt}{t}\right]$$ $$=n[\ln(x)-\ln(n)]$$

But I just don't know how to show that this expression has an oblique asymptote $y=x$. I've thought that if there is an oblique asymptote as $n$ goes to infinity, than for a huge $n$, we have :

$$\ln\left(1+\frac{x}{n}\right)\approx \frac{x}{n}\approx0$$ Which looks correct but we could have any other function $f(x)$, $\ln\left(1+\frac{x}{n}\right)\approx\frac{f(x)}{n}\approx 0$. Which doesn't prove the oblique asymptote because $x$ is constant.

So how can prove $e^x=\lim \limits_{n\to +\infty } \left(1+\frac{x}{n}\right)^n$? And where did I go wrong?

share|improve this question
8  
What is your definition of $e^x$? –  Vectk Jun 14 '13 at 6:39
3  
What is your definition of $\exp (x)$? –  Git Gud Jun 14 '13 at 6:40
    
@moray95: Your integral should be $\int_1^{nx}\frac{dt}{t}+ \int_{nx}^{nx+1} \frac{dt}{t}- \int_1^n \frac{dt}{t}$. –  Seirios Jun 14 '13 at 6:41
    
I'm using the definition $\exp(\ln(x))=\ln(\exp(x))=x$ –  moray95 Jun 14 '13 at 6:50
1  
Are you defining $\exp$ as the inverse of $\ln$, is that it? –  Git Gud Jun 14 '13 at 6:53

9 Answers 9

Let's start with the "where did I go wrong" part of the question. Where you wrote

$$=n\left[\int_1^{n}\frac{dt}{t}+\int_{n}^{x}\frac{dt}{t}-\int_1^{n}\frac{dt}{t}\right]$$

you should have written

$$=n\left[\int_1^{n}\frac{dt}{t}+\int_{n}^{n+x}\frac{dt}{t}-\int_1^{n}\frac{dt}{t}\right]$$

Note, the correct upper limit in the middle integration is $n+x$, not just $x$. Otherwise you were on the right track. The corrected integral leaves us with

$$n\ln\left(1+\frac{x}{n}\right)=n\int_n^{n+x}{dt\over t}$$

Now as soon at $n\gt x$, we have

$${1\over n+|x|}\le{1\over t}\le{1\over n-|x|}$$

on the interval $t\in[n-|x|,n+|x|]$, which certainly includes the interval between $n$ and $n+x$. If you are careful with the minus signs, you can conclude that

$${nx\over n+x}\le n\int_n^{n+x}{dt\over t}\le{nx\over n-x}$$

and it now follows easily that

$$\lim_{n\to\infty}n\ln\left(1+\frac{x}{n}\right)=x$$

share|improve this answer

If you're allowed to use Taylor (power) expansions this is pretty simple:

$$n\log\left(1+\frac xn\right)=n\sum_{k=1}^\infty (-1)^{k+1}\frac{x^k}{k\,n^k}=n\left(\frac xn+\mathcal O\left(\frac1{n^2}\right)\right)=$$

$$=x+\mathcal O\left(\frac1n\right)\xrightarrow[n\to\infty]{}x$$

share|improve this answer
1  
Great response but I'd like to prove it without the power series... –  moray95 Jun 14 '13 at 15:51
1  
Ok...did you read my comment under your question? Because you haven't yet corrected your work there... –  DonAntonio Jun 14 '13 at 15:53
    
Edited my question with your remark but still, I still have the same thing at the end... –  moray95 Jun 14 '13 at 17:07

I don't know if it helps you, it is just a suggestion, if you know the fundamental limite: $$\lim_{n\to \infty}(1+\frac{1}{n})^n=e$$ Then you have for $$\lim_{n\to \infty}(1+\frac{x}{n})^n$$ replacing $k=\frac{n}{x}$ we get $$\lim_{n\to \infty}(1+\frac{1}{k})^{kx}= (\lim_{k\to \infty}(1+\frac{1}{k})^{k})^x =e^x$$

share|improve this answer
    
I've thought about something like that but then I'll need to prove first $\lim_{n\to \infty}(1+\frac{1}{n})^n=e$. –  moray95 Jun 15 '13 at 10:23
    
@moray95 This is often taken as the definition of $e$. The only other definition I know of is the power series, which you can show is equal to this expression using binomial expansion. –  you-sir-33433 Jan 7 at 20:44

Start with the functions $$ f_n(x) = \left(1 + \frac{x}{n}\right)^n. $$ Then $$ f'_n(x) = \left(1 + \frac{x}{n}\right)^{n-1} = \left(1 + \frac{x}{n}\right)^{-1}f_n(x) $$ If we take the limit and call $f(x) = \lim_{n\rightarrow\infty}f_n(x)$, then $$ f'(x) = f(x) $$ and $f(0) = 1$. This first-order ODE has the unique solution $f(x) = e^x$.

share|improve this answer

If you already know the exponential function, then you probably also know the inequality $e^x\ge 1+x$. Then also

$$e^x=\frac1{e^{-x}}\le\frac1{1-x}=(1+x)\frac1{1-x^2}$$

Now $e^x=(e^{\frac xn})^n$, so for $n>|x|$

$$ \left(1+\tfrac xn\right)^n \le e^x \le \left(1+\tfrac{x}{n}\right)^n\frac1{\left(1-\tfrac{x^2}{n^2}\right)^n}\le\left(1+\tfrac{x}{n}\right)^n\frac1{1-\tfrac{x^2}{n}} $$

The last inequality is by the Bernoulli inequality.

share|improve this answer

Here's a way to do it with the integral: \begin{align*} n\log{\left(1+{x\over n}\right)} & = n\int_1^{1+{x/ n}}{dt\over t} \end{align*} and by the simplest conceivable estimate \begin{align*} {x\over 1+x/n} = n{x/n\over1+{x/ n}}\leq n\int_1^{1+{x/ n}}{dt\over t} \leq n{x\over n} = x. \end{align*} (The inequalities hold when $x$ is negative too, provided the expressions are defined.) Now make $n\to\infty$ and apply the squeeze theorem.

share|improve this answer

$f(x) = e^x$ is the only solution to the differential equation $\dfrac{dy}{dx} = y$ with $f(0)=1$.

To approximate $f(a)$, we can use Euler's method on the interval $[0,a]$ with $n$ subintervals.

$f(0) = 1, f'(0)=1 \implies f(\frac{a}{n}) \approx 1+\frac{a}{n}$

$f(\frac{a}{n}) \approx 1+\frac{a}{n}, f'(\frac{a}{n}) \approx 1+\frac{a}{n} \implies f(\frac{2a}{n}) \approx 1+\frac{a}{n} + \frac{a}{n}(1+\frac{a}{n}) = (1+\frac{a}{n})^2$

.

.

.

$f(a) \approx (1+\frac{a}{n})^n$

Since Euler's method actually converges in the limit, we have

$$e^a = \lim_{n \to \infty} (1+\frac{a}{n})^n$$

share|improve this answer
    
p.s. If you think long and hard about this same proof, only applied to the line segment from $0$ to $i$ in the complex plane, you will understand why $e^{i\pi}=-1$ –  Steven Gubkin Jan 7 at 19:54

Let's say you define $e^x$ by the derivative $\frac{d}{dx} e^x=e^x$. Then, we try the same to what you have here. $\frac{d}{dx}(1+\frac{x}{n})^n=(1+\frac{x}{n})^{n-1}$. And, as n approaches infinity, n and n-1 essentially boils down to the same thing. Then, $\frac{d}{dx}(1+\frac{x}{n})^n=(1+\frac{x}{n})^n$.

share|improve this answer

Since $X=e^{\log X}$, we can write $$\lim_{n \to \infty} \bigg(1+\frac{x}{n}\bigg )^n=e^{\log\lim_{n \to \infty} \bigg(1+\frac{x}{n}\bigg )^n} =e^{\lim_{n \to \infty} n\log \bigg(1+\frac{x}{n}\bigg )}$$

Notice that the exponent on $e$ is an indeterminate form of type $\infty\cdot 0$, so rewrite the limit in the exponent as $\frac{\log \big(1+\frac{x}{n}\big)}{\frac{1}{n}}$ and use L'Hopital's Rule to show the the limit in the exponent of $e$ is indeed $x$.

share|improve this answer

Your Answer

 
discard

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.