Using several times L'Hospital Rule I got $$\lim_{x \rightarrow +\infty}{e^x \left (e - \left(1+\dfrac{1}{x}\right )^x\right)} = +\infty.$$ Is it possible find this limit without L'Hospital?
-
1It is always possible to find a limit without l'Hopital as at some point l'Hopital was proving (hopefully without invoking l'Hopital). – Fabian Aug 15 '12 at 20:49
-
5what we allowed to use? – Norbert Aug 15 '12 at 20:55
-
You are right. I never watched that. Thanks. – jon jones Aug 16 '12 at 00:17
2 Answers
The natural thing to do is to look at the logarithm of $\left(1+\frac{1}{x}\right)^x$, that is, at $x\log\left(1+\frac{1}{x}\right)$. Use the series $$\log(1+t)=t-\frac{t^2}{2}+\frac{t^3}{3}-\frac{t^4}{4}+\cdots.$$ From this we can obtain good estimates of the difference between $e$ and $(1+1/x)^x$ when $x$ is large. For the calculation, the series expansion of $e^t$ is useful.

- 507,029
-
4
-
Taylor series is much more abvanced approch than L'Hopital rule. So I think this is not a solution or a hint. – Norbert Aug 15 '12 at 21:11
-
-
1@AsafKaragila: We may be cursed with a similar sense of humour. I thought of but this time resisted writing, instead of logarithm, natural logarithm (with the emphasis). – André Nicolas Aug 15 '12 at 21:14
-
3@Norbert: The Taylor series is only "more advanced" in the sense that it's taught later in the curriculum. – Aug 15 '12 at 21:18
-
@jonjones: We basically want to know how close $(1+1/x)^x$ and $e$ are, for large $x$. We can get good estimates by using the Binomial Theorem, which is in a sense more elementary, but requires much more attention to detail. "Elementary" arguments can require more mathematical experience than less elementary ones. – André Nicolas Aug 15 '12 at 21:19
-
how are taylor series more advanced than l'hopital? if youre working with analytic functions you might as well use series expansions, some people would say it's much more natural – butt Aug 15 '12 at 22:20
-
1The OP only said "without l'Hospital", not "without l'Hospital or Taylor series". AND he did not include label "homework" so why not use any method? – GEdgar Aug 15 '12 at 23:35
We need to prove that, $$\lim_{x \rightarrow +\infty}{e^x \left (e - \left(1+\dfrac{1}{x}\right )^x\right)} = +\infty.$$
consider $$\lim_{x \rightarrow +\infty}{e^x \left (e - M\right)} $$
where, $$M = \left(1+\dfrac{1}{x}\right )^x$$
if we prove that $M$ has a finite limit, we are done.
Note that,
1. M is increasing function of x
2. M is bounded above
first one you can prove as an exercise, for second $$M = \left(1+\dfrac{1}{x}\right )^x = \left(\left(1+\dfrac{1}{x}\right )^{x/k} \right)^k < \left(\frac{1}{1-\frac{1}{x} \cdot\frac{x}{k}}\right)^k = \left ( \frac{1}{\left(1-\frac{1}{k}\right)^k}\right)$$
so that, $$M< \frac{1}{\left(1-\frac{1}{k}\right)^k}$$ for any whole k.
$$\lim_{x \rightarrow +\infty}{e^x \left (e - M\right)} = \lim_{x \rightarrow +\infty}{e^x L} = +\infty $$

- 1,132
-
$\lim\limits_{x\to\infty} \left(1+\frac1x\right)^x = e$. So the argument doesn't work. – Daniel Fischer Nov 28 '13 at 18:57
-
-
1So if you look at $\lim\limits_{x\to\infty} e^x(e-M)$, you look at $\lim\limits_{x\to\infty} e^x\cdot 0$. – Daniel Fischer Nov 28 '13 at 19:02
-
@DanielFischer, ahh I see. thanks for pointing out that, I will try to find better answer. – chatur Nov 28 '13 at 19:14