Instead of using L'Hospital's rule can this be proved by using the definition of the limit and or the squeeze theorem.
-
1What have you tried? You could solve this quickly using the Taylor series for the exponential. – Joel Mar 22 '18 at 03:59
-
I have tried using a epilson on two arguement and the Taylor series didn't work out too well for me. I'm trying to prove this as apart of a problem set for my real analysis class and I think the necessary proof is from a squeeze theorem arguement. – Alexander Quinn Mar 22 '18 at 04:01
2 Answers
Use Mean Value Theorem to deduce that $1+u+u^{2}/2\leq e^{u}\leq 1+u+u^{2}$ for $0\leq u\leq 1/2$, then for $n\geq 2$, we have $1/n\leq 2$ and hence $1+(1/n)+1/(2n^{2})\leq e^{1/n}\leq 1+(1/n)+1/n^{2}$, so $n(e^{1/n}-1)\rightarrow 1$.
The use of Mean Value Theorem here:
Let $\varphi(u)=1+u+u^{2}-e^{u}$, $0\leq u\leq 1/2$, then for $0<u\leq 1/2$, $\varphi(u)=\varphi(u)-\varphi(0)=\varphi'(\xi)u$, where $\xi$ lies in between $0$ and $u$.
Since $\varphi'(\xi)=1+2\xi-e^{\xi}$, we investigate the function $\eta(u)=1+2u-e^{u}$ for $0\leq u\leq 1/2$. Note that for $0<u\leq 1/2$, $\eta(u)=\eta(u)-\eta(0)=\eta'(\omega)u=(2-e^{\omega})u>0$ for some $\omega$ lies in between $0$ and $u$, in particular, $\omega$ lies in between $0$ and $1/2$.
So $\eta(\xi)>0$ and hence $\varphi'(\xi)=\eta(\xi)>0$, so $\varphi(u)>0$.

- 55,591
-
Could you please give more information a bit more how did you use the mean value theorem – Alexander Quinn Mar 22 '18 at 04:03
-
Thank you very much is it possible to solve this by using a epilson on two arguement or by using the squeeze theorem? – Alexander Quinn Mar 22 '18 at 04:10
-
My reasoning finally goes with Squeeze Theorem, just the middle with the aid of Mean Value Theorem, no? I see no hope in doing $\epsilon$-argument, as you know, we all are supposed to know the answer of the limit, and then running $\epsilon$-argument to confirm the accuracy, but not the other way around. – user284331 Mar 22 '18 at 04:12
-
-
So $1+1/(2n)\leq n(e^{1/n}-1)\leq 1+1/n$ by the inequalities. Now $1+1/(2n)\rightarrow 1$ and $1+1/n\rightarrow 1$, so squeeze it, the limit is then $1$. – user284331 Mar 22 '18 at 04:16
-
Would that be a full proof to answer this question using the squeeze theorem I feel like you would need a bit more? – Alexander Quinn Mar 22 '18 at 04:17
-
Yes, that is enough, I don't see any gap. For Squeeze Theorem we have a rigorous proof, so it is totally safe to use that. – user284331 Mar 22 '18 at 04:19
-
So once you have done that could you not just do a epilson arguement just for the sake of doing so. – Alexander Quinn Mar 22 '18 at 04:21
-
Then the $\epsilon$-argument is in some sense the proof of Squeeze Theorem instantiating to this question, but why doing that? The point is that, Squeeze Theorem has proved in most of the textbooks. – user284331 Mar 22 '18 at 04:23
-
-
Are there any other unique methods of forming a proof for this question? – Alexander Quinn Mar 22 '18 at 04:33
-
What do you mean by other unique methods of forming a proof? Like what? – user284331 Mar 22 '18 at 04:34
-
Well besides using the squeeze theorem or Mean value theorem and within the realm of real analysis – Alexander Quinn Mar 22 '18 at 04:35
-
-
@user284331 I'm not certain that Taylor's Theorem is the "fastest way." See my solution as a fairly efficient way forward. – Mark Viola Mar 22 '18 at 04:48
In THIS ANSWER, I showed using only the limit definition of the exponential function and Bernoulli's Inequality that the exponential function satisfies the inequalities
$$1+x\le e^x\le \frac1{1-x}\tag 1$$
for $x<1$.
Letting $x=\frac1n$ for $n>1$ in $(1)$, subtracting $1$, and multiplying the result by $n$ reveals
$$1\le n(e^{1/n}-1)\le \frac{1}{1-\frac1n}$$
whence applying the squeeze theorem yields the coveted result.

- 179,405
-
So how would you do this from the beginning without using that previous answer you gave – Alexander Quinn Mar 22 '18 at 04:45
-
@AlexanderQuinn I would develop the inequalities in $(1)$ using the approach in the referenced answer or develop a different set of inequalities using other tools such as Taylor's Theorem. Another approach would be to transform variables. – Mark Viola Mar 22 '18 at 04:47
-
-
-
@AlexanderQuinn The tools I used were about as elementary as one can use to fulfill your request efficiently. Tools: Bernoulli's Inequality, the limit definition of the exponential function, arithmetic of inequalities, and the squeeze theorem. – Mark Viola Mar 22 '18 at 04:53
-
-
-