0

I/
I have a very simple question.
If $\forall n \in \mathbb{N}, a_n>0$ and $\lim_{n\rightarrow \infty }a_n=L>0$ does this imply that $\lim_{n\rightarrow \infty }\sqrt[n]{a_n}=\lim_{n\rightarrow \infty }\sqrt[n]{L}=1$?

II/ I ve answered yes to it and here is my proof.
Let define $b_{n\geq 2}=\frac{a_n}{a_{n-1}}\Rightarrow \lim_{n \to \infty }b_n=\lim_{n \to \infty }\frac{a_n}{a_{n-1}}=\frac{L}{L}=1$ (By limit arithmetic).
It is obvious too that $\forall n \geq 2 \in\mathbb{N}, b_n>0$ and that $\frac{a_n}{a_1}=b_2b_3...b_n \Rightarrow a_n=b_2b_3...b_na_1$
Now because we know that: $\lim_{n \to \infty }\sqrt[n]{b_2b_3...b_n}=1=\lim_{n \to \infty }b_n$ we can writte: $\lim_{n \to \infty }a_n^{1/n}=\lim_{n \to \infty }\sqrt[n]{b_2b_3...b_na_1}=\lim_{n \to \infty }\sqrt[n]{b_2b_3...b_n} \sqrt[n]{a_1}=\lim_{n \to \infty }\sqrt[n]{b_2b_3...b_n} \lim_{n \to \infty }\sqrt[n]{a_1}=1\cdot 1=1$. By arithmetic of limit and by knowing that:$\lim_{n \to \infty }\sqrt[n]{a_1}=1$ Hence $\lim_{n \to \infty }a_n^{1/n}=1=\lim_{n\rightarrow \infty }\sqrt[n]{L}$ Q.E.D.

I have two question:
1-Is my proof correct?
2-Does somebody know how to proove it by definition of limit (in the case this is true).

Thank you for your help.

  • 1
    Can you use that $\lim_{n\to \infty }\sqrt[n]{c}=1$ for every constant $c > 0$? – Martin R Nov 17 '22 at 12:39
  • @MartinR For what i ve learned and what i ve founded ont he net it seems to me that yes: https://math.stackexchange.com/questions/2681154/proof-verification-limc1-n-1-for-c0 – X0-user-0X Nov 17 '22 at 12:40
  • 3
    Well, then note that $0 < L/2 < a_n < 2L$ for all sufficiently large $n$, and squeeze ... – Martin R Nov 17 '22 at 12:41
  • 1
    If you have not seen the theorem you linked, you probably need to prove it – Taladris Nov 17 '22 at 12:43
  • @MartinR There is a lot questions and answers on mathstack prooving that if $c>0$ so $lim , c^{1/n}=1$ for exemple here: https://math.stackexchange.com/questions/3348516/if-a0-prove-that-lim-a1-n-1 . So i don't really understand your remark. – X0-user-0X Nov 17 '22 at 12:45
  • @Taladris Thk for your answer. We have all ready prooved the theorem concerning the "sequence meaning" as for exemple here the geometric mean. – X0-user-0X Nov 17 '22 at 12:47
  • @MartinR Oh sorry i ve understood your remark now. Yes i can use the fact that: $lim , c^{1/n}=1$ So is my proof correct? And what concerning my segond question? – X0-user-0X Nov 17 '22 at 12:49
  • 1
    You claim that $(b_2b_3...b_n)^{\frac{1}{n}}$ converges to 1. This is not clear to me at all, and if it's true, it needs a proof. So I would not consider your proof correct/complete.

    I think you are overcomplicating it, try out what @MartinR wrote, you will need to make lots of similar arguments to his in doing analysis anyway, so it's good to get practicing!

    – Richard Jensen Nov 17 '22 at 12:58
  • 1
    @RichardJensen I understand your point but i ve learned in class this. https://math.stackexchange.com/questions/1220020/limit-of-a-geometric-mean So i just use it without the need to prove it. I don't see any simpler/shorter proof. If you have one or any other improvement please writte it as an answer so i can mark this topic solved. :-) – X0-user-0X Nov 17 '22 at 13:01
  • 1
    I see. Well, then your proof is correct. I will write up a short proof bases on the other idea mentioned, then you can judge which you find simpler. – Richard Jensen Nov 17 '22 at 13:05
  • Thk a lot. Now i see what @MartinR means by using the sandwitch theorem. You can writte the answer so i ll writte this topic as solved thk – X0-user-0X Nov 17 '22 at 13:07

1 Answers1

2

Your proof is correct but uses a (relatively) sophisticated theorem to prove a basic result. Here is a simpler proof (alternatively, follow Martin R's advice).

If $a_n\to L>0$ then $(\ln a_n)$ is convergent hence bounded.

Therefore, $c_n:=\frac{\ln a_n}n\to0$ and $\sqrt[n]{a_n}=e^{c_n}\to e^0=1.$

Anne Bauval
  • 34,650
  • Thk for your answer so does my segond proof here is too correct? https://math.stackexchange.com/questions/4578556/proof-let-a-n-be-a-positive-sequence-and-0q1-a-constant-number-s-t-for (using sub sequence)? – X0-user-0X Nov 17 '22 at 13:11
  • No, I answered there. – Anne Bauval Nov 17 '22 at 13:20