0

So looking back at this problem: $\lim_{n\rightarrow \infty } (\frac{(n+1)^{2n^2+2n+1}}{(n+2)^{(n+1)^2}\cdot n^{n^2}})$ , I did a while ago, I thought of something rather interesting. The author of that problem originally got it from using the ratio test on a series. Now, I used the root test on that series, and got 1/e as my answer. Of course, if one evaluates the above limit by brute force, they will also yield 1/e as an answer. So that makes me wonder, why is this so?

tl;dr - For an absolutely convergent series, why is it that the root test and the ratio test give the same exact limit as n -> infinity? As a follow up question, is there ever an instance (for an absolutely convergent series), where the root and ratio test wouldn't result in the same limit?

  • 3
    Relevant: https://www.maa.org/sites/default/files/0025570x33450.di021200.02p0190s.pdf (middle of first page) – Clement C. May 06 '17 at 02:15

1 Answers1

1

The improved ratio test says that given a series $\sum_{n=1}^{\infty}x_{n}$ with $x_{n}\geq0$ for all $n\in\mathbb{N}$, if$$ \limsup_{n\rightarrow\infty}\sqrt[n]{x_{n}}<1, $$ then the series converges, while if $$ \limsup_{n\rightarrow\infty}\sqrt[n]{x_{n}}>1, $$ then the series diverges to $\infty$. On the other hand, the ratio test says that given a series $\sum_{n=1}^{\infty}x_{n}$ with $x_{n}>0$ for all $n\in\mathbb{N}$, if$$ \limsup_{n\rightarrow\infty}\frac{x_{n+1}}{x_{n}}<1, $$ then the series converges. If $$ \liminf_{n\rightarrow\infty}\frac{x_{n+1}}{x_{n}}>1, $$ then the series diverges to $\infty$. Now given a sequence $\left\{ x_{n}\right\} $ of real numbers, with $x_{n}>0$ for all $n\in\mathbb{N}$. Then (see inequalities $$ \liminf\limits_{n\rightarrow\infty}\frac{x_{n+1}}{x_{n}}\leq\liminf \limits_{n\rightarrow\infty}\sqrt[n]{x_{n}}\leq\limsup\limits_{n\rightarrow \infty}\sqrt[n]{x_{n}}\leq\limsup\limits_{n\rightarrow\infty}\frac{x_{n+1} }{x_{n}}. $$ All the inequalities can be strict. It follows that if $\lim\limits_{n\rightarrow\infty}\frac{x_{n+1}}{x_{n}}=\ell$ exists then there exists $\lim\limits_{n\rightarrow \infty}\sqrt[n]{x_{n}}=\ell,$ so the two tests give the same result. However in general the the ratio test is worse than the root test. Consider the series $\sum_{n=1}^{\infty}x_{n}$, where $$ x_{n}=\left\{ \begin{array} [c]{ll}% \frac{1}{2^{n}} & \text{if }n\text{ is even,}\\ \frac{1}{3^{n}} & \text{if }n\text{ is odd.}% \end{array} \right. $$ Then $$ \liminf\limits_{n\rightarrow\infty}\frac{x_{n+1}}{x_{n}}=0,\quad \limsup\limits_{n\rightarrow\infty}\frac{x_{n+1}}{x_{n}}=\infty, $$ and so the ratio test gives NO information. However,$$ \limsup\limits_{n\rightarrow\infty}\sqrt[n]{x_{n}}=\frac{1}{2}, $$ and so the series converges by the root test.

Gio67
  • 20,905
  • (+1) Well constructed tutorial. You might consider adding a reference such as This One that proves that $$ \liminf \frac{a_{n+1}}{a_n} \le \liminf (a_n)^ {1/n}\le\limsup (a_n)^ {1/n}\le \limsup \frac{a_{n+1}}{a_n}$$ – Mark Viola May 06 '17 at 03:25
  • thanks. I just added the link – Gio67 May 06 '17 at 10:40