5
$\begingroup$

Prove that $$\prod_{n=1}^{\infty}\left(\frac{2n+3}{2n+2}\right)$$ converges or diverges.

Taking analysis this year has really intrigued me. We do not cover this topic in analysis, but I did some research on my own after learning about conditionally convergent sequences. I have become interested in learning about infinite products. From my research, it seems that a criteria of a convergent infinite product is that the sequence converges to 1. This sequence does converge to 1 so it meets this criteria. I believe I need to show that

$$\sum_{n=1}^{\infty} \ln\left(\frac{2n+3}{2n+2}\right)$$

converges. I tried to prove this using the ratio test, but I got nowhere with that. Any suggestions?

$\endgroup$
5
  • 1
    $\begingroup$ Do you know what the linear approximation is for $\ln$ at $1$? I.e., $\ln(1+x) \approx \ $? $\endgroup$ Dec 5, 2016 at 1:05
  • 1
    $\begingroup$ Possible duplicate of How to prove that $\lim_{n \to\infty} \frac{(2n-1)!!}{(2n)!!}=0$ $\endgroup$ Dec 5, 2016 at 1:07
  • 1
    $\begingroup$ @ArnaldoNascimento That doesn't work. $\endgroup$ Dec 5, 2016 at 1:08
  • $\begingroup$ Note that there aren't very many infinite product questions on this site, and so it only takes a few minutes to briefly check if your question is a duplicate. $\endgroup$ Dec 5, 2016 at 1:10
  • $\begingroup$ All the terms are greater than 1. Product of such infinite numbers greater than 1 has to be infinite. $\endgroup$
    – jnyan
    Dec 5, 2016 at 16:37

6 Answers 6

3
$\begingroup$

$\newcommand{\bbx}[1]{\,\bbox[15px,border:1px groove navy]{\displaystyle{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$ \begin{align} \prod_{n = 1}^{\infty}\pars{2n + 3 \over 2n + 2} &= \lim_{N \to \infty}\,\,\prod_{n = 1}^{N}\pars{n + 3/2 \over n + 1} = \lim_{N \to \infty}\,\,{\pars{5/2}^{\overline{N}} \over \pars{2}^{\overline{N}}} = \lim_{N \to \infty}\,\,{\Gamma\pars{5/2 + N}/\Gamma\pars{5/2} \over \Gamma\pars{2 + N}/\Gamma\pars{2}} \end{align}

As $\ds{N \to \infty}$, the Stirling Asymptotic Formula yields: \begin{align} {\Gamma\pars{5/2 + N} \over \Gamma\pars{2 + N}} &\sim {\root{2\pi}\pars{5/2 + N}^{N + 3}\,\,\expo{-\pars{N + 5/2}} \over \root{2\pi}\pars{2 + N}^{N + 5/2}\,\,\expo{-\pars{N + 2}}} = \root{N}\,{\bracks{1 + \pars{5/2}/N}^{N }\,\,\expo{-1/2} \over \pars{1 + 2/N}^{N}} \\[5mm] & \sim \root{N}\,{\expo{5/2}\expo{-1/2} \over \expo{2}} = \root{N} \end{align}


$$ \mbox{Then,}\quad\bbx{\ds{% \prod_{n = 1}^{N}\pars{n + 3/2 \over n + 1} \sim {4 \over 3\root{\pi}}\,\root{N} \quad\mbox{as}\quad N \to \infty}} $$

$\endgroup$
2
  • $\begingroup$ Seems a bit overkill just to determine convergence :). $\endgroup$
    – Erick Wong
    Dec 5, 2016 at 2:04
  • $\begingroup$ @ErickWong I agree with you. $Stirling$ is the first 'thing' which comes to our mind. Thanks. $\endgroup$ Dec 5, 2016 at 2:43
3
$\begingroup$

A First Approach

Since $(2n+3)^2\gt(2n+3)^2-1=(2n+2)(2n+4)$, we get $$ \frac{2n+3}{2n+2}\ge\frac{2n+4}{2n+3} $$ Therefore, we have $$ \begin{align} \left(\prod_{n=1}^m\frac{2n+3}{2n+2}\right)^2 &\ge\prod_{n=1}^m\frac{2n+3}{2n+2}\frac{2n+4}{2n+3}\\ &=\prod_{n=1}^m\frac{2n+4}{2n+2}\\ &=\frac{2m+4}4 \end{align} $$ Thus, $$ \prod_{n=1}^m\frac{2n+3}{2n+2}\ge\sqrt{\frac{m+2}2} $$ and the infinite product diverges.


A Second Approach

For $a,b\ge0$, we have $(1+a)(1+b)\ge1+a+b$. Therefore, by induction, we get $$ \prod_{n=1}^m(1+a_n)\ge1+\sum_{n=1}^ma_n $$ Thus, $$ \begin{align} \prod_{n=1}^m\left(1+\frac1{2n+2}\right) &\ge1+\sum_{n=1}^m\frac1{2n+2}\\ &=1+\frac12\sum_{n=2}^{m+1}\frac1n \end{align} $$ which diverges by comparison with the Harmonic Series.

$\endgroup$
2
$\begingroup$

The series diverges by the integral test. Consider $f(x) = \log\left(\frac{2x+3}{2x+2}\right) = \log (2x+3) - \log(2x+2)$ defined on $[1, +\infty)$.

Integrating this, by parts, yields the antiderivative$$ F(x)=\left(x+\frac32\right)(\log (2x+3)) - (x+1)\log(2x+2) = \frac 12(\log(2x+3)) +(x+1)(\log(2x+3)) - (x+1)(\log (2x+2)) = (x+1) \log\left(\frac{2x+3}{2x+2}\right) \ + $$

$$\frac 12\log(2x+3) \xrightarrow[x\to+\infty]{} +\infty$$

which implies $\int_1^{+\infty} f(t) \ dt = \lim_{x \to \infty}\left(F(x) - F(1)\right) = + \infty $.

Of course, this implies the product diverges as well.

$\endgroup$
1
  • $\begingroup$ Thank you, this answered my question. $\endgroup$
    – MathGuy
    Dec 5, 2016 at 20:20
1
$\begingroup$

Two useful general results about real sequences:

(1). If $a_n\geq 0$ for all $n$ then $\prod_{n=1}^{\infty}(1+a_n)<\infty \iff \sum_{n=1}^{\infty}a_n<\infty.$

(2). If $0\leq a_n<1$ for all $n$ then $\prod_{n=1}^{\infty}(1-a_n)\ne 0\iff \sum_{n=1}^{\infty}a_n<\infty.$

These are not hard to prove. A useful part of proving (2) is that if the product is not $0$ then $0\leq a_n<1/2$ for all but finitely $n.$

Apply (1) with $a_n=1/(2n+2).\;$We have $\sum_{n=1}^{\infty}a_n=\infty.$ Therefore $\prod_{n=1}^{\infty}(2n+3)/(2n+2)=\prod_{n=1}^{\infty}(1+a_n)=\infty.$

$\endgroup$
0
$\begingroup$

Just as a complement to the answer by Felix (being too long for a comment thereto), consider that we may resort to the definition of the Partial Gamma, as: $$ \begin{gathered} \Gamma (z) = \mathop {\lim }\limits_{n\, \to \,\infty } \left( {\Gamma _{\,n} (z) = n^{\,z} \frac{{\,n!}} {{z^{\,\overline {\,n + 1\,} } }} = n^{\,z} \frac{1} {z}\prod\limits_{1\, \leqslant \,k\, \leqslant \,n} {\frac{k} {{z + k}}} } \right) \hfill \\ \mathop {\lim }\limits_{n\, \to \,\infty } \prod\limits_{1\, \leqslant \,k\, \leqslant \,n} {\frac{k} {{z + k}}} = \mathop {\lim }\limits_{n\, \to \,\infty } \,\frac{{z\,\Gamma (z)}} {{n^{\,z} }} \hfill \\ \end{gathered} $$ wherefrom we get $$ \begin{gathered} \prod\limits_{1\, \leqslant \,n} {\frac{{2n + 3}} {{2n + 2}}} = \mathop {\lim }\limits_{N\, \to \,\infty } \prod\limits_{1\, \leqslant \,n\, \leqslant \,N} {\frac{{\left( {n + 3/2} \right)}} {{\left( {n + 1} \right)}}} = \mathop {\lim }\limits_{N\, \to \,\infty } \frac{{\prod\limits_{1\, \leqslant \,n\, \leqslant \,N} {\frac{n} {{\left( {n + 1} \right)}}} }} {{\prod\limits_{1\, \leqslant \,n\, \leqslant \,N} {\frac{n} {{\left( {n + 3/2} \right)}}} }} = \hfill \\ = \mathop {\lim }\limits_{N\, \to \,\infty } \frac{{1\,\Gamma (1)}} {{N^{\,1} }}\tfrac{{2\,N^{\,3/2} }} {{3\,\Gamma (3/2)}} = \frac{2} {{3\,\Gamma (3/2)}}\mathop {\lim }\limits_{N\, \to \,\infty } \sqrt N \hfill \\ \end{gathered} $$

$\endgroup$
0
$\begingroup$

In fact if $a_n>0,$ then $\prod_{n=1}^\infty (1+a_n)$ converses iff $\sum a_n$ converges. You can obtain this result by applying the logarithm and using $\ln (1+u) \sim u$ as $u\to 0.$

$\endgroup$
1
  • $\begingroup$ And this was downvoted why? $\endgroup$
    – zhw.
    Dec 5, 2016 at 18:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.