Convergence of a Recursively Defined Series: $a_n =5$, $a_{n+1} = 1/(6-a_n)$?

348 Views Asked by At

Determine whether the series converges or diverges. I have no idea how to go about this when it isn't defined in terms of n. Help is appreciated. $$a_n =5, a_{n+1} = 1/(6-a_n)$$

3

There are 3 best solutions below

0
On BEST ANSWER

One can prove by induction that $a_n <1$ for all $n>2$.

Induction basis: $a_3=\frac {1}{5} <1$

Induction step: $$a_{n+1}=\frac {1}{6-a_n}<\frac {1}{6-1}<\frac {1}{5}<1$$ We used the induction hypothesis for the first inequality sign.

One can then prove that the sequence is strictly decreasing: $$a_{n+1}=\frac {1}{6-a_n}<\frac {1}{6a_n-a_n}=\frac {1}{5}a_n<a_n$$ We used our statement we proved before for the first inequality sign.

And since we proved that $a_n<1$ for $n>2$ it follows that $6-a_n >0$ and therefore all the elements of the sequence are positive.

All in all, we have proven that the sequence is strictly decreasing and bounded from below which implies its convergence.

1
On

$\newcommand{\bbx}[1]{\,\bbox[8px,border:1px groove navy]{\displaystyle{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$ Write $\ds{a_{n}}$ as $\ds{x_{n} \over y_{n}}$ with $\ds{x_{n + 1} = y_{n}}$ and $\ds{y_{n + 1} = -x_{n} + 6y_{n}}$. $\ds{x_{n} \over y_{n}}$ satisfies the recurrence and \begin{align} &{x_{n + 1} \choose y_{n + 1}}\ =\ \overbrace{\pars{\begin{array}{rr}0 & 1 \\ -1 & 6\end{array}}} ^{\ds{\mathsf{M}}}\ {x_{n} \choose y_{n}} = \pars{\begin{array}{rr}0 & 1 \\ -1 & 6\end{array}}^{n}{x_{1} \choose y_{1}} \,\,\,\stackrel{\mrm{as}\ n\ \to\ \infty}{\sim}\,\,\, \lambda^{n}\,\mathbf{u}\mathbf{u}^{T} {x_{1} \choose y_{1}} \propto \lambda^{n}\,\mathbf{u} \\[5mm] &\ \mbox{with}\quad \left\{\begin{array}{l} \ds{x_{1} = 5} \\ \ds{y_{1} = 1} \end{array}\right. \end{align} where $\ds{\lambda}$ is the largest, in magnitude, eigenvalue of $\ds{\mathsf{M}}$ and $\ds{\mathbf{u} \propto {3 - 2\root{2} \choose 1}}$ the correspondent eigenvector. Note that $\ds{0 \not= \mathbf{u}^{T} {x_{1} \choose y_{1}} \propto 16 - 10\root{2}}$.

$$ \mbox{So,}\qquad {x_{n + 1} \over y_{n + 1}} \,\,\,\stackrel{\mrm{as}\ n\ \to\ \infty}{\to}\,\,\, {3 - 2\root{2} \over 1} = \bbx{\ds{3 - 2\root{2}}} $$

0
On

If $(a_n)_n$ converges to a limit $L$ then $L=1/(6-L),$ equivalently $L^2-6L+1=0,$ so $L=3\pm 2\sqrt 2.$ Experimentally, the values $a_1=5,$ $a_2=1,$ $a_3=1/5,$ $a_4=5/29$ suggest that $(a_n)_n$ appears to be converging to the smaller value $3-2\sqrt 2.$

So let $A=3-2\sqrt 2$ . Let $a_n=A+d_n.$

We have $1-6A+A^2=0$ and $|A|=A<1.$

Suppose $|d_n|<1$ for some $n$. Then $|6-a_n|=6-(A+d_n)|>4.$ And also $$|d_{n+1}|=|a_{n+1}-A|=\left|\frac {1}{6-a_n}-A\right|=\left|\frac {1}{6-(A+d_n)}-A\right|=$$ $$=\frac {|1-6A+A^2-Ad_n|}{|6-(A+d_n)|}|=\frac {|Ad_n|}{|6-(A+d_n)|}\leq|d_n|/4.$$ Therefore by induction on $m\geq n$ we have $|d_m|<1$ for all $m\geq n .$

But then,also by induction on $m\geq n ,$ we have $|d_{m+1}|\leq |d_m|/4$ for all $m\geq n.$

So $\lim_{m\to \infty}d_m=0$ because $|d_m|\leq 4^{-(m-n)}|d_n|$ for all $m\geq n.$ And $d_m\to 0$ is equivalent to $a_m\to A.$