2

The problem I'm working on asks for the radius of convergence for the Maclaurin series of $\frac{1}{x^2 + x - 6}$

Here is my attempt at a solution. I would highly appreciate anyone who spots the error(s).

By partial fraction decomposition we have

$$\frac{1}{x^2 + x - 6} = \frac{\frac{1}{5}}{x - 2} - \frac{\frac{1}{5}}{x + 3}$$

$$ = \biggl(\frac{1}{5}\biggl)\frac{1}{(-2)(1 -\frac{x}{2})} - \biggl(\frac{1}{5}\biggl)\frac{1}{(-3)(1 -\frac{x}{3})} $$

$$ = \biggl(\frac{1}{5}\biggl) \biggl[ \biggl(-\frac{1}{2}\biggl) \frac{1}{(1 -\frac{x}{2})} + \biggl(\frac{1}{3}\biggl)\frac{1}{(1 -\frac{x}{3})} \biggl]$$

Converting to power series and absorbing the fractions into the respective summations:

$$ = \biggl(\frac{1}{5}\biggl)\biggl( \sum_{n=0} ^{\infty} \frac{x^n}{3^{n + 1}} -\sum_{n=0} ^{\infty} \frac{x^n}{2^{n + 1}} \biggl)$$

Combining summations and computing a common denominator we have

$$ = \biggl(\frac{1}{5}\biggl)\biggl( \sum_{n=0} ^{\infty} \frac{x^n(2^{n + 1} - 3^{n + 1})}{3^{n + 1}2^{n + 1}} \biggl)$$

We now apply the ratio test to the general term of our power series:

$$ \lim_{n\to\infty} \biggl|\frac{x^{n + 1}(2^{n + 2} - 3^{n + 2})}{3^{n + 2}2^{n + 2}} \cdot \frac{3^{n + 1}2^{n + 1}}{x^n(2^{n + 1} - 3^{n + 1})} \biggl| $$

$$ = \lim_{n\to\infty} \biggl|\frac{x^{n + 1}(2^{n + 2} - 3^{n + 2})}{3^{n + 2}2^{n + 2}} \cdot \frac{3^{n + 1}2^{n + 1}}{x^n(2^{n + 1} - 3^{n + 1})} \biggl| $$

$$ = \frac{x}{6} \biggl( \lim_{n\to\infty} \biggl|\frac{2^{n + 2}}{2^{n + 1} - 3^{n + 1}} \biggl| - \lim_{n\to\infty} \biggl|\frac{3^{n + 2}}{2^{n + 1} - 3^{n + 1}}\biggl| \biggl)$$

Throwing the $\frac{1}{5}$ back in we get:

$$ \biggl(\frac{1}{5} \biggl) \biggl(\frac{x}{6}\biggl) (0 - 3)$$

$$ \Longrightarrow $$

$$ \biggl|- \frac{x}{90}\biggl| < 1 $$

So $ x < 90 $.

But the radius of convergence is supposed to be $ 2 $ so this is clearly wrong.

(As a humorous aside, ChatGPT was unable to go through these steps without making numerous errors, often arriving at a completely different answer when asked if it might have made a mistake).

Anne Bauval
  • 34,650
RTF
  • 424

2 Answers2

3

Indeed, since $\displaystyle \sum_{0}^{+\infty}x^k=\frac{1}{1-x}$ with convergence for $|x|<1$ so one can write

\begin{align*} \frac{1}{x^2+x-6}&=\frac{1/5}{x-2}+\frac{-1/5}{x+3}\\ &=\frac{1}{5}\sum_{0}^{+\infty}(-\frac{1}{2})(\frac{x}{2})^k-\frac{1}{5}\sum_{0}^{+\infty}\frac{(-1)^k}{3} (\frac{x}{3})^k\\ \end{align*}

The first series converges for $|x|<2$ and the second series converges to $|x|<3$ so the sums of the series should be converge for $|x|<2$ (because between $2$ and $3$ diverges). One can see it as follows

Adding series over its intervals of convergence one can write

\begin{align*} \frac{1}{5}\sum_{0}^{+\infty}(\frac{-1}{2^{k+1}}+\frac{(-1)^{k+1}}{3^{k+1}})x^k&=\frac{1}{5}\sum_{0}^{+\infty}(-(\frac{1}{2})^{k+1}+(-\frac{1}{3})^{k+1})x^k\\ &:=\frac{1}{5}\sum_{0}^{+\infty}c_kx^k \end{align*}

Since one has that
\begin{align*} |\frac{c_{k}}{c_{k+1}}|&=|\frac{-(\frac{1}{2})^{k+1}+(-\frac{1}{3})^{k+1}}{-(\frac{1}{2})^{k+2}+(-\frac{1}{3})^{k+2}}|\\ &\underset{k\to+\infty}{\longrightarrow \frac{1}{1/2}}=2 \end{align*}

Since the limit there exists and it is $2$ so the radius of convergence is the limit, that is, $2$.

NB: The Maclaurin series is Taylor series centered at $a=0$. The map $x\mapsto \frac{1}{x^2+x-6}$ has singularity when $x^2+x-6=0$ so $x_1=-3$ and $x_2=2$. Since

The radius of convergence of a power series $f$ centered on a point $a$ is equal to the distance from $a$ to the nearest point where $f$ cannot be defined in a way that makes it holomorphic. (cf.https://en.m.wikipedia.org/wiki/Radius_of_convergence)

Hence the radius of convergence is $R=2$, again.

A. P.
  • 5,978
  • +1 but the question was only "what is the radius" (which your last paragraph readily gives) and "where is my error(s)" (which you pointed in the comment section), so the expansion was useless (there are tons of very similar calculations on this site and the OP showed being able to do them, apart from that mistake). – Anne Bauval May 15 '23 at 07:55
  • 1
    I prefer to write answers that help the OP and show a bit more information that I consider relevant. It is true that I may have repeated a bit of what the OP wrote (as I suggested in the comments), but it was not a waste of time at all. This is a matter of style. If my answer gets upvotes it's OK, if not it's OK too. As long as it's a correct answer it's good enough for me. – A. P. May 15 '23 at 11:52
  • Allright. I also prefer that. My reaction was only due to having been traumatized by repeated severe criticisms from some moderator(s) who try to prevent such "duplicating". – Anne Bauval May 15 '23 at 12:02
2

Combining summations and computing a common denominator we have

$$ \biggl( \sum_{n=0} ^{\infty} \frac{1}{5}\frac{x^n(2^{n + 1} - 3^{n + 1})}{3^{n + 1}2^{n + 1}} \biggl)$$

We now apply the ratio test to the general term of our power series:

$\begin{align} \lim_{n\to\infty} \biggl|\frac{1}{5}\frac{x^{n + 1}(2^{n + 2} - 3^{n + 2})}{3^{n + 2}2^{n + 2}} \cdot \frac{5}{1}\frac{3^{n + 1}2^{n + 1}}{x^n(2^{n + 1} - 3^{n + 1})} \biggl|&=\lim_{n\to\infty} \biggl|\frac{x(2^{n + 2} - 3^{n + 2})}{6(2^{n + 1} - 3^{n + 1})} \biggl|\\ &=\lim_{n\to\infty} \biggl|\frac{x(\frac{2^{n + 2}}{3^{n + 2}}-1)3^{n+2}}{6(\frac{2^{n + 2}}{3^{n + 2}}-1)3^{n+1}} \biggl| \end{align}$

and we get $$ \frac{|x|}{2}\lim_{n\to\infty} \biggl|\frac{\frac{2^{n + 2}}{3^{n + 2}}-1}{\frac{2^{n + 1}}{3^{n + 1}}-1} \biggl|=\frac{|x|}{2}<1. $$ Also, you didn't need to combine the series: The radius of convergences of the series $\sum_{n=0} ^{\infty} \frac{x^n}{2^{n + 1}}$, $\sum_{n=0} ^{\infty} \frac{x^n}{3^{n + 1}}$ are $2$ and $3$ respectively and we have to take the minumum to find the radius of the convergence of their difference.

Bob Dobbs
  • 10,988
  • Thanks, although I'm having difficulty connecting the dots between these two steps. Canceling terms I get $\lim_{n\to\infty} \biggl|\frac{x^{n}(2^{n + 2} - 3^{n + 2})}{3 \cdot 2} \cdot \frac{1}{(2^{n + 1} - 3^{n + 1})} \biggl|$ Could you spell out some of the details? Also, what happened to the original $ \frac{1}{5} $ fraction? – RTF May 14 '23 at 02:57
  • @RTF That $1/5$ will not affect the radius of convergence in any way. I am sure you can convince yourself about that. – Gary May 14 '23 at 06:29