10
$\begingroup$

Is something wrong with the calculation below?

$$ \begin{align} \lim_{x \to \infty} \frac{4x^2}{x-2} &= \lim_{x \to \infty} \frac{4x}{1-2/x} \\ & = \frac{(\lim_{x \to \infty} 4x)}{(\lim_{x \to \infty} 1-2/x)} \\ & = \frac{(\lim_{x \to \infty} 4x)}{1} \\ & = \lim_{x \to \infty} 4x \\ & = \infty. \end{align} $$

I ask because if there isn't then the following would seem correct,

$$ \begin{align} \lim_{x \to \infty} \frac{4x^2}{x-2} - 4x &= \left( \lim_{x \to \infty} \frac{4x^2}{x-2} \right) - \left( \lim_{x \to \infty} 4x \right) \\ & = \left( \lim_{x \to \infty} 4x \right) - \left( \lim_{x \to \infty} 4x \right) \\ & = \lim_{x \to \infty} 4x - 4x \\ & = \lim_{x \to \infty} 0 \\ & = 0. \end{align} $$

But it is not, since

$$ \begin{align} \lim_{x \to \infty} \frac{4x^2}{x-2} - 4x &= \lim_{x \to \infty} \frac{4x^2 - 4x(x-2)}{x-2} \\ &= \lim_{x \to \infty} \frac{8x}{x-2} \\ &= \lim_{x \to \infty} \frac{8}{1-2/x} \\ &= 8. \\ \end{align} $$

In what way is this wrong? Where is the mistake?

$\endgroup$
  • 5
    $\begingroup$ I think you have pretty well pinned down the mistake by your careful exposition. After all, $x$ and $x+8$ both tend to infinity, but that doesn't mean that you can find the limit of their difference by taking the difference of the two (infinite) limits. $\endgroup$ – hardmath Nov 25 '16 at 1:20
  • $\begingroup$ But getting to the limit of their difference is exactly how I would find the difference of their limits. I was thinking despite the first calculation being correct, the mistake is substituting something for infinity in the second calculation. $\endgroup$ – henriq Nov 25 '16 at 1:31
  • 1
    $\begingroup$ The limit of the difference $(x+8)-x$ is obvious (since the expression simplifies to a constant). But it doesn't make sense to subtract $\infty - \infty$, the difference of the limits. Something that makes sense is not best found by asking for something that does not make sense. $\endgroup$ – hardmath Nov 25 '16 at 1:34
  • $\begingroup$ That makes sense. So can I or can't I use the rules for calculating with limits when one of the terms tends to infinity? And if I can, what is the mistake after all? $\endgroup$ – henriq Nov 25 '16 at 1:44
  • 1
    $\begingroup$ @henriq.cd You can do operations with limits unless you have an indeterminate form. In your first calculation you have $\infty/1$ which is not indeterminate ($\infty/1=\infty$). Look at Jack's wikipedia link for the rules and also you may look here for a list of indeterminate forms $\endgroup$ – Momo Nov 25 '16 at 2:13
11
$\begingroup$

It is incorrect that

$$ \lim_{x \to \infty} \frac{4x^2}{x-2} - 4x = \left( \lim_{x \to \infty} \frac{4x^2}{x-2} \right) - \left( \lim_{x \to \infty} 4x \right) $$

since in order to apply the limit rule $$ \lim_{x\to\infty}\big(f(x)-g(x)\big)=\lim_{x\to\infty}f(x)-\lim_{x\to\infty}g(x) $$ one needs both at least one of $\lim_{x\to\infty}f(x)$ and $\lim_{x\to\infty}g(x)$ being real numbers. ([Added later:] In the case $\infty-(-\infty)$, one needs to define the arithmatic operation for the extended real numbers first.)


[Added:] It is dangerous to view $\infty$ as a number unless you know exactly what extended real numbers are and what arithmetic operations are (and are not) allowed for them. There are more examples for nonsense by considering $\infty$ as a real number: $$ \lim_{x\to\infty}\left(x\cdot\frac{1}{x}\right)=\infty\cdot 0=0; $$ $$ \lim_{x\to\infty}\big(2x-x\big)=\infty-\infty=0 $$


[Added later:] Also, the step $$ \lim_{x \to \infty} \frac{4x}{1-2/x} = \frac{(\lim_{x \to \infty} 4x)}{(\lim_{x \to \infty} 1-2/x)} $$ would not be mathematically correct unless one is working in the extended real numbers and has defined what is $\dfrac{\infty}{a}$ for a non-zero real number $a$.

$\endgroup$
  • $\begingroup$ Splitting works even if the limits are infinite, unless you have an indeterminate form. In this case there is an undeterminate form. $\endgroup$ – Momo Nov 25 '16 at 1:18
  • 2
    $\begingroup$ Much better after correction. Although "you need at least one of $\lim_{x\to\infty}f(x)$ and $\lim_{x\to\infty}g(x)$ being real numbers." is still not correct. For example $\infty+\infty$ is not undeterminate. $\endgroup$ – Momo Nov 25 '16 at 2:03
4
$\begingroup$

$\infty-\infty$ is an indeterminate form, so the result of it can be anything.

Your mistake is exactly here:

$$\left( \lim_{x \to \infty} 4x \right) - \left( \lim_{x \to \infty} 4x \right)= \lim_{x \to \infty} (4x - 4x)$$

On the left side you have $\infty-\infty$, which has to be solved on a case by case basis by returning to the original limit. So splitting in this case does not work.

$\endgroup$
  • 6
    $\begingroup$ No - their mistake is before that, when they split the limit term by term without making sure that the limits of the terms exist. $\endgroup$ – user361424 Nov 25 '16 at 1:15
  • $\begingroup$ It works even if the limits are infinite, unless you have an undeterminate form. And this case has an undeterminate form, as I mentioned. $\endgroup$ – Momo Nov 25 '16 at 1:17
3
$\begingroup$

You can't split a limit term-by-term like that unless both terms converge. In this case, neither does, so you can't.

$\endgroup$

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.