$$\lim_{x \to 0} \frac{e^{\sin^2 x} - \cos x}{x^3}$$


This is indeterminate of the form $\frac{0}{0}$. Applying L'Hopital's rule twice results in,

$$\lim_{x \to 0} \frac{e^{\sin^2 x}(\sin^2(2x) + \sin^2(4x)) + \cos x}{6x}.$$

Wolfram Alpha says the two sided limit does not exist.


Can I deduce the result by splitting it up into,

$$\frac{1}{6}\lim_{x \to 0} \left(e^{\sin^2 x}(\sin^2(2x) + \sin^2(4x)) + \cos x \right)\left(\frac{1}{x}\right).$$

And saying that the limit does not exist because $\frac{1}{x}$ does not have a two sided limit?

  • $\begingroup$ I suppose. Maybe specify $x\to0^+$ is $+\infty$, $x\to0^-$ is $-\infty$ hence $f$ does not have a two sided limit at 0. Not mandatory though. $\endgroup$ – user39572 Oct 2 '12 at 12:05
  • $\begingroup$ @JulienGodawatta - Possibly weird qns: I know I can't say the same for $\frac{sin x}{x}$ (which tends to 1 as $x$ tends to 0), but why not? I mean, what is the "algorithm" for working out such things? $\endgroup$ – Legendre Oct 2 '12 at 12:08
  • $\begingroup$ It matters whether it is of the form $1/0$, or $0/0$.. $\endgroup$ – Berci Oct 2 '12 at 12:09
  • $\begingroup$ @Legendre L'Hopital's works well in this case (since it is a 0/0) - reduces it to cos! $\endgroup$ – user39572 Oct 2 '12 at 12:11
  • $\begingroup$ @JulienGodawatta - Yes I know. Just wondering if I can "algorithmize" the thinking this way: 1) if $\frac{0}{0}$ then L'Hopital, 2) if $\frac{a}{0}$, $a$ constant, then split into fractions and check if two sided limits exists for the parts. I suppose I can. Thanks. $\endgroup$ – Legendre Oct 2 '12 at 12:13

$\lim_{x\to 0}\frac{e^x-1}{x}=1\implies e^x\approx 1+x $when $x\to 0$,thus, $e^{\sin^2x}\approx 1+\sin^2x$ as $x\to 0$

Hence, $$\lim_{x \to 0} \frac{e^{\sin^2 x} - \cos x}{x^3}=\lim_{x \to 0} \frac{1+ \sin^2 x - \cos x}{x^3}$$

Now you can easily apply L'Hopital's rule to conclude that limit doesn't exist.

  • $\begingroup$ Nice answer! Is it always "legit" to truncate the Taylor series like this for evaluating limits when $x \to 0$? $\endgroup$ – Legendre Oct 2 '12 at 12:17
  • $\begingroup$ More technically , one should use $e^x=1+x+o(x)$ $\endgroup$ – Aang Oct 2 '12 at 12:21
  • $\begingroup$ Yes, and thanks. +1 $\endgroup$ – Legendre Oct 2 '12 at 12:27

If your calculation is good, then yes. You don't even need to split it into two terms, the fraction already tells you that, when $x\to \pm 0$, the numerator $\to 1$ but the denominator $\to \pm 0$, so it is of the form $\displaystyle\frac1{\pm0}$..


The limit is $\infty$. You can see this in your second limit, as you did, by using L'Hopital.

  • $\begingroup$ Wolfram Alpha says limit does not exist? wolframalpha.com/input/… $\endgroup$ – Legendre Oct 2 '12 at 12:10
  • $\begingroup$ Yes it is right. I evaluate only by the right side. By the left side it is $\infty$, $\endgroup$ – Tomás Oct 2 '12 at 13:40

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.