I want to find the following limit:

$$\lim_{x\rightarrow -\infty}{e^{\frac {1}{2-x}}\cdot\frac{x^2+2x-1}{x-2}}-x$$

This is what I do. I change the variable $t=-x$ and I have the following limit:

$$\lim_{t\rightarrow +\infty}{e^{\frac {1}{2+t}}\cdot{\frac{t^2-2t-1}{-t-2}}+t}=\lim_{t\rightarrow+\infty}{h(x)}$$

We have $e^{\frac{1}{2+t}}\rightarrow1$ for $t\rightarrow+\infty$

Therefore I think (this is the passage I'm less sure about)

$$h(x)\sim \frac{t^2-2t-1}{-t-2}+t=\frac{t^2-2t-1+t(-t-2)}{-t-2}=\frac{t^2-2t-1-t^2-2t}{-t-2}=\frac{-4t-1}{-t-2}\sim{\frac {-4t}{-t}}\rightarrow4$$

The solution should actually be $3$. Any hints on what I'm doing wrong?

  • $\begingroup$ You went wrong taking $e^{\frac{1}{2+t}}\rightarrow1$, we need to expand to one term more. $\endgroup$ – user Sep 3 '18 at 10:23
  • $\begingroup$ In general one can't replace a small part of a large expression by its limit while evaluating the limit of the large expression. Such replacements are allowed in specific cases and you may have a look at this answer to get more details on the specific cases. $\endgroup$ – Paramanand Singh Sep 3 '18 at 15:11

ALternatively: $$\lim_{x\rightarrow -\infty}\left[{e^{\frac {1}{2-x}}\cdot\frac{x^2+2x-1}{x-2}}-x\right]=\\ \lim_{x\to-\infty} \left[e^{\frac{1}{2-x}} \left(x+4+\frac7{x-2}\right)-x\right]=\\ \lim_{x\to-\infty} \left[e^{\frac{1}{2-x}}\left(x+4\right)-x\right] +\underbrace{\lim_{x\to-\infty} e^{\frac{1}{2-x}} \cdot \frac7{x-2}}_{=0}=\\ \lim_{x\to-\infty} \left[e^{\frac{1}{2-x}}\left(x+4\right)-(x+4)+4\right]=\\ \lim_{x\to-\infty} \left[\frac{e^{\frac1{2-x}}-1}{\frac1{2-x}}\cdot \frac{x+4}{2-x}+4\right]=\\ 1\cdot (-1)+4=3,$$ where it was used: $$\lim_{x\to 0} \frac{e^x-1}{x}=1.$$

  • $\begingroup$ Thanks, this looks clean. If I may ask how do you get $\frac{x^2+2x-1}{x-2}=(x+4+\frac{7}{x-2})$. When I try to find the solutions for $x^2+2x-1$ I get roots $\endgroup$ – Cesare Sep 3 '18 at 11:13
  • $\begingroup$ you can rearrange: $\frac{x^2-4+2x-4+7}{x-2}$ $\endgroup$ – farruhota Sep 3 '18 at 11:14
  • $\begingroup$ Somebody like me will definitely need a certain "eye" to do some of the things you did. There's certainly some good level of experience involved here. But I think this is by far the cleanest and most straightforward solution. Thanks $\endgroup$ – Cesare Sep 3 '18 at 11:30
  • $\begingroup$ You are welcome. There are several overlapping points with other solutions, so the credit goes to all. Good luck. $\endgroup$ – farruhota Sep 3 '18 at 11:36

I think the first substitution isn't too useful, but let's keep it nontheless. $$e^{1/(2+x)}\frac{-x^2+2x+1}{x+2}+x=(4-x)e^{1/(2+x)}-7\frac{e^{1/(2+x)}}{x+2}+x=\\=4e^{1/(x+2)}-\frac7{x+2}e^{1/(2+x)}+xe^{1/(2+x)}\left(e^{-1/(2+x)}-1\right)=\\=4e^{1/(x+2)}-\frac7{x+2}e^{1/(2+x)}-\frac x{x+2}e^{1/(2+x)}\frac{e^{-1/(2+x)}-1}{-\frac1{2+x}}$$

Now, since $-\frac1{x+2}\to 0^-$ as $x+2\to\infty$ and $\lim_{s\to 0}\frac{e^s-1}s=1$, $$\lim_{x\to\infty}4e^{1/(x+2)}-\frac7{x+2}e^{1/(2+x)}-\frac x{x+2}e^{1/(2+x)}\frac{e^{-1/(2+x)}-1}{-\frac1{2+x}}=\\=\left[4e^0-\frac7\infty e^0-1\cdot e^0\cdot 1\right]=3$$

Now that you should know one more idea than before, you can see for yourself how your initial guess fares against $$\lim_{x\to\infty} xe^{1/(\alpha +\beta x)}-x,\qquad \beta\ne 0$$ as opposed to the correct answer.

  • $\begingroup$ Thanks Saucy, honestly I don't understand most of this. I'm sure this is correct +1 $\endgroup$ – Cesare Sep 3 '18 at 10:45
  • $\begingroup$ You shouldn't trust it, because the first version had a mistake of algebra. Fortunately, in a term that then decays to $0$. $\endgroup$ – Saucy O'Path Sep 3 '18 at 10:45
  • $\begingroup$ Are you using Taylor here? $\endgroup$ – Cesare Sep 3 '18 at 10:46
  • $\begingroup$ Not really. Just multiplication-and-division, a couple of fundamental limits and an eye for substituitions. $\endgroup$ – Saucy O'Path Sep 3 '18 at 10:47

Although $\lim\limits_{x\to-\infty}e^{\frac1{2-x}}=1$, we cannot simply replace it by $1$ in the limit because it is actually $1+\frac1{2-x}+O\!\left(\frac1{(2-x)^2}\right)$

For $x$ near $\pm\infty$, $$ \begin{align} e^{\frac1{2-x}}\frac{x^2+2x-1}{x-2}-x &=\overbrace{\left(1+\frac1{2-x}+O\!\left(\frac1{(2-x)^2}\right)\right)}^{e^{\frac1{2-x}}}\overbrace{\left(x+4-\frac7{2-x}\right)\vphantom{\left(\frac1{()^2}\right)}}^{\frac{x^2+2x-1}{x-2}}-x\\ &=\frac x{2-x}+4+O\!\left(\frac1{2-x}\right) \end{align} $$ Now let $x\to-\infty$.

Simply replacing $e^{\frac1{2-x}}$ by $1$ misses the $\frac x{2-x}$ in the formula above and leaves us only with $4$ in the limit.


Let set $\frac {1}{2-x}=y \to 0$

$$\lim_{x\rightarrow -\infty}{e^{\frac {1}{2-x}}\cdot\frac{x^2+2x-1}{x-2}}-x=\lim_{y\rightarrow 0}{e^{y}\cdot\frac{\frac1{y^2}-\frac 6 y+7}{-\frac1y}}-2+\frac1y$$$$=\lim_{y\rightarrow 0}e^{y}\cdot\left(-\frac1{y}+6-7y\right)-2+\frac1y$$

then use $e^y=1+y+o(y)$ to obtain


$$-\frac1{y}+6-7y-1+6y-7y^2+o(1)-2+\frac1y=3+o(1) \to 3$$

  • $\begingroup$ I don't understand how to replace $x^2$ with $\frac{1}{2-x}$. I tried to get $x$ out of $\frac{1}{2-x}$.... $\endgroup$ – Cesare Sep 3 '18 at 10:15
  • $\begingroup$ Do I need to find the inverse? Replacing $x$ with something that contains the variable $y$ isn't easy for me $\endgroup$ – Cesare Sep 3 '18 at 10:18
  • $\begingroup$ @Cesare I add some detail more. $\endgroup$ – user Sep 3 '18 at 10:19
  • $\begingroup$ Thanks, are you using Taylor? $\endgroup$ – Cesare Sep 3 '18 at 10:22
  • $\begingroup$ @Cesare Yes of course, but note that it always necessary tak einto account the remainder terms vie little-o or big-O notation otherwise we can easily get wrong. $\endgroup$ – user Sep 3 '18 at 10:24

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.