0
$\begingroup$

I need help proving this limit:

$$ \lim_{x \to \infty}\left[% {2{\rm Li}_{2}\left(1-{\rm e}^{x}\right) \over x} -{x\,{\rm e}^{x} \over 1 - {\rm e}^{x}}\right] =0 $$

Where the ${\rm Li}_{2}$ is polylogarithm function of order $2$.

The difficulty is because each component go to negative infinity, so we have to evaluate the limit of the whole expression. $$ \mbox{I was able to show that}\quad\frac{\displaystyle{\frac{x\,{\rm e}^{x}}{1 - {\rm e}^{x}}}} {\displaystyle{\frac{2\,{\rm Li}_{2}\left(1 - {\rm e}^{x}\right)}{x}}}=1 $$ but I have a hunch that the implication do not go that way. Please help quickly. My homework due soon. Thank you.

$\endgroup$
1
$\begingroup$

Using the identity \begin{align} Li_{2}(1-t) + Li_{2}\left(1 - \frac{1}{t}\right) = - \frac{1}{2} \ln^{2}(t) \end{align} with $t= e^{x}$ it is sen that \begin{align} Li_{2}(1-e^{x}) = - \frac{x^{2}}{2} - Li_{2}(1-e^{-x}). \end{align} Now the limit is \begin{align} \lim_{n \rightarrow \infty} \left\{ \frac{2}{x} \ Li_{2}(1-e^{x}) - \frac{x}{e^{-x}-1} \right\} &= \lim_{n \rightarrow \infty} \left\{ - \frac{2}{x} \ Li_{2}(1-e^{-x}) - x - \frac{x}{e^{-x} -1} \right\} \\ &= \lim_{n \rightarrow \infty} \left\{ - \frac{2}{x} \ Li_{2}(1-e^{-x}) - \frac{x}{1-e^{x}} \right\} \\ &= \lim_{n \rightarrow \infty} \left\{ - \frac{2}{x} \ Li_{2}(1-e^{-x}) - \frac{x}{1-(1 + x + \frac{x^{2}}{2} + \mathcal{O}(x^{3}))} \right\} \\ &= \lim_{n \rightarrow \infty} \left\{ - \frac{2}{x} \ Li_{2}(1-e^{-x}) + \frac{1}{\frac{x}{2} + \mathcal{O}(x^{3})} \right\} \\ &= 0. \end{align}

$\endgroup$
4
  • $\begingroup$ Thanks. Funny, I actually use the identity to "simplify" the original equation into the form I posted. Guess that made thing worse. $\endgroup$ – Panicnow May 27 '14 at 3:58
  • $\begingroup$ Can I ask how you're taking the limit with regards to the polylogarithm? I can't find a reference anywhere that it is continuous and don't have any specific information on how it behaves under limit (other than what mathematica is providing). Thanks $\endgroup$ – Michael Dec 5 '14 at 15:52
  • $\begingroup$ The dilogarithm is the series $\begin{align} Li_{2}(x) = \sum_{n=1}^{\infty} \frac{x^{n}}{n^{2}} \end{align}$ for which properties can be obtained. In the above case the limit is for $x \rightarrow infty$ the term $\frac{1}{x}$ dominates and makes the term zero. $\endgroup$ – Leucippus Dec 15 '14 at 7:10
  • $\begingroup$ Instead of $n$ is should be the limit as $x \rightarrow \infty$ $\endgroup$ – Leucippus Dec 15 '14 at 7:11

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.