An article I'm reading states a bound on some value, I believe they are using Stirling's approximation but I have been unable to derive their result, perhaps someone here can help.
The claim:
For $\alpha,\epsilon >0$ we have $$ \frac{\alpha^n}{n!}<\epsilon $$ when $$ n>e\alpha + \log(1/\epsilon). $$
My attempt so far:
By Stirling's approximation we have $$ \frac{\alpha^n}{n!}\sim\frac{e^n\alpha^n}{\sqrt{2\pi n}\;n^n}<\frac{e^n\alpha^n}{n^n}, $$ inserting into the first inequality and taking the log of both sides we have $$ n\log(e\alpha/n) < \log\epsilon $$ then, taking the inverse of the log arguments and rearranging we have $$ n\log n > n \log e\alpha + \log(1/\epsilon). $$ I am unsure how to proceed, is there another approximation I should be using?