Can you please explain how $(\log n)! = \theta((\log n)*(\log \log n))$? I myself used the equation $n! = \theta(n^n)$ and replaced $\log n$ for $n$ as below $(\log n)! = \theta((\log(n))^{\log(n)})$. But it is different. Why shouldn't we use this equation?
Asked
Active
Viewed 119 times
-3
-
Here are tips on How to ask a good question, and on How to use MathJax. – Anthony Sep 02 '20 at 13:09
-
Many thanks dear Anthony!!!! – learn9909 Sep 02 '20 at 13:11
1 Answers
0
In addition to following @Anthony's advice, notice the following: $$ f_n = \Theta(g_n) \Rightarrow \\ (1) \lim_n |\frac{f_n}{g_n}| \geq c >0 \\ (2) \lim_n |\frac{f_n}{g_n}| \leq c < \infty $$ Both of these conditions must fulfill, otherwise you have different orders. Also notice, $$ (\log n)! = \prod_{k=2}^n \log k = e^{\sum_{k=2}^{n} \log \log k} \geq e^{\int_{2}^{n}\log \log x dx} $$ For this integral, after a bit of algebra, you get $$ n \log \log n - \int_{2}^{\log n}\frac{e^t dt}{t} $$ The second integral is an Exponential integral, you should find the solution yourself, but, given the upper bound, it should $\Theta(n)$. Can you handle from here?
Alex
- 19,262
-
Excellent explanation, many, many thanks dear Alex!!! But sorry from what you've earned $(logn)!$ is \theta(nloglogn) not (lognloglogn), no? – learn9909 Sep 02 '20 at 13:51
-
You should find the asymptotic bounds on the exponential integral up to some order, then just compare the expression you get to the $\log n \log \log n$ by taking the limit of the ratio, since for $n>N$ for some $N$ both expressions are positive – Alex Sep 02 '20 at 15:07
-
Many, many thanks dear Alex for all your help!!!! Much appreciated!!! I have forgotten integrals I should read them again. – learn9909 Sep 02 '20 at 17:40