Proving Both Upper and Lower Bounds for Stirling's Approximation

  • Context: Comp Sci 
  • Thread starter Thread starter rxh140630
  • Start date Start date
Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
1 reply · 2K views
rxh140630
Messages
60
Reaction score
11
Homework Statement
prove lg(n!) = thetha(nlgn) where lg n is log with base = 2. Use Sterling's approximation as a hint
Relevant Equations
Sterling's approximation: [itex]n! = \sqrt(2*pi*n)(\frac{n}{e})^n(1+ \theta{1/n}) [/itex]
Sterling's approximation: [itex]n! = \sqrt{2*pi*n}(\frac{n}{e})^n(1+ \theta(1/n))[/itex]

So I need to prove

[itex]c_1nlgn ≤lg(\sqrt{2*pi*n}) + lg((\frac{n}{e})^n) + lg(1+ \theta(1/n))) ≤ c_2nlgn[/itex]

My question is:

assume I've proven [itex]lg(\sqrt{2*pi*n})[/itex] as [itex]\theta(lgn)[/itex]

Do I need to now prove that [itex]c_1nlgn ≤ lgn ≤ c_2nlgn[/itex] ??

What if we assume I found that [itex]lg(\sqrt{2*pi*n})[/itex] as [itex]\theta(\sqrt{n})[/itex]

Do I need to now prove that [itex]c_1nlgn ≤ \sqrt{n} ≤ c_2nlgn[/itex] ??

Assuming the other terms in the sum [itex]g(\sqrt{2*pi*n}) + lg((\frac{n}{e})^n) + lg(1+ \theta(1/n)))[/itex] are proven to be [itex]\theta(nlgn)[/itex]
 
Physics news on Phys.org
The upper and lower bounds of Stirling's formula are so good, that I would use both of them just to be on the safe side.
$$
1< e^{1/(12n+1)} < \dfrac{n!}{\sqrt{2\pi n}\cdot\left(\dfrac{n}{e}\right)^n} < e^{(1/12n)}< 1+\dfrac{1}{11n}
$$

The formal definition of the theta notation is:
$$
f(x)=\theta(g(x)) \Longleftrightarrow 0<\operatorname{lim\,inf}_{x\to a}\left|\dfrac{f(x)}{g(x)}\right|\leq \operatorname{lim\,sup}_{x\to a}\left|\dfrac{f(x)}{g(x)}\right|<\infty
$$
so, yes, you have to manage both ends.