0

How to find which of $2^{n}$ to $n^{a}$ is an (upper bound, tight bound asymptotically larger, etc..)? I tried to use the formula:

$$ \lim_{n\to\infty} \frac{f(n)}{g(n)} $$

where $f(n) = n^{a}$ and $f(n) = 2^{n}$

But I cannot figure out how to use L'Hospital's rule to solve it. We are guaranteed that $a \in \mathbb{N}$.

D.W.
  • 159,275
  • 20
  • 227
  • 470
Moody
  • 137
  • 3

3 Answers3

3

As a hint, $n^\alpha = 2^{\alpha \log_2 n}$. Now, you can compute the ratio between these functions by subtracting the exponents. There's no need to use l'Hôpital's rule.

templatetypedef
  • 9,102
  • 1
  • 30
  • 60
2

Hint: $\frac{(n+1)^\alpha}{n^\alpha} \to 1$, so for any $C>1$ you can show that $n^\alpha = O(C^n)$. If you take $C < 2$ then you get that $n^\alpha = o(2^n)$.

Yuval Filmus
  • 276,994
  • 27
  • 311
  • 503
1

If you want to use L'Hopital's rule to prove $n^{a}=o(2^n)$ that is fairly easy.

Lets consider the lemma in this question's accepted answer. The problem reduces to calculate the limit:

$$ \lim_{n\to\infty} \frac{n^a}{2^n} $$

This limit is of the type "infinite divided by infinite". So we can try to use L'Hopital's rule.

Let $f^i(n)$ be the ith derivative of $f(n)$ and $g^i(n)$ be the ith derivative of $g(n)$. Consider the following properties of $f(n)=n^a$ and $g(n)=2^n$:

$$f^i(n) = a(a-1)(a-2)...(a-i+2)(a-i+1)n^{a-i} $$ $$g^i(n) = 2^n*log^i(2)$$

As you can see the following LHopital's Rule's conditions comply:

  • $f$ and $g$ are differentiable in the interval $]1,\infty[$.
  • $g$ and $g'$ are nonzero in the interval $]1, \infty[$.
  • As $n \rightarrow \infty$:

$$\frac{f(n)}{g(n)}\rightarrow \frac{\infty}{\infty}$$

This conditions also comply for the derivatives of $f$ and $g$.

Now we use L'Hopital's rule "a" times: We have to derive $f(n)$ and $g(n)$ "a" times and the result is:

$$ \lim_{n\to\infty} \frac{n^a}{2^n} = \lim_{n\to\infty} \frac{a(a-1)(a-2)...(2)(1)*n^0}{2^n*log^a(2)} \rightarrow \frac{a(a-1)(a-2)...(2)(1)*1}{2^\infty*log^a(2)}= 0 $$

Finally the last condition of L'Hopital rule complies:

  • The limit exist(in this case is defined).

and the limit is zero as requested so $n^{a}=o(2^n)$.

Renato Sanhueza
  • 1,335
  • 8
  • 21
  • 1
    a) What if $a \not\in\mathbb{N}$? b) Your limit calculation is wrong. You don't get to insert $\infty$ as a symbolic value. 3) L'Hospital has certain conditions you have to check. – Raphael Dec 05 '15 at 09:23
  • Hi. The OP asked for $a\in \mathbb{N}$. What do you mean about inserting $\infty$ as a symbolic value?. Finally L'Hopital's conditions meet so I can't see the problem. – Renato Sanhueza Dec 05 '15 at 14:57
  • 1
    Ahh, my mistake: I missed the title. (I've edited the question to make sure it appears in the body, so others don't miss it.) So, that addresses Raphael's point a), and your edit to the answer addresses Raphael's point 3). Thank you for the nice improvements! Seems like a nice answer to me. – D.W. Dec 14 '15 at 01:05