# Thread: Why can I ignore a monotonic function?

1. ## Why can I ignore a monotonic function?

I am attempting to tackle a problem in a computer science course which involves taking a limit. I had gotten a piece of advice but I'm not quit sure about some of the details and I was hoping someone could give me some clarification.

The problem is...

Prove the is $n\sqrt{\log{n}}$ is $\omega(\sqrt{n}\log{n})$

My book says that it comes down to showing that $\lim_{x \to \infty} \frac{f(n)}{g(n)}} = \infty$
So I have to find $\lim_{x \to \infty} \sqrt{\frac{n}{\log{n}}}}}$

I'm told that this is equivalent to finding $\lim_{x \to \infty} \frac{n}{\log{n}}}}$ since the square root is monotonic.

I can use L'Hopital's rule and show that the limit converges to infinity. I do understand why the square root function is monotonic but I don't understand why I'm allowed to disregard the affect of the square root on the function.
Could someone explain why?

2. ## Re: Why can I ignore a monotonic function?

The important thing is not that $\sqrt{n}$ is monotonic, but that it tends to infinity as n -> ∞. For example, h(n) = 1 - 1 / n = (n - 1) / n is also monotonic for n > 0, but h(n / log(n)) = 1 - log(n) / n -> 1 as n -> ∞. Here h(n) -> 1 as n -> ∞ instead of h(n) -> ∞. On the other hand, h(n) = 2sin(n) + n is not monotonic, but h(n / log(n)) -> ∞.

Can you prove that if j(n) -> ∞ and h(n) -> ∞ as n -> ∞, then h(j(n)) -> ∞ as n -> ∞?

3. ## Re: Why can I ignore a monotonic function?

Originally Posted by restin84
I am attempting to tackle a problem in a computer science course which involves taking a limit. I had gotten a piece of advice but I'm not quit sure about some of the details and I was hoping someone could give me some clarification.
The problem is...
Prove the is $n\sqrt{\log{n}}$ is $\omega(\sqrt{n}\log{n})$
My book says that it comes down to showing that $\lim_{x \to \infty} \frac{f(n)}{g(n)}} = \infty$
So I have to find $\lim_{x \to \infty} \sqrt{\frac{n}{\log{n}}}}}$
I'm told that this is equivalent to finding $\lim_{x \to \infty} \frac{n}{\log{n}}}}$ since the square root is monotonic.
I can use L'Hopital's rule and show that the limit converges to infinity. I do understand why the square root function is monotonic but I don't understand why I'm allowed to disregard the affect of the square root on the function.
Will you please review what you posted.
You explain nothing about the notation you are using.
What is the actual question?
What does $\omega(\sqrt{n}\log{n})$ mean?

4. ## Re: Why can I ignore a monotonic function?

I suspect that was supposed to be what some of use would have called " $\theta$" or " $\Theta$" or "O"- f is "O(g)" if and only if f(x)/g(x) goes to 1 as x goes to some limit such as infinity. Saying that $x\sqrt{ln(x)}\epsilon \sqrt{x}ln(x)$ means that $\lim_{x\to\infty}\frac{x \sqrt{ln(x}}{\sqrt{x}ln(x)}= \frac{x}{\sqrt{x}}\frac{\sqrt{ln(x)}}{ln(x)}= \sqrt{\frac{\sqrt{x}}{ln(x)}}$ goes to 1 as x goes to infinity. Of course, since the square root function is continuous for x positive, that will be true as long as $\lim_{x\to 0}\frac{\sqrt{x}}{ln(x)}= 1$. And that can be shown by, for example, L'Hopital's rule.

5. ## Re: Why can I ignore a monotonic function?

Originally Posted by Plato
What is the actual question?
I think the OP is asking why $\lim_{x \to \infty} \frac{n}{\log{n}}}}=\infty$ implies $\lim_{x \to \infty} \sqrt{\frac{n}{\log{n}}}}}=\infty$.

Originally Posted by Plato
What does $\omega(\sqrt{n}\log{n})$ mean?
This is not really relevant to the question (because the problem reduces to proving $\lim_{n\to\infty}\frac{n\sqrt{\log n}}{\sqrt{n}\log n}=\infty$), but apparently this is one of the asymptotic notations.

6. ## Re: Why can I ignore a monotonic function?

Originally Posted by HallsofIvy
Saying that $x\sqrt{ln(x)}\epsilon \sqrt{x}ln(x)$ means that $\lim_{x\to\infty}\frac{x \sqrt{ln(x}}{\sqrt{x}ln(x)}= \frac{x}{\sqrt{x}}\frac{\sqrt{ln(x)}}{ln(x)}= \sqrt{\frac{\sqrt{x}}{ln(x)}}$ goes to 1 as x goes to infinity. Of course, since the square root function is continuous for x positive, that will be true as long as $\lim_{x\to 0}\frac{\sqrt{x}}{ln(x)}= 1$. And that can be shown by, for example, L'Hopital's rule.
Of course, neither $\lim_{x\to 0}\frac{\sqrt{x}}{\ln(x)}$ nor $\lim_{x\to \infty}\frac{\sqrt{x}}{\ln(x)}$ equals 1 (and it should say $\lim_{x\to \infty}\frac{x}{\log(x)}$). And none of big-O, big-Θ, and small-ω notations involve a limit that equals 1.

7. ## Re: Why can I ignore a monotonic function?

Originally Posted by emakarov
I think the OP is asking why $\lim_{x \to \infty} \frac{n}{\log{n}}}}=\infty$ implies $\lim_{x \to \infty} \sqrt{\frac{n}{\log{n}}}}}=\infty$.

This is not really relevant to the question (because the problem reduces to proving $\lim_{n\to\infty}\frac{n\sqrt{\log n}}{\sqrt{n}\log n}=\infty$), but apparently this is one of the asymptotic notations.
Why are we expected to make a guess as to what is meant?

8. ## Re: Why can I ignore a monotonic function?

Originally Posted by Plato
Why are we expected to make a guess as to what is meant?
I feel the same way about some questions on this site, but in this case I think the question was pretty clear.

9. ## Re: Why can I ignore a monotonic function?

Originally Posted by emakarov

Can you prove that if j(n) -> ∞ and h(n) -> ∞ as n -> ∞, then h(j(n)) -> ∞ as n -> ∞?
I was trying to find some information I could use in order to prove the above statement. I'm not really sure where to start.

10. ## Re: Why can I ignore a monotonic function?

Originally Posted by restin84
I was trying to find some information I could use in order to prove the above statement. I'm not really sure where to start.
You need prove that for every M > 0 there exists an N such that for every n > N it is the case that h(j(n)) > M. Fix an arbitrary M. Since h(n) -> ∞ as n -> ∞, there exists a K such that n > K implies h(n) > M. Now you need to guarantee that j(n) > K from some point onward. This would mean that h(j(n)) > M. Can you do this?

11. ## Re: Why can I ignore a monotonic function?

Originally Posted by emakarov
You need prove that for every M > 0 there exists an N such that for every n > N it is the case that h(j(n)) > M. Fix an arbitrary M. Since h(n) -> ∞ as n -> ∞, there exists a K such that n > K implies h(n) > M. Now you need to guarantee that j(n) > K from some point onward. This would mean that h(j(n)) > M. Can you do this?
Honestly no. I've been reading it over for the last fifteen minutes now. I can't really get any kind of "mental picture" I guess.

12. ## Re: Why can I ignore a monotonic function?

The fact that j(n) > K from some point onward, i.e. that there exists an N such that n > N implies j(n) > K, follows by definition from the fact that j(n) -> ∞ as n -> ∞. Verify that h(j(n)) > M for n > N.