10
$\begingroup$

I've frequently used “intuition” to solve limits at infinity. For example, if someone asked me what is: $$ \lim_{x \to \infty} f(x) = \frac {x^5 + x^3 + x}{x^2} $$

Or a sequence that can be represented by such a function, I would quickly argue that the only thing that matters as go through higher and higher numbers for $x$ would be the terms with the higher powers, and so the limit becomes:

$$ \lim_{x \to \infty} f(x) = \frac {x^5 }{x^2} $$

The numerator is growing at a much faster rate than the denominator, and so the function will diverge to $ + \infty $. But, when I was going through a problem from Paul's online notes, my intuition didn't go well with solving it by the book. The problem is:

$$ \left\{ {\frac{{\ln \left( {n + 2} \right)}}{{\ln \left( {1 + 4n} \right)}}} \right\}_{n = 1}^\infty $$

Does this converge to a value? Now, if we use L'Hopital's rule, this is a easy enough problem. It converges to $1$.

$$ \mathop {\lim }\limits_{n \to \infty } \frac{\ln ( n + 2 )}{\ln (1 + 4n )} = \mathop \lim \limits_{n \to \infty } \frac{^{1}/_{(n + 2)}}{^{4}/_{(1 + 4n)}} = \mathop \lim \limits_{n \to \infty } \frac{1 + 4n}{4( {n + 2} )} = 1 $$

But, as I said, sometimes I like to do these by intuition, and what I did was this: as $ n $ approaches infinity, the integers “$2$” and “$1$” won't matter. So the limit becomes: $$ \lim_{n \to \infty} \frac { \ln(n) }{\ln(4n)} $$

The denominator is increasing at a much faster rate than the numerator, and so the limit will converge to $0$.

I understand that something is wrong with my intuition, and intuition is probably not a good way to solve mathematical problems — indeed, the tutorial this problem is from itself mentions that intuition can sometimes lead you astray — but I'd love to gather some insight as to where I'm going wrong.

$\endgroup$
  • $\begingroup$ Base-$a$ logarithm of $b$ is the number of digits you'll get if you write down $b$ in the base-$a$ numeral system. From this intuition you can see that base-$e$ representation of $4n$ is only a constant number of digits longer than of $n$, hence the limit must equal one. (To be precise, I don't even know if there are numeral systems with non-integer bases, but this intuition can still apply. You can read more of this stuff about logarithms at arbital.com/p/logarithm/?l=3wj $\endgroup$ – CrabMan May 6 at 0:05
  • $\begingroup$ Note that in your first two shown equations $\lim_{x\to\infty}$ is missing. $\endgroup$ – ComFreek May 6 at 6:45
  • 2
    $\begingroup$ The introduction is completely wrong. A limit of any real-valued function $f(x)$ is a number (provided it exists), not another function! Possibly you mean a limit $\lim f(x)$ of a function $f(x)=\frac {x^5+\dots}{x^2}$...? But then also that limit is not equal to the function $x\mapsto \frac{x^5}{x^2}$, but rather those two functions have equal limits, which (after appropriate reasoning) allows us to artificially replace one function with another inside the limit expression. $\endgroup$ – CiaPan May 6 at 8:32
  • $\begingroup$ You are using waaaay more brackets than you need to. $\endgroup$ – Acccumulation May 6 at 15:30
41
$\begingroup$

When you have$$\lim_{n\to\infty}\frac{\ln(n)}{\ln(4n)},$$the denominator doesn't increase at a much faster rate than the numerator. As a matter of fact, since we have$$(\forall n\in\mathbb N):\ln(4n)=\ln(4)+\ln(n),$$they increase at the same rate. And now it is easy to see that the limit is indeed $1$.

$\endgroup$
  • 9
    $\begingroup$ And when you play around with logarithms enough (for example in computer science), it will become absolutely intuitive that $\ln(n)$ and $\ln(4n)$ increase at the same rate. $\endgroup$ – JiK May 5 at 19:14
11
$\begingroup$

Note that $\ln(4n)=\ln(4)+\ln n$ so that this function grows at the same rate as $\ln n$. In particular $$ \lim_{n\to \infty} \frac{\ln(n)}{\ln(4n)}= \lim_{n\to \infty} \frac{\ln(n)}{\ln(4)+\ln(n)}=\lim_{n\to \infty} \frac{1}{\frac{\ln(4)}{\ln (n)}+1}=1. $$ Also note that your intuition can be formalized. If $a_n\sim b_n$ and $c_n\sim d_n$ (where $\sim$ means that the ratio of the two sides goes to one), then $$ \lim_{n\to \infty}\frac{a_n}{c_n}=\lim_{n\to\infty}\frac{b_n}{d_n}. $$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.