-2

Is there an easy way to show that $$\lim_{n\to \infty}\frac {(\log n)^{100}}{\sqrt n}=0 $$

GEdgar
  • 111,679
Jack
  • 107
  • 4
    If the limit is 0 then $\sqrt n$ grows faster not slower than $\log^{100} n$ – Claude Apr 19 '16 at 12:42
  • 1
    Do you mean faster? – gammatester Apr 19 '16 at 12:42
  • 2
    Note: I reformatted your question pretty heavily. Please check to make sure I didn't introduce any errors. In particular, I guessed that you meant $(\log n)^{100}$. I might have gotten that wrong. – lulu Apr 19 '16 at 12:44
  • 1
    Consider the sequence $n=10^{2^k}$... – abiessu Apr 19 '16 at 12:44
  • I meant $$ (logn)^{100}$$ – Jack Apr 19 '16 at 12:57
  • 1
    You should react to the first comments. –  Apr 19 '16 at 13:17
  • Please improve the question by providing additional context, which ideally includes your thoughts on the problem and any attempts you have made to solve it. This information helps others identify where you have difficulties and helps them write answers appropriate to your experience level. – Fly by Night Apr 19 '16 at 16:52

5 Answers5

4

Consider the more general limit: $$ \lim_{x\to \infty}\frac{(\log(x))^{100}}{\sqrt{x}} $$ Then use L'Hopital once to get:

$$ \lim_{x\to \infty}\frac{(\log(x))^{100}}{\sqrt{x}}=200\lim_{x\to \infty}\frac{(\log(x))^{99}}{\sqrt{x}} $$ Now repeat sufficient times to get:

$$ \lim_{x\to \infty}\frac{(\log(x))^{100}}{\sqrt{x}}=K\lim_{x\to \infty}\frac{\log(x)}{\sqrt{x}}=0 $$

Hence for any sequence $a_n,\ n=1, ...$ for which $a_n\to \infty$ we have:

$$ \lim_{n\to \infty}\frac{(\log(a_n))^{100}}{\sqrt{a_n}}=0 $$

etc...

1

Let $m=\log(n)$, then since $e^x\ge1+x\implies e^x\ge\left(1+\frac xk\right)^k$, let $k=101$ $$ \frac{\log(n)^{100}}{\sqrt{n}}=\frac{m^{100}}{e^{m/2}}\le\frac{m^{100}}{\left(1+\frac{m}{202}\right)^{101}}\le\frac{202^{101}}m=\frac{202^{101}}{\log(n)} $$

robjohn
  • 345,667
0

Hint:

Take $n=e^t$, and you need to show that $e^{t/2}$ grows faster than $t^{100}$.

Or, taking the $100^{th}$ root, $e^{t/200}$ grows faster than $t$.

Or by rescaling, $e^u$ grows faster than $200u$, which is the same as $e^{u}$ growing faster than $u$ (or $u$ faster than $\log(u)$).

Then for all $u>1$

$$\frac{\dfrac{e^{u+1}}{u+1}}{\dfrac{e^{u}}{u}}=e\frac{u}{u+1}>\frac e2$$

and $$\frac{e^u}u>\left(\frac e2\right)^u.$$

  • This uses induction, so the results are valid for integer $u$. However, since $u=\log(n)/200$, it is unlikely that $u$ is an integer. – robjohn Apr 19 '16 at 14:30
  • @robjohn: this is only a hint. I am sure you will fill the gap without toil. –  Apr 19 '16 at 15:57
0

More generally, for any $a>0$, $b > 0$, $\lim_{n \to \infty} \frac{(\log(n))^a}{n^b} = 0 $.

There are a number of ways to show this. I like reducing this to a standard form.

First, replacing $n$ by $n^{1/b}$,

$\begin{array}\\ \lim_{n \to \infty} \frac{(\log(n))^a}{n^b} &=\frac{(\log(n^{1/b}))^a}{n}\\ &=\frac{(\log(n)/b)^a}{n}\\ &=\left(\frac{\log(n)}{bn^{1/a}}\right)^a\\ \end{array} $

Then, by replacing $n$ by $n^a$, this becomes

$\begin{array}\\ \lim_{n \to \infty} \frac{(\log(n))^a}{n^b} &=\left(\frac{\log(n)}{bn^{1/a}}\right)^a\\ &=\left(\frac{\log(n^a)}{bn}\right)^a\\ &=\left(\frac{a\log(n)}{bn}\right)^a\\ \end{array} $

So, if we can show that $\lim_{n \to \infty} \frac{\log(n)}{n} =0 $, by working backwards, we can show that $\lim_{n \to \infty} \frac{(\log(n))^a}{n^b} =0 $.

Note that these substitutions can be combined by replacing $n$ by $n^{a/b}$. We get

$\begin{array}\\ \lim_{n \to \infty}\frac{(\log(n))^a}{n^b} &=\lim_{n \to \infty}\frac{(\log(n^{a/b}))^a}{(n^{a/b})^b}\\ &=\lim_{n \to \infty}\frac{((a/b)\log(n))^a}{n^{a}}\\ &=\lim_{n \to \infty}\left(\frac{(a/b)\log(n)}{n}\right)^a\\ \end{array} $

marty cohen
  • 107,799
0

(1).For any $A>0$ we have $\lim_{x\to \infty}(\ln x)/x^A=0.$

Proof: For $x>0$ we have $$\ln x=\int_1^x (1/t)\;dt \quad\text { and }\quad x^A= A(1+ \int_1^x (1/t^{1-A}\;dt\;).$$ Take any $E>0.$ Take $x_0>1$ such that $1/(x_0)^A<E A.$ For $t\geq x_0$ we have $$(1/t)/(1/t^{1-A})= 1/x^A\leq 1/(x_0)^A<E A.$$ $$\text {For } x\geq x_0 \text { we have } \ln x=\ln x_0+\int_{x_0}^t(1/t)\;dt\leq$$ $$\leq \ln x_0+\int_{x_0}^x (E A)(1/t^{1-A})\;dt=$$ $$=E(x^A-(x_0)^A).$$So for $x\geq x_0$ we have $$0<(\ln x)/x^A\leq (-E (x_0)^A+\ln x_0)/x^A+E<(\ln x_0)/x^A+E.$$ From this, since $A>0$ implies $\lim_{x\to \infty}(\ln x_0)/x^A=0,$ we have $$0\leq \lim_{y\to \infty} \sup_{x>y}\;[(\ln x)/x^A]\leq E \quad \text {for every } E>0.$$ $$\text {Therefore }\quad \lim_{x\to \infty}(\ln x)/x^A=0.$$ (2a). For any $B \ne 0$ and $C>0$ let $A=C/|B|.$ We have $A>0$ and we have $$\ln x>1\implies 0\leq |(\ln x)^B/x^C|\leq (\ln x)^{|B|}/x^C=$$ $$((\ln x))/x^A)^{|B|}$$ which tends to zero as $x\to \infty.$

(2b). For $B=0$ and $C>0$ we have,trivally,$x>1\implies (\ln x)^B/x^C=1/x^C$, tending to $0$ as $x\to \infty.$

(3). For $r>0$ we have $\log_r x=(\ln x)/\ln r$ so for $C>0$ we have $|\log_r x)|^B/x^C=|\ln r|^{-B} \cdot |\ln x|^B/x^C$ which tends to $0$ as $x\to \infty.$