1
$\begingroup$

Please help me to calculate the following limit $$\lim_{n \to \infty} (\sqrt[3]{n^3+3n^2}-\sqrt{n^2+2n} )$$ I factored out $n$ from both radicals but it didn't work and tried to use the identity $a^2-b^2$ and $a^3-b^3$.

$\endgroup$
  • 1
    $\begingroup$ What have you tried? $\endgroup$ – Robert Shore May 21 at 0:19
  • $\begingroup$ I factored out n from both radicals but it didn't work and tried to use the identity a^2-b^2 and a^3-b^3. $\endgroup$ – Andrei Umbriță May 21 at 0:21
  • 1
    $\begingroup$ Please edit your question to indicate your attempts (rather than putting that information in the comments). Your question is much more likely to get a positive reception if you do so. We are sensitive to people simply showing up and asking a question without any indication they've put effort into finding the solution for themselves. $\endgroup$ – Robert Shore May 21 at 0:23
  • $\begingroup$ are you familiar with binomial series? $\endgroup$ – J. W. Tanner May 21 at 0:27
  • $\begingroup$ Yes, but I would like to see an answer without these or approximations. $\endgroup$ – Andrei Umbriță May 21 at 0:28
4
$\begingroup$

For this type of problem, my preferred weapon is the generalized binomial theorem in the form $(1+x)^a =1+ax+a(a-1)x^2/2+O(x^3) $.

Often the simpler form $(1+x)^a =1+ax+O(x^2) $ is enough.

So,

$\begin{array}\\ \sqrt[3]{n^3+3n^2}-\sqrt{n^2+2n} &=n\sqrt[3]{1+3/n}-n\sqrt{1+2/n}\\ &=n(1+(3/n)(1/3)+O(1/n^2))-n(1+(2/n)(1/2)+O(1/n^2))\\ &=n(1+1/n+O(1/n^2))-n(1+1/n+O(1/n^2))\\ &=n+1+O(1/n)-n-1-O(1/n)\\ &=O(1/n)\\ &\to 0 \qquad\text{as } n \to \infty\\ \end{array} $

Note that if what was wanted was $n$ times the difference, the additional term in the expansion would have been needed.

You might try this for practice.

$\endgroup$
  • $\begingroup$ by "difference", did you mean the case in Riemann's answer? is that rigorous? $\endgroup$ – farruhota May 21 at 6:29
  • $\begingroup$ Yes and yes. The additional term will give the 1/2n result. $\endgroup$ – marty cohen May 21 at 15:10
  • $\begingroup$ without the additional term it is indeterminate $0/0$, isn't it? Thank you. +1 $\endgroup$ – farruhota May 21 at 15:24
  • $\begingroup$ No. It comes to O(1). $\endgroup$ – marty cohen May 21 at 16:43
  • $\begingroup$ which is not demonstrated, $=\frac{\sqrt[3]{1+\frac{3}{n}}-1-\left(\sqrt{1+\frac{2}{n}}-1\right)}{\frac{1}{n}}$ with $\sqrt[3]{1+\frac{3}{n}}-1 \sim \frac{1}{n},\sqrt{1+\frac{2}{n}}-1\sim \frac{1}{n}$ are not sufficient, aren't they? $\endgroup$ – farruhota May 21 at 17:50
4
$\begingroup$

Another way to solve the problem is the following, where no Differential Calculus is needed.

By elementary algebraic manipulation of radicals, you get:

$$ \begin{split} \sqrt[3]{n^3 + 3n^2} - \sqrt{n^2 + 2n} &= \underbrace{\sqrt[6]{(n^3 + 3n^2)^2}}_{=:\sqrt[6]{a}} - \underbrace{\sqrt[6]{(n^2 + 2n)^3}}_{=:\sqrt[6]{b}} \\ & = \frac{a - b}{\sqrt[6]{a^5} + \sqrt[6]{a^4b} + \sqrt[6]{a^3b^2} + \sqrt[6]{a^2b^3} + \sqrt[6]{ab^4} + \sqrt[6]{b^5}} \\ &\approx \frac{-3n^4}{6n^5} \\ &= - \frac{1}{2n} \end{split} $$

with the latter member going to $0$ as $n \to \infty$.

$\endgroup$
3
$\begingroup$

First let's get rid of the square root:

$$\lim_{n \to \infty} (\sqrt[3]{n^3-3n^2}-\sqrt{n^2+2n})=\lim_{n \to \infty} \frac{\sqrt[3]{(n^3+3n^2)^2}-(n^2+2n)}{\sqrt[3]{(n^3+3n^2)}+\sqrt{n^2+2n}}\\= -\lim_{n \to \infty} \frac{n^2+2n-\sqrt[3]{(n^3+3n^2)^2}}{\sqrt[3]{(n^3+3n^2)}+\sqrt{n^2+2n}}.$$

Now let's get rid of the cube root:

$$-\lim_{n \to \infty} \frac{n^2+2n-\sqrt[3]{(n^3+3n^2)^2}}{\sqrt[3]{(n^3+3n^2)}+\sqrt{n^2+2n}}$$

$$=-\lim_{n \to \infty} \frac{(n^2+2n)^3-(n^3+3n^2)^2}{(\sqrt[3]{(n^3+3n^2)}+\sqrt{n^2+2n})((n^2+2n)^2+(n^2+2n)\sqrt[3]{(n^3+3n^2)^2}+\sqrt[3]{(n^3+3n^2)^4})}.$$

The numerator yields $-3n^4$ and a bunch of lower-order terms, and the denominator yields $6n^5$ and a bunch of lower order terms, so dividing numerator and denominator by $n^5$ shows that the answer is $0$.

$\endgroup$
3
$\begingroup$

Use the following fact: when $x\to 0$, $$(1+x)^{\alpha}-1\sim\alpha x.$$ So $$\sqrt[3]{n^3+3n^2}-\sqrt{n^2+2n}=n\left(\sqrt[3]{1+\frac{3}{n}}-1-\left(\sqrt{1+\frac{2}{n}}-1\right)\right)$$ $$=\frac{\sqrt[3]{1+\frac{3}{n}}-1-\left(\sqrt{1+\frac{2}{n}}-1\right)}{\frac{1}{n}},$$ use the above fact: $$\sqrt[3]{1+\frac{3}{n}}-1 \sim \frac{1}{n},\sqrt{1+\frac{2}{n}}-1\sim \frac{1}{n},$$ and this implies $$\lim_{n \to \infty} (\sqrt[3]{n^3+3n^2}-\sqrt{n^2+2n} )=0.$$

In fact: $$\sqrt[3]{n^3+3n^2}-\sqrt{n^2+2n}\sim -\frac{1}{2n},n\to\infty.$$

$\endgroup$
2
$\begingroup$

For positive $n$, let $a=\sqrt[3]{n^3+3n^2}$, and let $b=\sqrt{n^2+2n}$.

Note that $a < n+1$ and \begin{align*} 3n +1 &= (n+1)^3-a^3\\[4pt] &= \bigl((n+1)-a\bigr)\bigl((n+1)^2+a(n+1)+a^2)\\[4pt] &> \bigl((n+1)-a\bigr)(n+1)^2\\[4pt] \end{align*} hence $0 < n+1-a < {\large{\frac{3n+1}{(n+1)^2}}}$, which implies $\displaystyle{\lim_{n\to \infty}}\;(n+1)-a = 0$.

Also, $b < n+1$ and \begin{align*} 1&=(n+1)^2-b^2\\[4pt] &= \bigl((n+1)-b\bigr)\bigl((n+1)+b\bigr)\\[4pt] &> \bigl((n+1)-b\bigr)(n+1)\\[4pt] \end{align*} hence $0 < (n+1)-b < {\large{\frac{1}{n+1}}}$, which implies $\displaystyle{\lim_{n\to \infty}}\;(n+1)-b = 0$.

Then we get $$\lim_{n\to \infty}\;a-b = \lim_{n\to \infty}\;\bigl(a - (n+1)\bigr) + \bigl((n+1)-b\bigr) = 0 + 0 = 0$$

$\endgroup$
  • $\begingroup$ The question has changed. $\endgroup$ – Nosrati May 21 at 0:52
  • $\begingroup$ Ok, I edited my answer to match your edited question. $\endgroup$ – quasi May 21 at 1:01
2
$\begingroup$

There is a way to use directly the two binomial formulas you mentioned. It only needs a little trick:

$$\sqrt[3]{n^3+3n^2}-\sqrt{n^2+2n} = \sqrt[3]{n^3+3n^2}\color{blue}{-n + n}-\sqrt{n^2+2n}$$ Now, consider \begin{eqnarray*} \sqrt[3]{n^3+3n^2}-n & \stackrel{n=\sqrt[3]{n^3}}{=} & \frac{n^3+3n^2 - n^3}{(\sqrt[3]{n^3+3n^2})^2 + n\sqrt[3]{n^3+3n^2}+ n^2}\\& = & \frac{3}{\left(\sqrt[3]{1+\frac{3}{n}}\right)^2 + \sqrt[3]{1+\frac{3}{n}} + 1}\\ & \stackrel{n \to \infty}{\longrightarrow} & 1 \end{eqnarray*}

Similarly, you get $n - \sqrt{n^2+2n} \stackrel{n \to \infty}{\longrightarrow} -1$.

Hence, you get

$$\lim_{n \to \infty} (\sqrt[3]{n^3+3n^2}-\sqrt{n^2+2n} ) = \lim_{n \to \infty} (\sqrt[3]{n^3+3n^2}-n ) + \lim_{n \to \infty} (n-\sqrt{n^2+2n} ) = 1-1 = 0$$

$\endgroup$
0
$\begingroup$

Set $1/x=h,h\to0^+$

The limit becomes $$\lim_{h\to0}\dfrac{\sqrt[3]{1+3h}-1}h-\lim...\dfrac{\sqrt{1+2h}-1}h$$

Now for $F=\lim_{...}\dfrac{\sqrt[m]{1+mh}-1}h,$

set $\sqrt[m]{1+mh}-1=u,$

$\implies mh=(1+u)^m-1=mu+O(u^2)$

Can you take it from here?

$\endgroup$

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.