I have the following proof for $\lim_{n\rightarrow\infty} \sqrt{4n^2+n} - 2n = \frac{1}{4}$ and was wondering if it was correct. Note that $\sqrt{4n^2+n} - 2n = \frac{n}{\sqrt{4n^2+n} + 2n}$. $$\left|\frac{n}{\sqrt{4n^2+n} + 2n} - \frac{1}{4}\right| \\ = \left|\frac{2n - \sqrt{4n^2+n}}{4(\sqrt{4n^2+n} + 2n)}\right|=\left|\frac{\sqrt{4n^2+n} - 2n}{4(\sqrt{4n^2+n} + 2n)}\right|\\ = \left|\frac{n}{4(\sqrt{4n^2+n} + 2n)^2}\right| \leq \left|\frac{n}{4(4n)^2}\right| = \left|\frac{n}{64n^2}\right| \\ = \left|\frac{1}{64n}\right| < \epsilon \\ \implies n>\frac{1}{64\epsilon}$$

  • 3
    $\begingroup$ Seems quite right, though the final implication must be in the other direction. $\endgroup$
    – user65203
    Aug 6, 2019 at 13:30
  • 1
    $\begingroup$ It's wrong, for the reason Yves gave. $\endgroup$ Aug 6, 2019 at 13:34
  • $\begingroup$ Did you finish this problem @user100000001? $\endgroup$
    – Axion004
    Sep 20, 2020 at 18:48

3 Answers 3


You want to show that $\lim_{n\rightarrow\infty} \sqrt{4n^2+n} - 2n = \frac{1}{4}$. To do this, I would split up the analysis into scratch work and the formal proof.

For the scratch work, you need to find a suitable upper bound. You have done this by showing

\begin{align} \left|\frac{n}{\sqrt{4n^2+n} + 2n} - \frac{1}{4}\right| & = \left|\frac{2n - \sqrt{4n^2+n}}{4(\sqrt{4n^2+n} + 2n)}\right|\\&=\left|\frac{\sqrt{4n^2+n} - 2n}{4(\sqrt{4n^2+n} + 2n)}\right|\\& = \left|\frac{n}{4(\sqrt{4n^2+n} + 2n)^2}\right| \\&\leq \left|\frac{n}{4(4n)^2}\right| \\&= \left|\frac{n}{64n^2}\right| \\& = \left|\frac{1}{64n}\right| \\&< \epsilon \end{align}

which means that $n>\frac{1}{64\epsilon}$ is the upper bound.

For the formal proof:

Let $\epsilon>0$ (you need to fix $\epsilon$ as a small positive constant). It follows from $\frac{1}{\epsilon}>0$ that $\frac{1}{64\epsilon}>0$. Then by the Archimedean property there exists a $N\in\mathbb N$ such that $N>\frac{1}{64\epsilon}$. Then if $n\geq N > \frac{1}{64\epsilon}$, we have

$$\left|\frac{n}{\sqrt{4n^2+n} + 2n} - \frac{1}{4}\right|\leq \frac{1}{64n}<\epsilon $$

where the last inequality follows from

$$n\geq N > \frac{1}{64\epsilon} \implies n>\frac{1}{64\epsilon} \implies\epsilon > \frac{1}{64n}$$


Like the other commenters have mentioned, the implication is in the wrong direction. Since you want $|\frac{1}{64n}| < \epsilon$, you are required to have $n > \frac{1}{64 \epsilon}$. In other words, $n > \frac{1}{64 \epsilon}$ implies $|\frac{1}{64n}| < \epsilon$. What you have written is the other way around.


You did the scratch work correctly. But I wouldn't call this a proof (of course it contains all ingredients of a good proof!).

You didn't introduce $\epsilon$. Of course, everyone knows what you mean but one should write it out if one wants to be fully rigorous.

So, your proof should read:

Let $\epsilon >0$. Let $N$ be an integer greater than --insert your choice for $N$ here--.

If $n \geq N$, we have

--insert your estimations here--

Thus, by definition of limit, the result follows. QED


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.