Fix $p \ge 0$. I'm having trouble determining whether or not $$\sum_{n=1}^\infty \frac{\sqrt{n+1}-\sqrt n}{n^p}$$ converges.

Clearly, the limit criterion is satisfied and I have tried to apply the ratio test to no success (I receive $|a_{n+1}/a_n| = 1$ and thus the test is inconclusive. I don't think the root test nor integral test would help since it is almost impossible to integrate or take the $n$ root of the summand.

Thus, I believe the key to solving this problem is the comparison test. I however cannot think of a series that I may use for comparison. Does anybody have a suggestion?

  • $\begingroup$ What if you knew that $\sqrt{n+1}-\sqrt{n}=O(n^{-1/2})$? $\endgroup$ – uniquesolution Nov 6 '15 at 9:31
  • $\begingroup$ Well, $\sqrt{n+1}-\sqrt n\approx{1\over2\sqrt n}$ $\endgroup$ – Ivan Neretin Nov 6 '15 at 9:31
  • $\begingroup$ @IvanNeretin Yes I did already try that I was able to compare the series to $\sum \frac{1}{2n^{p+1/2}}$ but this will only converge for $p > 1/2$. I am looking for convergence when $p \ge 0$ in general. $\endgroup$ – MathMajor Nov 6 '15 at 9:33

$\frac{\sqrt{n+1}-\sqrt n}{n^p} = \frac{1}{n^p(\sqrt{n+1}+\sqrt{n})},$ and the latter is equivalent to $\frac1{2n^{p+1/2}}$ You just have to compare with Riemann series.

Without $\sim$:

Let $a_n = \frac{\sqrt{n+1}-\sqrt n}{n^p}$, and $b_n = \frac1{2n^{p+1/2}}$. Then $\frac{a_n}{b_n} \xrightarrow[n\to\infty]{} 1$, ie $a_n = O(b_n)$. Thus, if $p>1/2$, as $\Sigma b_n$ converges, then $\Sigma a_n$ converges.

Notice also that $\frac{b_n}{a_n} \xrightarrow[n\to\infty]{} 1$, ie $b_n = O(a_n)$. The reciprocal of the comparison test asserts that if $\Sigma b_n$ diverges, then $\Sigma a_n$ diverges.

Thus, if $p\leq \frac12$, as $\Sigma b_n$ diverges, then $\Sigma a_n$ diverges too.

  • $\begingroup$ Please see my comment to @IvanNeretin. Here $p \ge 0$. And clearly if $p = 1/2$ the harmonic series $\sum \frac{1}{2n^{1/2+1/2}} = \sum \frac{1}{2n}$ does not converge. $\endgroup$ – MathMajor Nov 6 '15 at 9:34
  • $\begingroup$ Sure. The convergence or not of your serie depends on the value of $p$. If $p\leq\frac12$, then it diverges, and if $p>\frac12$, then it converges. $\endgroup$ – Kevin Quirin Nov 6 '15 at 9:35
  • $\begingroup$ Divergence of the strictly greater series does not imply divergence of the original series. $\endgroup$ – MathMajor Nov 6 '15 at 9:36
  • $\begingroup$ The comparison test says : If $a_n \sim b_n$, then $\Sigma a_n$ converges iff $\Sigma b_n$ converges. Here, $a_n$ is $\frac{\sqrt{n+1}-\sqrt n}{n^p}$ and $b_n$ is $\frac1{2n^{p+1/2}}$. $\endgroup$ – Kevin Quirin Nov 6 '15 at 9:40
  • $\begingroup$ Sorry we have not used the $\sim$ notation. Can this be explained without asymptotics? $\endgroup$ – MathMajor Nov 6 '15 at 9:42

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.