Show that $$\sum\limits_{k=1}^n \frac{1}{k} \geq \log(n)$$ Use this to deduce that the series $$\sum\limits_{k=1}^\infty \frac{1}{k}$$ diverges. Hint: Use the estimate $$\frac{1}{k} \geq \int_k^{k+1} \frac{1}{x}\,\mathrm{d}x$$

Note that all instances of "$\log$" mean the natural logarithm, the inverse of the exponential. After doing the integration, I've concluded that $\frac{1}{k} \geq \log(1+\frac{1}{k})>-\log(k)$

I suspect I'm supposed to use the hint to arrive at the first inequality, and then since $\log(n)$ can become arbitrarily large as $n$ increases, so must the harmonic series (essentially the comparison test).

I'm having trouble making this jump. Hints are appreciated, but please no solutions.


$$\frac1k\ge\int_k^{k+1}\frac{dx}x=\log\frac{k+1}k\implies\sum_{k=1}^n\frac1k\ge\log\left(\prod_{k=1}^n\frac{k+1}k\right)\ge\log n$$

  • $\begingroup$ I'm with you up until the last equality - I keep getting $\log(n+1)$. Not that it matters, since $\log(n+1)>\log(n)$, so the conclusion I need can be reached regardless. $\endgroup$ – Ducky Feb 2 '15 at 21:40
  • $\begingroup$ Yes, you get $\;\log(n+1)\;$ indeed...and indeed it doesn't matter for the reason you mention. Editing the typo. $\endgroup$ – Timbuc Feb 2 '15 at 21:41

we can use the fact that $$\dfrac{1}{k-1} + \dfrac{1}{k}+\dfrac{1}{k+1} = \dfrac{1}{k}+\dfrac{2k}{k^2 - 1} > \dfrac{1}{k}+\dfrac{2}{k} = \dfrac{3}{k}\text{ for } k \ge 2$$ to show that $s=1 + \frac12 + \frac13 + \cdots$ cannot be finite, therefore the harmonic series must diverge.

$\begin{align} s &= \dfrac{1}{1} + \dfrac{1}{2} + \dfrac{1}{3} +\dfrac{1}{4} + \dfrac{1}{5}+ \cdots \\ &=1+\left(\dfrac{1}{2} + \dfrac{1}{3}+\dfrac{1}{4} \right)+ \left(\dfrac{1}{5} + \dfrac{1}{6}+\dfrac{1}{7} \right) + \left(\dfrac{1}{8} + \dfrac{1}{9}+\dfrac{1}{10} \right) + \left(\dfrac{1}{11} + \dfrac{1}{12}+\dfrac{1}{13} \right) + \cdots \\ &> 1+3\left(\frac13 + \frac16+\frac19+\frac1{12} \cdots\right) = 1+1 +\frac12 + \frac13+\frac14 + \cdots\\ &=1 + s \end{align}$

we have $ > s+1$, therefore $s$ cannot be a finite number.

  • $\begingroup$ Why is $s>s-\frac16$ a contradiction? $\endgroup$ – LutzL Feb 2 '15 at 23:02
  • $\begingroup$ @LutzL, thanks for spotting the error. i have fixed the problem. $\endgroup$ – abel Feb 3 '15 at 16:28

An other way:

Let $S_n=\sum_{k=1}^n\frac{1}{k}$.

$$|S_{2n}-S_n|=\sum_{k=n+1}^{2n}\frac{1}{k}=\underbrace{\frac{1}{n+1}}_{\geq \frac{1}{2n}}+\underbrace{\frac{1}{n+2}}_{\geq \frac{1}{2n}}+...+\underbrace{\frac{1}{2n}}_{\geq \frac{1}{2n}}\geq \frac{n}{2n}=\frac{1}{2}$$

and thus, $(S_n)$ is not a Cauchy sequence, therefore it doesn't converge.



  • 1
    $\begingroup$ Did you read the question, or just the title? $\endgroup$ – Akiva Weinberger Jul 21 '15 at 3:06

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.