In my real analysis exam I had a problem in which I proved that $|x_{n+1} - x_n|\lt {a^n}$ for all natural numbers $n$ and for all positive number $a\lt 1$ then $(x_n)$ is a Cauchy sequence.

This was solved successfully but the question is if $|x_{n+1} - x_n|\lt \frac 1n$ does that mean $(x_n)$ is Cauchy? Well my answer was yes because I could write this in the form of the first one, but now I am somehow confused with what I have answered since $1/n$ is a sequence of $n$ so maybe the answer is not necessarily true... Can you please provide me with the correct answer for this question?

  • $\begingroup$ Are you about the first question? I am confused the the "for all a" and even more by the "a<1" (which makes the upper bound on the differences rather large). $\endgroup$ – Dirk Nov 26 '18 at 5:31
  • $\begingroup$ Seems to be some sort of error. The proper correction should be either changing "$\frac 1{a^n}$" to "$a^n$" or changing "$a<1$" to "$a>1$" (minding the radius of convergence for geometric series). $\endgroup$ – Matt A Pelto Nov 26 '18 at 5:44

Consider the harmonic series: $\sum_{n=1}^{\infty}\frac 1n$.

$\mid a_{n+1}-a_n\mid=\frac1{n+1}\lt\frac1n$.

But it diverges.

  • $\begingroup$ This is the difference between $1/n$ and $1/(n+1)$, not between two terms of the sequence of partial sums. The sequence $(1/n)_{n\in \mathbb{N}}$ is indeed Cauchy. $\endgroup$ – Michael Lee Nov 26 '18 at 0:55
  • $\begingroup$ Yes. I'm referring to the sequence of partial sums of the series. It's not Cauchy because it doesn't converge. I was addressing the OP's example. This shows that $\mid a_{n+1}-a_n\mid$ can be less than $\frac1n$, but the sequence can still not be Cauchy. $\endgroup$ – Chris Custer Nov 26 '18 at 1:08
  • $\begingroup$ The difference between partial sums is $1/(n+1)$, not $1/n(n+1)$. You add $1/(n+1)$ to get from the $n$th partial sum to the $(n+1)$th. $\endgroup$ – Michael Lee Nov 26 '18 at 1:08
  • $\begingroup$ Oh yeah. My mistake. $\endgroup$ – Chris Custer Nov 26 '18 at 1:10

No, $\lvert x_{n+1}-x_n\rvert < 1/n$ does not imply that $(x_n)_{n\in \mathbb{N}}$ is Cauchy. Consider $x_n = \sum_{k=1}^n 1/2k$, which does not converge.


Take $x_n=1+\frac 1 2+\cdots+\frac 1 n$. This is not Cauchy because the harmonic series $1+\frac 1 2+\cdots$ is divergent.


This kind of thing works only if you can show $|x_{n + 1} - x_{n}| < a_n$ where $\sum_{k = 0}^\infty a_k < \infty$ because, if this condition holds,

\begin{align} |x_{n} - x_{n + m}| &= |x_{n} - x_{n + 1} + x_{n + 1} - x_{n + 2} + x_{n + 2} - \cdots + x_{n + m - 1} - x_{n + m}| \\ &\le |x_{n} - x_{n + 1}| + \cdots + |x_{n + m - 1} - x_{n + m}| \\ &\le a_n + a_{n + 1} + \dots + a_{n + m - 1} \\ &\le \sum_{k = n}^\infty a_k \end{align}

Now convergence of $\sum a_k$ to $A$ means that for any $\varepsilon > 0$ there exists $N$ such that for all $n \ge N$,

$$ \left| A - \sum_{k = 0}^{n - 1} a_k \right| = \sum_{k = n}^\infty a_k < \varepsilon $$

Comparing this with the above, we have for every $n \ge N$ and every $m \ge 0$,

$$ |x_n - x_{n + m}| < \varepsilon $$

Which means the sequence $(x_n)$ is Cauchy.

If the bound on $|x_{n + 1} - x_n|$ does not converge as a series, you need to use a different trick.

  • 1
    $\begingroup$ +1 for the most informative response and using \varepsilon...no effort has been spared here. I will however nitpick at the last sentence by noting that a different trick may or may not still work, depending on whether the sequence is indeed Cauchy which I presume you know but just add as clarification for people such as the question asker. An example where a different trick might apply is the sequence of partial sums of the alternating series $\sum_{n=1}^\infty \frac{(-1)^{n+1}}{n}$. $\endgroup$ – Matt A Pelto Nov 26 '18 at 1:28

For each $n \in \mathbb{N}$, define $x_n:=1^{-1}+2^{-1}+\cdots+n^{-1}$. Notice $|x_{n+1}-x_n|=\frac{1}{n+1}$ but the sequence $\{x_n\}_{n=1}^\infty$ is not Cauchy as its terms are just the partial sums of the harmonic series which is known to not converge and $\mathbb{R}$ is complete.


This would imply that any series with a general term which tends to $0$ is convergent. This is false (except for $p$-adic numbers…).


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.