If a sequence ($a_n$) is monotonically increasing, and ($b_n$) is a decreasing sequence, with $\lim_{n\to\infty}\,(b_n-a_n)=0$, show that $\lim a_n$ and $\lim b_n$ both exist, and that $\lim a_n=\lim b_n$.

My attempt:

To show that the limits of both sequences exist, I think I should be using the Monotone Convergence Theorem (MCT). For that I would need to show that the sequences are bounded.

($a_n$) is increasing, and so it should be bounded below. ($b_n$) is decreasing, so it should be bounded above. The challenge here is to show that ($a_n$) can be bounded above and ($b_n$) can be bounded below. This should utilise the third condition, from which I get:

$$\begin{align*} & \lim_{n\to\infty}\,(b_n-a_n)=0 \\[3pt] \iff & \forall\varepsilon>0,\ \exists N\in \mathbb{N} \text{ s.t. } \forall n\geq N,\ |{b_n-a_n}|<\varepsilon \end{align*}$$

I then tried using the triangle inequality: $$ |b_n|-|a_n|\leq|b_n-a_n|<\varepsilon$$

but I'm not sure where to go from here.

  • 1
    $\begingroup$ (+1) for the fluffy little dog (and a well posed question). $\endgroup$ – tired Feb 4 '17 at 9:41
  • 1
    $\begingroup$ @tired It's my favourite dog breed. Thanks, haha. :) $\endgroup$ – Troy Feb 4 '17 at 10:05

The leftmost part of the inequality $|b_n|-|a_n|\leq|b_n-a_n|<\epsilon$ isn't needed in the proof.

Claim 1     $(a_n)$ is bounded above.

Proof    Let $\epsilon>0$. From $|b_{n+1}-a_{n+1}|<\epsilon$, $a_{n+1}<b_{n+1}+\epsilon$. Use the monotonicity of $(a_n)$ and $(b_n)$. We have $$a_1 \le \dots \le a_n\le a_{n+1} < b_{n+1}+\epsilon \le b_n+\epsilon \le \dots \le b_1 + \epsilon.$$ Since the choice of $n$ in the above inequality is arbitrary, we have $a_n < b_1 + \epsilon$ for all $n \in \Bbb N$. Therefore, $(a_n)$ is bounded above by $b_1 + \epsilon$.

Similarly, we have another claim.

Claim 2     $(b_n)$ is bounded below.

Now, recall that $(a_n)$ and $(b_n)$ are increasing and decreasing sequences respectively, and apply MCT to $(a_n)$ and $(b_n)$ to establish the existence of $\lim a_n$ and $\lim b_n$. Finally, use $\lim\limits_{n\to+\infty}(b_n-a_n)=0$ to conclude that $\lim a_n = \lim b_n$.

Sorry for using others' ideas in my solution. I would like to draw a commutative diagrams in the comments, but the system forbids me from posting comments with two or more @ characters, so I can't post the following diagram in a comment. Hoping that others can benefit from his answer at the first glance, I draw this diagram for fun.

A graphical explanation to DonAntonio's answer

$\require{AMScd}$ \begin{CD} @. a_n \\ @. @AA \vdots A \\ @. a_{N+1} \\ @. @AA (a_n)\uparrow A \\ @. a_N \\ \text{Suppose }a_N > b_K. \\ b_K @. \\ @V(b_n)\downarrow VV @.\\ b_{K+1} @. \\ @V \vdots VV @.\\ b_n @. \end{CD} It's clear from the diagram that we have to take $n \ge \max\{K,N\}$. But $\lim\limits_{n\to+\infty}(b_n-a_n)=0$, contradiction.

  • 1
    $\begingroup$ Nice explanation to my answer. +1 $\endgroup$ – DonAntonio Feb 4 '17 at 12:57


Suppose there exists $\;K\in\Bbb N\;$ such that for some $\;N\in\Bbb N\;,\;\;a_N>b_K\;$ , say $\;a_N-b_K=\epsilon>0\;$ , but then

$$\forall\,n\ge N\,,\,\,\forall\,m\ge K\;,\;\;\begin{cases}a_n\ge a_N>b_K,&\text{since $\,\{a_n\}\,$ is monotone ascending}\\{}\\ b_m\le b_K<a_N,&\text{since $\,\{b_n\}\,$ is monotone descending}\end{cases}$$

and from here we'd get that for any $\;n\ge\max\,\{K,N\}\;$ :

$$a_n\ge a_N>b_K\ge b_n\implies a_n-b_n>\epsilon>0\implies \lim_{n\to\infty}(a_n-b_n)\neq0$$

and the above shows not only both sequence are bounded in the right direction, but also that they both bound each other resp.

  • $\begingroup$ If I employ a graphical argument, if I have an increasing sequence $(a_n)$ and decreasing sequence $(b_n)$ and they have the same limit, I don't see how and when $a_N>b_K$. $\endgroup$ – Troy Feb 4 '17 at 10:04
  • $\begingroup$ @Troy Exactly the answer's point ! It can't be, and thus the whole sequence $\;a_n\;$ is bounded above by the $\;b_n\,$'s . Didn't you understand what the above is trying to tell you? $\endgroup$ – DonAntonio Feb 4 '17 at 10:22
  • 1
    $\begingroup$ I did not, but now I do. Thanks for taking the time to explain your answer! $\endgroup$ – Troy Feb 4 '17 at 10:33
  • 1
    $\begingroup$ @Troy No worries. Hoping that you can benefit from this answer, I attempted to post a commutative diagram in a comment, but the system doesn't let me post it due to the presence of more than one @ character, so I have to post it in my answer. Btw, +1 $\endgroup$ – GNUSupporter 8964民主女神 地下教會 Feb 4 '17 at 11:12

You choose an $\varepsilon>0$ and have $$|(a_n-b_n)|\lt\varepsilon, \forall n> N_0$$ because $\lim (a_n-b_n)=0$. To finish your proof, assume that $\lim a_n=a$ and $\lim b_n=b$. Then you have $$|a_n-a|<\varepsilon, \;\forall n>N_1$$ and $$|b_n-b|<\varepsilon, \; \forall n>N_2$$ So select an $n$ such that $n>N_0, n>N_1, n>N_2$ to get $$|a-b|\le|a-a_n|+|a_n-b_n|+|b_n-b|\lt 3\varepsilon$$

$\varepsilon$ was chosen arbitrary therefore $$|a-b|=0$$


Since $\lim_{n\to\infty}(b_n-a_n)=0$, there is an $N$ such that $|a_n-b_n|<1$ for all $n\ge N$. ($1$ is a number that I have just chosen for $\varepsilon$.) Since $b_n$ is decreasing, we have $a_n<b_n+1\le b_N+1$ for all $n\ge N$. Therefore $a$ is bounded from above, indeed by $\max\left\{a_0,a_1,\dots,a_{N-1},b_N+1\right\}$.

Therefore $a$ converges. Since $b_n = a_n + (b_n-a_n)$ and both $a$ and $b-a$ converge, $b$ also converges.


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.