11
$\begingroup$

The formulation I'm looking at goes: If $\lbrace x_n\rbrace$, $\lbrace y_n\rbrace$ and $\lbrace z_n \rbrace$ are sequences such that $x_n \le y_n \le z_n$ for all $n \in \mathbb N$, and $x_n \to l$ and $z_n \to l$ for some $l \in \mathbb R$, then $y_n \to l$ also.

So we have to use the definition of convergence to a limit for a sequence: $$\forall \varepsilon > 0, \space \exists N_\varepsilon \in \mathbb N, \space \forall n \ge N_\varepsilon, \space |a_n - l| < \varepsilon$$

I've been trying to say something like: $|y_n - l| < |x_n - l| + |z_n - l| \le \frac\varepsilon 2 + \frac\varepsilon 2 = \varepsilon$ for every $\varepsilon > 0$, but I'm not sure how to get there or if there may be a better way to prove the theorem. Any help would be greatly appreciated.

$\endgroup$
14
$\begingroup$

Write

$$|y_n-l|\le |y_n-x_n|+|x_n-l|\le( z_n-x_n)+|x_n-l|$$ Can you take it from here?

$\endgroup$
  • 1
    $\begingroup$ @silva: And as another tiny hint, there's a metric space axiom you use over and over again that you will want to use here. $\endgroup$ – nomen Feb 5 '15 at 19:43
  • $\begingroup$ @nomen Yeah, I've been taking that inequality over and over. The trick is just using that to get to that final statement I mentioned. $\endgroup$ – Reinstate Monica Feb 5 '15 at 19:46
9
$\begingroup$

Let $\varepsilon > 0$. Since $x_n \to l$, there exists $N_1 = N_1(\varepsilon)$ such that $|x_n - l| < \varepsilon$ for all $n \ge N_1$. Since $z_n \to l$, there exists $N_2 = N_2(\varepsilon)$ such that $|z_n - l| < \varepsilon$ for all $n \ge N_2$. Set $N = \max\{N_1,N_2\}$. If $n \ge N$, then $$y_n - l \le z_n - l < \varepsilon$$ and $$y_n - l \ge x_n - l > -\varepsilon$$ Hence $|y_n - l| < \varepsilon$ for all $n \ge N$. Since $\varepsilon$ was arbitrary, $y_n \to l$.

$\endgroup$
  • 2
    $\begingroup$ Clear and concise. +1. $\endgroup$ – Mark Viola Sep 22 '16 at 2:01
7
$\begingroup$

Ok, so this is what I have now:

Let $x_n \le y_n \le z_n \space\space \forall n \in \mathbb N$ and $\lim_{n \to \infty} x_n = \lim_{n \to \infty} z_n = l$ for some $l \in \mathbb R$.

Then we have: $$\begin{align} &\forall \varepsilon_1 > 0,\, \exists N_{\varepsilon_1} \in \mathbb N,\, \forall n \ge N_{\varepsilon_1}\space |x_n - l| < \varepsilon_1 \\ &\forall \varepsilon_2 > 0,\, \exists N_{\varepsilon_2} \in \mathbb N,\, \forall n \ge N_{\varepsilon_2}\space |z_n -l| < \varepsilon_2 \end{align}$$ Let $N = \max\lbrace N_{\varepsilon_1},N_{\varepsilon_2} \rbrace$ and $\varepsilon = \min\lbrace \varepsilon_1, \varepsilon_2 \rbrace$, so that we have $$\forall \varepsilon > 0,\, \exists N \in \mathbb N,\, \forall n \ge N\space |x_n - l| < \varepsilon \space \text{and} \space |z_n -l| < \varepsilon$$

Let $\frac\varepsilon3 > 0$. Then $\exists N \in \mathbb N,\, \forall n \ge N\space |x_n -l| < \frac\varepsilon3 \text{ and } |z_n -l| < \frac\varepsilon3$ so that $$|z_n - x_n| = |z_n - l + l - x_n| \le |z_n - l| + |x_n - l| < \frac\varepsilon3 + \frac\varepsilon3 = \frac{2\varepsilon}3$$ By assumption, $$\begin{align} x_n \le&\, y_n \le z_n \\ 0 \le&\, y_n - x_n \le z_n - x_n \\ &|y_n - x_n| \le |z_n - x_n| \\ |y_n - l| = |y_n - x_n + x_n - l| \le& |y_n - x_n| + |x_n - l| \le |z_n - x_n| + |x_n - l| < \frac{2\varepsilon}3 + \frac\varepsilon3 = \varepsilon \end{align} $$ And so $\lbrace y_n \rbrace$ also converges to $l$, thus, Q.E.D.

I think that that's satisfactory. If there are any ambiguities or mistakes, please let me know. Thanks to everybody for your advice and contributions.

$\endgroup$
  • $\begingroup$ Can you just assume that, since both $x_{n}$ and $z_{n}$ converge to the same limit, then for $\frac{\epsilon}{2}$ there is an $N$ such that $|z_{n}-x_{n}|<\frac{\epsilon}{2}$, and from here just going on with $|y_{n}-l|=|y_{n}-x_{n}+x_{n}-l|\leq|y_{n}-x_{n}|+|x_{n}-l|\leq|z_{n}-x_{n}|+|x_{n}-l|\leq\frac{\epsilon}{2}+\frac{\epsilon}{2}=\epsilon$? (Here I took $\frac{epsilon}{2}$ as a boundary for $|x_{n}-l|$) $\endgroup$ – Allonsy Dec 13 '17 at 18:01
3
$\begingroup$

How about $$ |y_n-l|\leq\max\{|x_n-l|,|z_n-l|\}<\varepsilon $$ where last inequality of the above holds for $n\geq N$ if $N$ is big enough?

$\endgroup$
3
$\begingroup$

An alternative proof is the following:

Proof: Since $x_n \leq y_n \leq z_n$ then $0\leq y_n-x_n\leq z_n-x_n$, thus $|y_n-x_n|\leq z_n-x_n$.

Combining the above with the fact that $\lim(z_n-x_n)=\lim z_n-\lim x_n=l-l=0 \ $, we get: $$\lim(y_n-x_n)=0$$

Now we can write the terms of $(y_n)$ as the sum of the terms of two converging sequences: $y_n=(y_n-x_n)+x_n$, so we have: $$ \lim y_n=\lim\big((y_n-x_n)+x_n \big)=\lim(y_n-x_n)+\lim x_n=0+l=l $$

$\endgroup$
  • 1
    $\begingroup$ You seem to be using the squeeze theorem in your proof of the squeeze theorem. How do you arrive at that limit y_n-x_n = 0 ?? $\endgroup$ – CogitoErgoCogitoSum Feb 4 '18 at 0:12
  • $\begingroup$ @CogitoErgoCogitoSum: I am using the property that: "if $\lim \beta_n=0$ and there is some $\kappa\in\mathbb{N}$, such that for all $n>\kappa$, $\ |\alpha_n|\leq |\beta_n|$, then $\lim\alpha_n=0$". This property is an immediate consequence of the $\epsilon$-$\delta$ definition of the limit of a sequence and it is generally not referred to as the "squeeze theorem". (although, it can obviously be understood as a special case of the squeeze theorem). $\endgroup$ – KonKan Feb 10 '18 at 2:24
1
$\begingroup$

As $n$ grows, you can get the distance from $x_n$ to $l$ to be less than any $\epsilon > 0$, and the distance from $z_n$ to $l$ to be less than $\epsilon$ as well. Just take the max $N$ of the two indices for $x_n$ and $z_n$ that guarantee this, and you will get that both $x_n$ and $z_n$ are within $\epsilon$ of $l$ for $n \geq N$. What does that say about the distance between $y_n$ and $l$, for $n \geq N$?

$\endgroup$

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.