Prob. 14, Sec. 3.4, in Bartle & Sherbert's INTRO TO REAL ANALYSIS, 4th ed: The supremum of a bounded sequence not in the range

371 Views Asked by At

Here is Prob. 14, Sec. 3.4, in the book Introduction to Real Analysis by Robert G. Bartle & Donald R. Sherbert, 4th edition:

Let $\left( x_n \right)$ be a bounded sequence and let $s \colon= \sup \left\{ \ x_n \ \colon \ n \in \mathbb{N} \ \right\}$. Show that if $s \not\in \left\{ \ x_n \ \colon \ n \in \mathbb{N} \ \right\}$, then there is a subsequence of $\left( x_n \right)$ that converges to $s$.

My Attempt:

As $s-1 < s$, so there exists a natural number $n_1$ such that $$ s-1 < x_{n_1} \leq s. $$ But $s \not\in \left\{ \ x_n \ \colon \ n \in \mathbb{N} \ \right\}$ and $x_{n_1} \in \left\{ \ x_n \ \colon \ n \in \mathbb{N} \ \right\}$. So we must have $$ s-1 < x_{n_1} < s. \tag{1} $$

As $s-1/2 < s$, so there exists a natural number $n_2$ such that $$ s - \frac{1}{2} < x_{n_2} \leq s. $$ But again $s \not\in \left\{ \ x_n \ \colon \ n \in \mathbb{N} \ \right\}$ and $x_{n_2} \in \left\{ \ x_n \ \colon \ n \in \mathbb{N} \ \right\}$. So we must have $$ s- \frac{1}{2} < x_{n_2} < s. \tag{2} $$

If there were no $n_2 > n_1$ for which (2) can hold, then we would obtain $$ x_n \leq s- \frac{1}{2} \ \mbox{ for every natural number } n > n_1. $$ So, for every natural number $n$, we would obtain
$$ x_n \leq \max \left\{ \ s- \frac{1}{2}, x_1, \ldots, x_{n_1} \right\} < s. $$ And this would then imply that $$ \sup \left\{ \ x_n \ \colon \ n \in \mathbb{N} \ \right\} < s. $$

So there does exist a natural number $n_2 > n_1$ for which (2) can hold.

Now suppose that $n_1, n_2, \ldots, n_k$ have been found such that $n_1 < n_2 < \cdots < n_k$ and such that $$ s - \frac{1}{j} < x_{n_j} < s \ \mbox{ for all } j = 1, \ldots, k. $$ Then just as we proceeded from $n_1$ to $n_2$, we can find a natural number $n_{k+1} > n_k$ such that $$ s - \frac{1}{k+1} < x_{n_{k+1} } < s. $$

In this way we obtain a subsequence $\left( x_{n_k} \right)$ of $\left( x_n \right)$ that satisfies $$ s - \frac{1}{k} < x_{n_k} < s $$ for all $k \in \mathbb{N}$ and hence converges to $s$.

Is this proof correct in each and every one of its steps? If not, then where is it incorrect? Is the presentation clear enough too?

1

There are 1 best solutions below

0
On

Your proof is correct and clear! As you show in your proof, we want to define our subsequence inductively:

Calling $\sup\{x_n: n\in\mathbb{N}\}=s$, we know that there must exist a term of the sequence, $x^*_n$, satisfying $s-1<x^*_n<s$. Let this be the first term of our subsequence: $x_{n_1}=x^*_n$.

Now, given $x_{n_k}$ for some $k\geq 1$, choose $x_{n_{k+1}}$ so that (i) $n_{k+1}>n_k$ and (ii) $s-\frac{1}{k+1}<x_{n_{k+1}}<s$.

Condition (i) ensures that the new sequence we're constructing is indeed a subsequence of $(x_n)$. Condition (ii) ensures that we have $$|x_{n_k}-s|<\frac{1}{k}$$ for all $k\in\mathbb{N}$, so that the subsequence has the desired property, namely, that $x_{n_k}\longrightarrow s$.