5
$\begingroup$

I am given a sequence $\{a_n\}$ with $a_n > 0$ and $b_n = a_n + 1/a_n$.

I am first asked to assume that $a_n \ge 1$ and show that the convergence of $\{b_n\}$ implies the convergence of $\{a_n\}$.

My argument is:

Since $\{b_n\}$ converges, $\{b_n\}$ is Cauchy, hence for every $\epsilon > 0$ some $N$ such that for $n_i,n_j > N$, $|b_{n_i} - b_{n_j}| < \epsilon$. Then $|b_{n_i} - b_{n_j}| = |a_{n_i} - a_{n_j} + 1/a_{n_i} - 1/a_{n_j}| < \epsilon$. Suppose that $a_{n_i} = a_{n_j}$. Then clearly $|a_{n_i} - a_{n_j}| < \epsilon$. Alternatively, suppose without loss of generality that $a_{n_i} > a_{n_j}$. Then $1/a_{n_i} - 1/a_{n_j} < 0$ while $a_{n_i} - a_{n_j} > 0$. It is straightforward to observe that $-(1/a_{n_i} - 1/a_{n_j}) \le a_{n_i} - a_{n_j}$ because of the condition $a_n \ge 0$, so it follows that $0 < |a_{n_i} - a_{n_j}| < |a_{n_i} - a_{n_j}|$, so then $|a_{n_i} - a_{n_j}| < |a_{n_i} - a_{n_j} + 1/a_{n_i} - 1/a_{n_j}| < \epsilon$, hence $\{a_n\}$ is Cauchy and therefore converges.

However, I am next asked to assume that $\{b_n\}$ converges but only that $a_n > 0$ and demonstrate, through construction of a counterexample, that it does not necessarily follow that $\{a_n\}$ converges. I am not entirely sure how to do this.

$\endgroup$
1
$\begingroup$

Hint: draw a graph of the function $f(x)= x + \frac{1}{x}, x>0$, what are solutions of $f(x) = b$? with $b = \dfrac{5}{2}$ for example

$\endgroup$
  • $\begingroup$ Oh, I think I see--is it correct to say that $f(x) = f(1/x)$, so if I take one of the convergent sequences $\{a_n\}$ where $a_n \ge 1$ and I change it so that every other term becomes $1/a_n$, that doesn't change $\{b_n\}$ but now $\{a_n\}$ doesn't converge? $\endgroup$ – Marcus Emilsson Oct 19 '14 at 23:46
  • 1
    $\begingroup$ @ElizabethLin Yeah you get it! $\endgroup$ – Petite Etincelle Oct 19 '14 at 23:47
1
$\begingroup$

You are really overthinking the second part.

Take the sequence $a_n = a, 1/a, a, 1/a, a, 1/a, ... $, then $b_n = a + 1/a$ and the sequence $b_n$ is convergent, but $a_n$ is not convergent unless a = 1.

$\endgroup$
0
$\begingroup$

Let $$a_n=2+\sqrt 3(-1)^n$$which is not convergent while $b_n=4$ is convergent to $4$.

$\endgroup$

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.