Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $\langle x_n\rangle $ be a sequence defined recursively by $ 0<a \le x_1 \le x_2\le b $ and $ x_{n+2} =\sqrt{x_nx_{n+1}} $ for each n $ \in \Bbb N $ show that $|x_{n+2} -x_{n+1}| \le \frac {b}{b+a} |x_n -x_{n+1}|$ Deduce that $\langle x_n \rangle$ is Cauchy and find its limit.

Can some one help me on this especially in deducing its Cauchy! I think the 1st part can be proven by induction though I had some problem there too.

share|cite|improve this question
To get real angle brackets for the sequences, use \langle ($\langle$) and \rangle ($\rangle$). – Brian M. Scott Jun 21 '13 at 7:50
To show that $|x_{n+2}-x_{n+1}|\leqslant\cdots$, did you try to write down $|x_{n+2}-x_{n+1}|$ as a function of $x_{n+1}$ and $x_n$? – Did Jun 21 '13 at 8:01

Well, this is not an answer, just an alternative approach.

Let $\theta_n = \ln x_n$. Then we have $\theta_{n+2} = \frac{1}{2} ( \theta_{n+1} + \theta_n)$. It should be clear from this formula that $\theta_n \in [\theta_0, \theta_1]$ for all $n$ (with appropriate adjustments for the relative order of $\theta_0, \theta_1$), so the sequence is well defined.

Then it is easy to compute that $\theta_{n+2}-\theta_{n+1} = \frac{1}{2}(\theta_n - \theta_{n+1})$, which gives $|\theta_{n+2}-\theta_{n+1}| = \frac{1}{2}|\theta_{n+1} - \theta_n|$, and so $|\theta_{n+1} - \theta_n| = \frac{1}{2^n} |\theta_2-\theta_1|$. Then summing gives (assuming $n\ge m$ for simplicity) $$ |\theta_{n}-\theta_{m}| =|\theta_{n}-\theta_{n-1}| + ... + |\theta_{m+1}-\theta_{m}| = \frac{2}{2^m} ( 1-\frac{1}{2^{n-m}} )|\theta_2-\theta_1| \le \frac{2}{2^m} |\theta_2-\theta_1| $$

It follows that $\theta_k$ is Cauchy, and since $x \mapsto e^x$ is uniformly continuous on $[\theta_0, \theta_1]$, the sequence $x_k$ is also Cauchy.

share|cite|improve this answer
This also makes it easy to compute that $\lim_{n\to \infty} \theta_n = \frac{2\theta_2 + \theta_1}{3}$ (notice OP's indices start at $1$). I love this answer. Ehm... I mean, alternative approach. – Arthur Jun 21 '13 at 8:21
@Arthur: Very nice observation! – copper.hat Jun 21 '13 at 8:56

We can estimate \begin{eqnarray} |x_{n+2}-x_{n+1}| & = & |\sqrt{x_{n+1} x_{n+1}}-\sqrt{x_n x_{n+1}}| = \frac{|x_{n+1}(x_{n+1}-x_n)|}{\sqrt{x_{n+1} x_{n+1}} + \sqrt{x_n x_{n+1}}} \\ & = & \frac{1}{1+\sqrt{x_n/x_{n+1}}}|x_{n+1}-x_n| \leq \frac{1}{1+\sqrt{a/b}}|x_{n+1}-x_n| \\ & \leq & \frac{1}{1+a/b/\sqrt{a/b}} |x_{n+1}-x_n| \leq \frac{1}{1+a/b} |x_{n+1}-x_n| \\ & = & \frac{b}{a+b} |x_{n+1}-x_n| \ . \end{eqnarray} Assume $m>n$. Now \begin{eqnarray} |x_m-x_n| & = & \bigg| \sum_{k=n}^{m-1} x_{k+1}-x_k \bigg| \leq \sum_{k=n}^{m-1} |x_{k+1}-x_k| \leq \sum_{k=n}^\infty \bigg(\frac{b}{a+b}\bigg)^{k-1} |x_2-x_1| \\ & = & \frac{1}{1-\frac{b}{a+b}} \bigg(\frac{b}{a+b}\bigg)^{n-1} |x_2-x_1| \rightarrow 0 \ , \end{eqnarray} as $m,n \rightarrow \infty$, where the second inequality is estalished by induction to the index $m-1$. This shows that the sequence is a Cauchy-sequence and has a limit $x$.

The sequence can be mapped using the logarithm function, $y_i = \ln(x_i)$, and we obtain $y_{i+2} = \frac{1}{2}(y_{i+1}+y_i)$ and hence $y_{i+2}-y_{i+1} = -\frac{1}{2}(y_{i+1}-y_i)$. Now \begin{eqnarray} y & = & y_1 + \sum_{k=1}^\infty y_{k+1}-y_k = y_1 + \sum_{k=1}^\infty \bigg(-\frac{1}{2}\bigg)^{k-1} (y_2-y_1) = y_1 + \frac{1}{1+\frac{1}{2}} (y_2-y_1) \\ & = & \frac{y_1+2y_2}{3} \ , \end{eqnarray} where the second equation is established termwise by induction. Hence \begin{eqnarray} x = e^y = e^\frac{y_1+2y_2}{3} = (e^{y_1} e^{y_2^2})^\frac{1}{3}= (x_1 x_2^2)^\frac{1}{3} \ . \end{eqnarray}

share|cite|improve this answer
Interesting spelling correction. – copper.hat Jun 21 '13 at 14:48

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.