3
$\begingroup$

I have to show that this sequence $$ x_n=\sqrt{a+x_{n-1}} \hbox{ with } x_1=\sqrt{a} $$ is a Cauchy sequence for every $a>0$. I have done the following calculations: $$ \left| x_{n+2}-x_{n+1} \right|=\left| \sqrt{a+x_{n+1}}-x_{n+1}\right|=\left|\frac{a+x_{n+1}-x_{n+1}^2}{\sqrt{a+x_{n+1}}+x_{n+1}} \right|= \left|\frac{x_{n+1}-x_n}{\sqrt{a+x_{n+1}}+x_{n+1}}\right| $$ I don't come up with a boundary for the denominator so that $$ \left|\frac{x_{n+1}-x_n}{\sqrt{a+x_{n+1}}+x_{n+1}}\right|<k\cdot\left|x_{n+1}-x_n\right| \hbox{ with } k<1 $$ Any hint? I know I can use induction to easily prove that the sequence is convergent, but I'd like to prove it's Cauchy without using convergence. Thank you in advance.

$\endgroup$
  • $\begingroup$ Just an idea - wouldn't it make sense to square both sides of the first equation and use $$x_n^2-a=x_{n-1}$$ to show the convergence? $\endgroup$ – Yuriy S Jun 9 '16 at 14:00
  • $\begingroup$ But you wrote $a>0$, it doesn't converge otherwise and $x_1$ becomes complex $\endgroup$ – Yuriy S Jun 9 '16 at 14:37
3
$\begingroup$

In order to prove that your sequence is a Cauchy sequence it is probably easier to prove that $$ \sqrt{a},\;\sqrt{a+\sqrt{a}},\;\sqrt{a+\sqrt{a+\sqrt{a}}},\;\ldots $$ is a convergent sequence. The candidate limit is the positive solution of $x^2-x=a$, namely $\frac{1+\sqrt{1+4a}}{2}>\sqrt{a}$. Your sequence is trivially increasing, so you just need to show it is bounded.
We have: $$\sqrt{a}<\frac{1+\sqrt{1+4a}}{2}$$ and the function $f(x)=\sqrt{a+x}$ is increasing over $\mathbb{R}^+$, hence by applying $f$ to both sides, $$ \sqrt{a+\sqrt{a}} < \frac{1+\sqrt{1+4a}}{2}$$ follows, and by induction it follows that $x_n<\frac{1+\sqrt{1+4a}}{2}$ holds for any $n\in\mathbb{N}$. On the other hand, $$ L=\lim_{n\to +\infty} x_n = \sup_{n\geq 1}x_n $$ has to fulfill $L^2-L=a$. That proves:

$$ \forall a>0,\quad \sqrt{a+\sqrt{a+\sqrt{a+\sqrt{a+\ldots}}}} = \frac{1+\sqrt{1+4a}}{2}.$$

$\endgroup$
  • $\begingroup$ Yes, I know I can use induction to show convergence but I'd like not to use convergence. Only a proper boundary. Thanks anyway. $\endgroup$ – Claudio Jun 9 '16 at 14:52
2
$\begingroup$

Claim. The given sequence is bounded from above by $$ \frac{1 + \sqrt{4a + 1}}{2} = \alpha $$

Proof. We proceed by induction. For $ k = 1 $ we have that

$$ \sqrt{a} = \frac{2\sqrt{a}}{2} = \frac{\sqrt{4a}}{2} < \frac{1 + \sqrt{4a + 1}}{2} $$

Now, assuming that the statement is true for $ k = n $, we have

$$ x_{n+1} = \sqrt{a + x_n} < \sqrt{ \frac{1 + 2a + \sqrt{4a + 1}}{2} } $$

Observe that

$$ \left( \frac{1 + \sqrt{4a + 1}}{2} \right)^2 = \frac{4a + 2 + 2\sqrt{4a + 1}}{4} = \frac{1 + 2a + \sqrt{4a + 1}}{2} $$

which establishes the bound for $ x_{n+1} $.

Now, we prove that the sequence is monotone increasing. For this, the following inequality is crucial:

Claim. For $ 0 \leq x < \alpha $, we have that $ x < \sqrt{a+x} $.

Proof. The given inequality is true precisely when $ x^2 - x - a < 0 $. The inequality factors as $ (x - \alpha)(x - \alpha') < 0 $ where $ \alpha' < 0 $, therefore the second factor is always positive when $ x \geq 0 $, and the first factor is negative precisely when $ x < \alpha $.

Now, since we have $ 0 \leq x_k < \alpha $ for all $ k $, it follows by this claim that

$$ x_{k+1} = \sqrt{a + x_k} > x_k $$

Since the sequence $ x_k $ is monotone increasing and bounded from above, by the monotone convergence theorem it is convergent; and thus Cauchy.

$\endgroup$
1
$\begingroup$

You could use another way $$ x_{n+2}-x_{n+1}={x_{n+1}-x_n\over x_{n+2}+x_{n+1}} $$ As $x_n>0\;\forall n$, the above relation shows that your sequence is monotone.

$x_{n+1}<\sqrt{a}+1+\sqrt[2^n]a$

Proof: Let's name the above statement $P(n)$. For positive numbers $x,y$ we have $$ \sqrt{x+y}<\sqrt x+\sqrt y$$ $x_1=\sqrt{a}<\sqrt{a}+1+a=\sqrt{a}+1+\sqrt[2^0]{a}$. Suppose $P(n)$ is true. $$ x_{n+2}=\sqrt{a+x_{n+1}}<\sqrt{a+\sqrt a+1+\sqrt[2^n]{a}}<\sqrt{(\sqrt a+ 1)^2+\sqrt[2^n]{a}}<\sqrt a+1+\sqrt[2^{n+1}]a $$ We complete our proof using induction.

As $$\lim_{n\to\infty}\sqrt[2^{n}]a=1,\\ \exists m: n>m\implies x_{n}<\sqrt a+2$$ Therefore your sequence is bounded above by $\max(x_1,x_2,...,x_m,\sqrt a+2)$ and below by $\sqrt a$. As every bounded monotone sequence is convergent and every convergent sequence in $\mathbb{R}$ (or a complete metric space) is a Cauchy sequence, your sequence is Cauchy.

$\endgroup$

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.