Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Let $\{u_k\}_{k\in \mathbb{N}}$ and $\{v_k\}_{k\in \mathbb{N}}$ be the two sequences given by $$ \begin{cases} u_0=v_0=1,&\\ u_{k+1}=\left(1-\frac{1}{(k+1)^2}\right)u_k+\left(\frac{1}{k+1}\right)v_k, & (k\in \mathbb{N})\\ v_{k+1}=\left(-\frac{1}{k+1}\right)u_k+\left(1-\frac{1}{(k+1)^2}\right)v_k. & (k\in \mathbb{N}) \end{cases} $$ Checking the convergence of the sequences $\{u_k\}_{k\in \mathbb{N}}$ and $\{v_k\}_{k\in \mathbb{N}}$.

share|improve this question
    
What have you tried? –  Alex Becker Mar 19 '12 at 22:29
    
Dear Alex Becker. I tried to do it. I guess that both the sequences are not convergent. Thank you for your consideration of my question. –  impartialmale Mar 19 '12 at 22:45

1 Answer 1

up vote 2 down vote accepted

Let $z_k = u_k + i v_k$. The recursion says $z_{k+1} = a_{k+1} z_k$, where $a_k = 1 - 1/k^2 - i/k$. Now $a_k = r_k e^{i\theta_k}$ where $r_k = |a_k| = \sqrt{1 - 1/k^2 + 1/k^4}$ and $\theta_k = -\arctan\left(\frac{1/k}{1 - 1/k^2}\right)$. Thus $z_n = z_0 \prod_{k=1}^n a_k = \left(\prod_{k=1}^n r_k \right) \exp\left(i \sum_{k=1}^n \theta_k\right)$.

We have $r_k = 1 - 1/(2 k^2) + O(1/k^4)$ and $\theta_k = -1/k + O(1/k^3)$ as $k \to \infty$. The infinite product $\prod_{k=1}^\infty r_k$ converges to a nonzero limit $R$, so $|z_n| \to R |z_0|$ as $n \to \infty$, but $\sum_{k=1}^\infty \theta_k$ diverges, so $\arg(z_n)$ diverges, and so do the sequences $u_n$ and $v_n$. In the limit as $n \to \infty$, the points $z_n$ spiral infinitely many times around the circle of radius $R |z_0|$ centred at $0$.

share|improve this answer
    
Thank you Robert Israel for your nice solution. I want to ask you more. If we take a point $\bar{z}$ in the circle of radius $R|z_0|$ centered at $0$, could we choose a subsequence $\{z_{n_i}\}\subset \{z_n\}$ converging to $\bar{z}$. –  impartialmale Mar 20 '12 at 5:08
    
Yes, we could do that. –  Robert Israel Mar 20 '12 at 6:20
    
Dear Robert Israel. Please give your arguments to prove this fact. Thank you. –  impartialmale Mar 20 '12 at 7:38
    
I am still confused. Why $\arg(z_n)$ diverges implies both $\{u_n\}$ and $\{v_n\}$ diverge? Let $z_n=u_n+iv_n=r_n(\cos(\theta_n)+i\sin(\theta_n))=r_ne^{i\theta_n}$. Let $r_n=r$ (constant) and $\theta_n=2n\pi$ for all $n\in \mathbb{N}$. Then $\theta_n$ is not convergent and $u_n=r$, $v_n=0$ for all $n\in\mathbb{N}$. This means that both $\{u_n\}$ and $\{v_n\}$ are convergent. –  impartialmale Mar 20 '12 at 10:01
    
Maybe I should have said $\text{Arg}(z_n)$ (i.e. the version of arg in the interval $(-\pi,\pi]$) diverges. –  Robert Israel Mar 20 '12 at 18:43

Your Answer

 
discard

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.