0
$\begingroup$

I have the complex-valued function

\begin{equation} f(z) = \frac{1}{1 + a(1 - z)}, \quad 0 < a < 1, ~ |z| \le 1. \end{equation}

For $|z| \le 1$ we have $|f(z)| \le 1$. Further, $z = f(z)$ has a single root in $|z| \le 1$, namely $z = 1$.

Define the sequence $z_n = f(z_{n - 1}), ~ n = 1,2,\ldots$ and set $|z_0| \le 1$. I want to show that $\lim_{n \to \infty} z_n = 1$.

Could anyone point me to some references? This should be easy, but I am not well-versed in complex-analysis nor in fixed-point iterations.

$\endgroup$
1
$\begingroup$

HINT.- First at all you have to prove that $$\left|\frac{a}{(1+a(1-z))(1+a(1-w))}\right|\le K\lt 1$$ for $f(z)$ be a contraction, i.e. $|f(z)-f(w)|\le K|z-w|$. The corresponding fixed point in the complet metric space $|z|\le 1$ is the limit of iterated $fofof....fof(z_0)$ for any arbitrary $z_0$ in the unit ball $|z|\le 1$.

Since $z_1=f(z_0),z_2=f(z_1),....,z_n=f(z_{n-1})$ one has $$|z_2-z_1|\le K|z_1-z_0|\\|z_3-z_2|\le K|z_2-z_1|\le K^2|z_1-z_0|\\.....\\|z_{n+1}-z_{n}|\le K^n|z_1-z_0|$$ It follows the Cauchy sequence $\{z_n\}$ because $$|z_{n+p}-z_n|\le |z_{n+p}-z_{n+p-1}|+|z_{n+p-1}-z_{n+p-2}|+.....+|z_{n+1}-z_n|$$ $$|z_{n+p}-z_n|\le (K^{p-1}+K^{p-2}+.....+K+1)K^n|z_1-z_0| \le\frac{K^n}{1-K}|z_1-z_0|$$ (we know that $\sum_{k=0}^{k=\infty}K^i=\frac{1}{1-K}$ and $K^n$ tends to $0$).

Hence $\{z_n\}$ has a limit $z$ and, because $f$ is continuous, if $z_n\to z$ then $f(z_n)\to f(z)$.

On the other hand, the uniqueness of this fixed point is clear because if $z$ and $w$ were distinct fixed points then one has the contradiction $$|z-w|\le K|z-w|\lt |z-w|$$ We finish with the fixed point $z=1$ having in hand and because of the uniqueness.

$\endgroup$
  • $\begingroup$ @Ritz: The proof of $K\lt 1$ is not difficult! $\endgroup$ – Piquito Sep 12 '16 at 13:06
  • $\begingroup$ Thanks for this. I am still working through it. For the first step, aren't we trying to show that for $w,z$ with $|w|,|z| \le 1$ and $w \neq z$ that $| \ldots | \le K < 1$ instead of $| \ldots | = K < 1$? In the first case, I find that $K = a$. $\endgroup$ – Ritz Sep 12 '16 at 13:19
  • $\begingroup$ Tha's right. It was a typo (Take the maximun). $\endgroup$ – Piquito Sep 12 '16 at 13:41
0
$\begingroup$

$\displaystyle g(z):=\frac{1}{1+a(1-z)}-1$

$\displaystyle \frac{g(z)}{g'(\Re(z))}=-(1-z)\frac{(1+a(1-\Re(z)))^2}{1+a(1-z)}$

Newton-Fourier method (https://en.wikipedia.org/wiki/Newton%27s_method below)

for $g(z)=0$: $\displaystyle \,\,z_{n+1}=z_n-\frac{g(z_n)}{g'(\Re(z_n))}$

Therefore $z=1$ or $\displaystyle \Re(z)=1+\frac{1}{a}$ and with $|z|\leq 1$ follows $z=1$.

$\endgroup$

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.