Let $P(z)=az-b$ and $Q(z)=cz-d$, where $a,b,c,d$ are nonzero complex numbers such that $bc\neq ad$.

Suppose that $|\frac{P(z)}{Q(z)}|=1$ whenever $|z|=1$. Show that

$$b\bar{d}=a\bar{c}\hspace{3mm}\text{and} \hspace{3mm} a\bar{b}=c\bar{d}$$

and deduce that

$$\frac{P(z)}{Q(z)}=\omega \frac{z-\alpha}{1-\bar{\alpha}z}$$

where $\alpha=\frac{b}{a}$ and $\omega$ is a complex number with modulus $1$.

$\textbf{My attempt so far:}$

Since $|\frac{P(z)}{Q(z)}|=1$ whenever $|z|=1$, then

$$|az-b|^2=|cz-d|^2 \rightarrow (az-b)\overline{(az-b)}=(cz-d)\overline{(cz-d)}$$

$$\rightarrow (az-b)(\bar{a}\bar{z}-\bar{b})=(cz-d)(\bar{c}\bar{z}-\bar{d})$$

$$\rightarrow a\bar{a}|z|^2-a\bar{b}z-\bar{a}b\bar{z}+b\bar{b}=c\bar{c}|z|^2-c\bar{d}z-\bar{c}d\bar{z}+d\bar{d}$$

$$\rightarrow |a|^2 -(a\bar{b}-c\bar{d})z+|b|^2=|c|^2-(\bar{c}d-\bar{a}b)\bar{z}+|d|^2$$

I'm stuck at this point. Am I on the right track, or is my method of approach wrong on a whole?

  • $\begingroup$ What happened to the $\bar{z}$s? $\endgroup$ – Chappers May 4 '18 at 15:36
  • $\begingroup$ @Chappers Ah alright I see the mistake. Allow me to rectify it. $\endgroup$ – Stoner May 4 '18 at 15:54
  • 1
    $\begingroup$ For alternative approaches, see here and here. $\endgroup$ – dxiv May 4 '18 at 19:11
  • $\begingroup$ @dxiv Thanks! I'll look them up! $\endgroup$ – Stoner May 5 '18 at 2:20

$\quad |a|^2 -(a\bar{b}-c\bar{d})z+|b|^2=|c|^2-(\bar{c}d-\bar{a}b)\bar{z}+|d|^2$

Almost done at this point, now multiply with $\,z\ne 0\,$ and use that $\,z \bar z = |z|^2 = 1\,$ one more time:

$$(a\bar{b}-c\bar{d})z^2- \left(|a|^2+|b|^2-|c|^2-|d|^2\right)z +(\bar{a}b-\bar{c}d) = 0$$

The latter quadratic has every $\,z\,$ on the unit circle as a root, thus infinitely many roots, and therefore must be the zero polynomial. So, in particular, $\,a\bar{b}-c\bar{d} = 0 \iff a \bar b = c \bar d\,$.

For the other equality, note that if $\,f(z) = \dfrac{az-b}{cz-d}\,$ preserves the unit circle, then so does its inverse $\,f^{-1}(z)=\dfrac{dz-b}{cz-a}\,$. Applying the argument above to $\,f^{-1}\,$ gives $\,d\bar b = c \bar a \iff a \bar c = b \bar d\,$.

  • $\begingroup$ Awesome thanks for the clarification! Just a few questions on my end - how did you derive the inverse in the last line above? $\endgroup$ – Stoner May 5 '18 at 2:20
  • 1
    $\begingroup$ @Stoner Let $f(z) = w = \dfrac{az-b}{cz-d}\,$, then $\,w(cz-d)=az-b \iff (cw-a)z=dw-b\,$, which means that $\,z= f^{-1}(w)=\dfrac{dw-b}{cw-a}\,$. Now replace $\,w \mapsto z\,$ to write it in the standard notation as $\,f^{-1}(z)=\dfrac{dz-b}{cz-a}\,$. $\endgroup$ – dxiv May 5 '18 at 2:37
  • 1
    $\begingroup$ Thank you for the clarification. I managed to get the next part of the question as a consequence of the above result! :) $\endgroup$ – Stoner May 5 '18 at 3:53

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.