1
$\begingroup$

The Argument Principle

Suppose a function $f$ is meromorphic on an open set that contains a circle $C$ and its interior. Further assume that $f$ has no zeroes on $C$ (but may have zeroes in the interior of $C$). Then,

$$\frac{1}{2\pi i} \oint_C \frac{f'(z)}{f(z)} dz \ =$$

the number of zeroes of $f$ in the interior of $C$ (counted with multiplicity) minus the number of poles of $f$ in the interior of $C$ (counted with multiplicity).

Rouche's Theorem

Suppose that $f$ and $g$ are holomorphic functions on a domain and let $C$ be a circle whose interior also lies in the domain. If $|f(z)-g(z)| \lt |f(z)|$ for all $z$ on $C$, then $f$ and $g$ have the same number of zeroes in the interior of $C$, counting multiplicities.

My Question

Why does the Argument Principle require $f$ to have no zeroes on $C$, whereas this requirement is relaxed for Rouche's Theorem?

Thank you.

$\endgroup$
  • $\begingroup$ It should be $|f-g|<|g|$ or $|f-g|<|f|$ on $C$, not $|f-g|<|f|+|g|$ on $C$. $\endgroup$ – Mercy King Jan 3 '13 at 22:26
  • $\begingroup$ If $f$ has a zero on $C$, the (correct) condition in Rouche's theorem can't be satisfied. $\endgroup$ – mrf Jan 3 '13 at 22:35
  • $\begingroup$ @Mercy, thanks, corrected. $\endgroup$ – Conan Wong Jan 3 '13 at 22:54
  • $\begingroup$ @mrf, not sure about your comment. Various sources I've seen do not have this assumption. $\endgroup$ – Conan Wong Jan 3 '13 at 22:55
  • 1
    $\begingroup$ @Mercy: Rouché's theorem also holds with the estimate $|f-g| < |f|+|g|$ on $C$. This is sometimes useful, apart from being nicely symmetric. See Remmert, page 390 $\endgroup$ – Martin Jan 4 '13 at 11:45
3
$\begingroup$

In the Argument Principle: If $f$ has a zero $z_0 \in C$, then $z_0$ is a pole for the quotient $f'/f$. Therefore you cannot define $\int_C f'(z)/f(z)\,dz$.

In Rouche's Theorem: The inequality $|f-g|<|f|$ on $C$ implies that $f$ and $g$ have no zeroes on $C$. In fact, if $g$ (resp. $f$) had a zero $z_0 \in C$, then $|f(z_0)|<|f(z_0)|$ (resp. $|g(z_0)|<0$), which is a contradiction.

$\endgroup$
  • $\begingroup$ Yes, and if there's some roots of $f$ on $C$, as $f$ is not zero everywhere and hence its roots doesn't have accumulation point, you can find a greater $C'$ whose interior contains $int(C)$ and then form the integral $\int_{C'}f'/f$, Am I right ? $\endgroup$ – Fardad Pouran Nov 23 '14 at 19:10

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.