Skip to content
Tutorials Center
My learning

Topics

Lesson 32 of 36 · Applications and Conformal Maps

The Argument Principle and Rouché's Theorem

Count zeros and poles inside a contour with the argument principle, and locate roots of polynomials with Rouché's theorem, with numerical winding checks.

How many roots does z⁵ + 3z + 1 have inside the unit disc? You could try to find the roots, but a quintic has no formula. The argument principle answers the counting question directly by integrating f'/f around the boundary, and Rouché's theorem turns it into a simple size comparison on the boundary. These tools prove many results in a few lines, and they are the basis of root-counting algorithms.

The argument principle

For a function f, the logarithmic derivative is f'/f. It has simple poles exactly at the zeros and poles of f:

  • If f has a zero of order m at a, then f(z) = (z - a)^m g(z), with g(a) ≠ 0, and f'/f = m/(z - a) + g'/g. The residue of f'/f at a is +m.
  • If f has a pole of order m at a, the residue of f'/f is -m.

Combine this with the residue theorem.

Theorem (argument principle). Let f be meromorphic (analytic except for poles) in a region containing a simple closed counterclockwise contour C and its interior, with no zeros or poles on C. Then

Text
(1 / 2πi) ∮_C f'(z)/f(z) dz = N - P

where N is the number of zeros and P the number of poles of f inside C, each counted with multiplicity.

Why "argument"? Write f(z) = |f| e^(iφ). Then f'/f dz = d(ln f) = d(ln|f|) + i dφ. Around the closed curve, ln|f| returns to its start, so the integral equals i times the total change of arg f along C. Dividing by 2πi gives the number of times the image curve f(C) winds around 0:

Text
N - P = winding number of f(C) around 0

So you can count zeros by watching how many times the curve w = f(z) circles the origin as z goes around C. This is the winding number from the earlier lesson applied to f(C).

flowchart LR
    A["z travels around C"] --> B["w = f(z) draws a curve f(C)"]
    B --> C["Count turns of f(C)<br/>around w = 0"]
    C --> D["Turns = zeros - poles inside C"]

The argument principle: count zeros by counting turns of the image curve.

Example 1

For f(z) = sin z and C: |z| = 4, the zeros inside are 0, ±π (since 2π ≈ 6.28 > 4), each simple, and there are no poles. So (1/2πi) ∮ cos z / sin z dz = 3.

Rouché's theorem

Theorem (Rouché). Let f and g be analytic on and inside a simple closed contour C. If

Text
|g(z)| < |f(z)|   for all z on C

then f and f + g have the same number of zeros inside C (with multiplicity).

Proof idea. Consider h_t = f + t g for t ∈ [0, 1]. On C, |h_t| ≥ |f| - t|g| > 0, so no zero ever crosses C as t changes. The number of zeros N(t) = (1/2πi) ∮ h_t'/h_t dz is an integer that varies continuously with t, hence constant. So N(0) = N(1). ∎

In practice you split a polynomial into a dominant part f and a small perturbation g on a circle, where f is simple enough that you know its zeros.

Example 2: zeros of p(z) = z⁵ + 3z + 1

  • On |z| = 1: take f = 3z, g = z⁵ + 1. Then |f| = 3 and |g| ≤ 2 < 3. So p has the same number of zeros as 3z, in the disc |z| < 1: one zero.
  • On |z| = 2: take f = z⁵, g = 3z + 1. Then |f| = 32 and |g| ≤ 7 < 32. So p has the same number of zeros as z⁵ in |z| < 2: five zeros.

All five zeros lie in |z| < 2, one lies in |z| < 1, so the other four are in the annulus 1 < |z| < 2. (No zeros lie on |z| = 1, since |p| ≥ |3z| - |z⁵ + 1| ≥ 1 > 0 there.)

Example 3: a proof of the fundamental theorem of algebra

For p(z) = a_n z^n + ... + a_0 with a_n ≠ 0, choose R so large that |a_(n-1) z^(n-1) + ... + a_0| < |a_n z^n| on |z| = R. By Rouché, p has the same number of zeros in |z| < R as a_n z^n, namely n. So there are exactly n roots in C.

Try it: count zeros by winding

The code below follows f(z) along the circle |z| = R, unwraps the argument of the image, and divides the total change by 2π. That is the number of zeros inside (for polynomials there are no poles). We apply it to z⁵ + 3z + 1 for radii 0.5, 1, 1.5, 2, and compare with the actual roots found by np.roots.

Python Winding number of the image curve

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

For R = 1 the Rouché count is 1, which matches the counting above (the real root near -0.33). The smallest root has modulus about 0.33, so the count is 0 for R = 0.2.

Try it: the integral of f'/f

Now with the integral itself: (1/2πi) ∮ f'/f dz on circles. We use f(z) = sin z (zeros at multiples of π) and f(z) = (z - 1)²/(z + 2) (a double zero at 1 and a simple pole at -2), so N - P should be 2 - 1 = 1 on a big circle, 2 on a circle containing only 1, and -1 on a circle containing only -2.

Python Counting zeros minus poles with an integral

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

When to use which

  • Argument principle: you want an exact count, or you can compute f(C) numerically and count turns.
  • Rouché: you have a polynomial or an analytic function that is dominated by a simple term on a circle, and you want to localize roots to discs and annuli. Always check the strict inequality |g| < |f| on the entire contour.

When splitting f + g, pick the term that is largest on the contour as f. On a small circle that is usually the low-degree terms; on a big circle it is the leading term.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. The argument principle says (1/2πi)∮ f'/f dz equals

    Zeros contribute +m and poles -m to the residues of f'/f, so the integral gives zeros minus poles.
  2. 2. The image curve f(C) winds twice around 0 and f has no poles. How many zeros does f have inside C?

    The winding number of f(C) around 0 equals N - P = N.
  3. 3. Rouché's theorem requires, on the contour C,

    If |g| < |f| everywhere on C, then f and f + g have the same number of zeros inside C.
  4. 4. z⁴ + 6z + 3 has how many zeros in |z| < 1? (Compare 6z with z⁴ + 3 on |z| = 1.)

    On |z| = 1, |6z| = 6 > 4 ≥ |z⁴ + 3|, so the number of zeros equals that of 6z, which is 1.

Summary

  • (1/2πi) ∮ f'/f dz = N - P: zeros minus poles inside C, with multiplicity.
  • Equivalently, f(C) winds N - P times around 0.
  • Rouché: if |g| < |f| on C, then f and f + g have the same number of zeros inside C.
  • Applications: localizing polynomial roots, and a proof of the fundamental theorem of algebra.

Next: Conformal mappings