Skip to content
Tutorials Center
My learning

Topics

Lesson 23 of 36 · Cauchy Theorems

The Cauchy-Goursat Theorem

Learn Cauchy's theorem: closed contour integrals of analytic functions on simply connected domains are zero. Includes deformation of contours and numerical checks.

Here is the central theorem of complex analysis. Roughly: if a function is analytic everywhere inside and on a closed curve, then its integral around that curve is zero. Everything in the rest of the course (the integral formula, power series, residues) is a consequence or a refinement of this statement.

In earlier lessons you saw hints: z², e^z and cos z all integrated to zero around closed curves, while conj(z) and 1/z did not. Now we state exactly why.

Statement

Cauchy-Goursat theorem. Let f be analytic on a simply connected domain D (a domain with no holes), and let C be a closed contour in D. Then

Text
∮_C f(z) dz = 0

Some vocabulary:

  • Analytic (or holomorphic) on D: complex differentiable at every point of D.
  • Simply connected: every closed curve in D can be shrunk to a point without leaving D. A disc, a half-plane and the whole plane are simply connected. A punctured disc or an annulus is not: it has a hole.

The word Goursat refers to the French mathematician Édouard Goursat, who proved the theorem without assuming that f' is continuous. Cauchy's original proof needed that extra assumption (it uses Green's theorem). It turns out never to be needed: a function that is complex differentiable everywhere on a domain automatically has a continuous derivative, and in fact infinitely many derivatives. You will see why in the derivative formula lesson.

Why you should believe it

For f = u + iv and dz = dx + i dy, the integral splits into two real line integrals:

Text
∮ f dz = ∮ (u dx - v dy) + i ∮ (v dx + u dy)

By Green's theorem each real integral equals a double integral over the enclosed region of (-v_x - u_y) and (u_x - v_y). The Cauchy-Riemann equations say exactly that u_x = v_y and u_y = -v_x, so both integrands are zero.

Goursat's own proof starts with a triangle, cuts it into four smaller triangles again and again, and shows that the integral over the original triangle is smaller than any positive number, using only the definition of the derivative. Then polygons are built from triangles, and curves are approximated by polygons.

Deformation of contours

A powerful consequence: you may deform a contour without changing the integral, as long as you do not cross a point where the function fails to be analytic.

Suppose f is analytic in a region except at a point a, and C₁ and C₂ are two counterclockwise simple closed curves that both enclose a, with C₂ inside C₁. Then the region between them has f analytic throughout. Cutting the region and applying the theorem gives

Text
∮_(C₁) f dz = ∮_(C₂) f dz
flowchart TD
    A["f analytic except at a"] --> B["Two loops around a"]
    B --> C["Integrals over both loops are equal"]
    C --> D["Shrink the inner loop to a tiny circle<br/>and compute there"]

Deformation: replace an awkward contour by a small circle around the singularity.

For f(z) = 1/z this explains why the integral is 2πi around the unit circle, a square, an ellipse, or any simple closed curve around 0: all can be shrunk to a small circle around 0 where we computed it directly in winding numbers.

If the curve does not enclose the singularity, then f is analytic inside it and Cauchy-Goursat gives 0.

The theorem needs analyticity at every point inside the curve, not only on the curve. The function 1/z is perfectly analytic on the unit circle, yet the integral is 2πi, because the point 0 inside is a singularity.

Try it: analytic or not?

We integrate six functions around the square with corners ±1 ± i, traversed counterclockwise. The first four are analytic on the entire plane, so the answers should be zero (up to rounding). The last two, conj(z) and |z|², are not analytic anywhere (except |z|² at the origin), so zero is not guaranteed. For the square with side 2, conj(z) gives 8i, which is 2i times the enclosed area 4. The integral of |z|² happens to vanish here only because the square is symmetric about the origin.

Python Closed integrals around a square

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.

Try it: deformation in action

Now f(z) = 1/z. We integrate around a circle, a square, an ellipse, and a lopsided curve around 0 (all give 2πi), and around a circle that does not contain 0 (gives 0). The curves are given as parametrizations over 0 ≤ t ≤ 2π, using the trapezoid rule, which is spectrally accurate for smooth periodic integrands.

Python Deforming the contour around a singularity

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.

The "lopsided" curve is shifted by 0.1 and has radius between 0.6 and 1.4, so it still encloses 0 (its distance from 0 is at least about 0.5).

How to use the theorem

  1. Check that f is analytic on and inside the contour. If so, the integral is 0: done.
  2. If f has singularities inside, deform the contour to small circles around them (see the next lessons).
  3. If you can find an antiderivative on a simply connected domain, you do not even need the theorem: it follows from path independence. In fact, on a simply connected domain, Cauchy-Goursat guarantees that every analytic function has an antiderivative.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. Which condition is needed for ∮_C f dz = 0 by the Cauchy-Goursat theorem?

    Analyticity on a simply connected domain that contains the closed contour is the hypothesis.
  2. 2. Why is ∮_(|z|=1) dz/z = 2πi although 1/z is analytic on the circle?

    The theorem needs analyticity everywhere inside the curve; 1/z fails at the center.
  3. 3. Two counterclockwise simple closed curves enclose the same single singularity of f and f is analytic elsewhere. Their integrals are

    The region between the curves has no singularity, so the contour can be deformed without changing the integral.
  4. 4. What did Goursat's version of the theorem remove from Cauchy's original assumptions?

    Goursat proved the theorem with no assumption that the derivative is continuous.

Summary

  • If f is analytic on a simply connected domain, then ∮_C f dz = 0 for every closed contour C in it.
  • The hypothesis includes every point inside C, not only the curve itself.
  • Contours may be deformed freely across regions where f is analytic; a singularity blocks the deformation.
  • Closed integrals of non-analytic functions such as conj(z) need not vanish.

Next: Cauchy's integral formula