Before any formula, you need to know what a complex number looks like.
Figure s01 — The point z=x+iy drawn as the red dot at the tip of the blue arrow. The yellow dashed leg is x=Re(z) (go right), the green dashed leg is y=Im(z) (go up); the blue arrow's length is ∣z∣=x2+y2. This is the "plane" every later picture lives on.
Why we need this. Residues live at pointsz0 in this plane, and we integrate along paths drawn on this plane. If z were just a number on a line, there would be no "going around" anything — you need two dimensions to loop.
Several later definitions lean on the phrase "power series", so we pin it down before using it.
Why we need it. The complex exponential (next) is a power series, and "analytic" will mean "locally equal to a power series". So this is the vocabulary those definitions speak.
To "go around" a point we need an angle. The tool that turns "angle" into "point on a circle" is the complex exponential, so we build it first.
Figure s02 — The blue arrow is eiθ, a point on the white dotted unit circle at the yellow angle θ. The green arrow shows the direction of travel: hold ρ fixed and let θ increase from 0 to 2π and the red dot sweeps once, counter-clockwise, around the whole circle. This sweep is the "loop" we integrate over.
Figure s03 — The blue-shaded region is the punctured disk 0<∣z−z0∣<R: f is analytic on the whole ring (green label) but the centre z0 is poked out (red hollow dot). The yellow arrows point inward toward z0 to suggest the values of f shooting off to infinity as you approach the bad point.
A Taylor series (a power series with only powers z0,z1,z2,…) can only describe nice points. To describe a blow-up we also allow negative powers. Those negative powers only make sense in a ring-shaped region, so we first name that ring.
We now compute ∮Cfdz around a small circle of radius ρ enclosing exactly the one singularity z0, step by step.
Step 1 — parametrize the loop. Let C be the circle ∣z−z0∣=ρ traced counter-clockwise:
z=z0+ρeiθ,θ∈[0,2π].
Differentiate this with respect to θ to get the tiny step dz:
dθdz=ρ⋅ieiθ⟹dz=iρeiθdθ.Why: it converts the abstract loop-integral into an ordinary integral in the single real variable θ.
Step 2 — insert the Laurent series and swap sum with integral. Choose ρ with r<ρ<R so the circle lies inside the annulus. There the series converges uniformly, and — as noted in Section 5 — uniform convergence is precisely the theorem that permits swapping the infinite sum and the integral (finite errors stay finite when integrated term by term). Hence:
∮Cfdz=∮C∑n=−∞∞an(z−z0)ndz=∑n=−∞∞an∮C(z−z0)ndz.
Step 3 — evaluate one term. With z−z0=ρeiθ and dz=iρeiθdθ:
∮C(z−z0)ndz=∫02π(ρeiθ)niρeiθdθ=iρn+1∫02πei(n+1)θdθ.Why: we collect all the ρ's and merge the two exponentials einθ⋅eiθ=ei(n+1)θ.
Step 4 — apply the averaging fact with k=n+1. In Section 2's formula the exponent index was k; here k=n+1 (and since n is an integer, so is k). The formula splits into two cases — and we must keep both:
∫02πei(n+1)θdθ={2π0n+1=0(i.e. n=−1)n+1=0(i.e. n=−1).
Feeding each case back into Step 3:
∮C(z−z0)ndz={iρ0⋅2π=2πiiρn+1⋅0=0n=−1n=−1.
Notice the radius ρ vanished from the surviving case (ρn+1=ρ0=1) — the answer does not depend on how big the circle is.
Step 5 — collapse the sum. In the sum of Step 2, every single term is multiplied by 0 except the one with n=−1, whose factor is 2πi. That surviving term carries the coefficient a−1:
∮Cf(z)dz=∑n=−1an⋅0+a−1⋅2πi=2πia−1=2πiRes(f,z0).
So only the a−1 term survives, and the whole loop-integral equals 2πi times that one number — the residue. That is the entire punchline.
The map below reads top-to-bottom as a chain of dependence: each arrow means "you must own the upper box before the lower one is meaningful". The two theorem boxes (Cauchy, Residue) sit at the bottom because they consume everything above.
Cover the right side and test yourself before reading the parent note.
What does z=x+iy look like on paper?
A point / arrow in a 2D plane: x right, y up.
What does ∣z−z0∣<R describe geometrically?
A filled disk of radius R centred at z0.
What does the extra "0<" in 0<∣z−z0∣<R remove?
The centre point z0 itself — a punctured (poked-out) disk.
What is a power series, in one phrase?
An "infinite polynomial" ∑an(z−c)n that converges inside a disk around its centre c.
How is ew defined for complex w?
By the power series ∑n≥0wn/n!, which converges for every complex w.
Why is eiθ=cosθ+isinθ?
Put w=iθ in the series; the real and imaginary groups are exactly the cos and sin series.
In z=z0+ρeiθ, what happens as θ runs 0→2π?
You travel once counter-clockwise around a circle of radius ρ about z0.
What is ∫02πeikθdθ for an integer k?
2π if k=0, otherwise 0 (holds for negative k too).
State the limit definition of f′(z).
limh→0(f(z+h)−f(z))/h, the same value from every direction of approach.
What does "analytic on a region" mean?
Complex-differentiable at every point there, hence locally a convergent power series.
What is an annulus r<∣z−z0∣<R?
The ring/washer of points whose distance from z0 is between the inner radius r and outer radius R.
Name the three isolated-singularity types.
Removable (no negative powers), pole of order m (finitely many), essential (infinitely many).
Why can we integrate the Laurent series term by term?
It converges uniformly on any circle inside its annulus, which licenses swapping sum and integral.
What is an antiderivative F of f?
A function with F′(z)=f(z); a closed loop returns F to its start, so the change is 0.
State Cauchy's Integral Theorem with its hypotheses.
If f is analytic on and inside a positively oriented simple closed piecewise-smooth C, then ∮Cfdz=0.
In the derivation, what is dz when z=z0+ρeiθ?
dz=iρeiθdθ.
Which index maps to the averaging fact, and which survives?
The exponent is k=n+1; only n=−1 (so k=0) survives, giving 2πi; all other terms are multiplied by 0.
State the Residue Theorem with its hypotheses.
For f analytic on and inside a positively oriented simple closed piecewise-smooth C except at isolated singularities zj inside, ∮Cfdz=2πi∑jRes(f,zj).