Helpful context:

You want to find the roots of $x^2 + 1 = 0$. In the real numbers, there are none - squaring any real number gives something nonnegative, so $x^2 \geq 0$ always, and $x^2 + 1 \geq 1 > 0$. You could declare “no solution” and move on.

Or you could do what mathematicians did in the 16th century: invent a new number $i = \sqrt{-1}$ and see where it leads.

This sounds like a trick - the kind of move a student makes when they want the answer to come out right. It is not. The invention of $i$ turned out to produce a complete, consistent number system - the complex numbers - in which every polynomial equation has roots (this is the Fundamental Theorem of Algebra). More surprisingly, functions defined on the complex numbers turn out to be far more rigid and beautiful than anything in real analysis. A function that is differentiable once in the complex sense is automatically differentiable infinitely many times, can be expanded as a power series, and is completely determined by its values on any open set - no matter how small that set is. Nothing like this is true in real analysis.

The roots of $x^2 + 1 = 0$ are just the entry point. The theory they unlock is one of the most powerful in all of mathematics.

Section 1: Complex Numbers as Geometry

A complex number is $z = a + bi$ where $a, b \in \mathbb{R}$ and $i^2 = -1$. The number $a$ is the real part, written $\text{Re}(z)$, and $b$ is the imaginary part, written $\text{Im}(z)$.

The first crucial step is to stop thinking of $z$ as an algebraic formula and start thinking of it geometrically. A complex number $z = a + bi$ is the point $(a, b)$ in the plane - the complex plane, also called the Argand plane. The horizontal axis is the real axis; the vertical axis is the imaginary axis.

Under this identification, addition is exactly vector addition:

$$(a + bi) + (c + di) = (a + c) + (b + d)i.$$

Add real parts, add imaginary parts. Draw the vectors and add them tip-to-tail. There is nothing new here - it is the same as adding vectors in $\mathbb{R}^2$.

Multiplication is more interesting. Before writing any formulas, let us build the geometric picture.

The modulus of $z = a + bi$ is $|z| = \sqrt{a^2 + b^2}$ - the distance from the origin to the point $(a, b)$.

The argument of $z$ is $\arg(z) = \theta$ where $\theta$ is the angle from the positive real axis to the vector $(a, b)$, measured counterclockwise.

The complex conjugate of $z = a + bi$ is $\bar{z} = a - bi$. Geometrically, conjugation reflects across the real axis. Algebraically: $z\bar{z} = (a+bi)(a-bi) = a^2 + b^2 = |z|^2$.

Now: when you multiply two complex numbers $z_1$ and $z_2$, what happens geometrically? The answer is:

$$|z_1 z_2| = |z_1||z_2| \qquad \text{and} \qquad \arg(z_1 z_2) = \arg(z_1) + \arg(z_2).$$

Moduli multiply. Arguments add. Multiplying by $z$ scales by $|z|$ and rotates by $\arg(z)$.

Here is the key test of this rule: what does multiplying by $i$ do? The number $i$ has modulus $|i| = \sqrt{0^2 + 1^2} = 1$ and argument $\arg(i) = \pi/2$ (it sits on the positive imaginary axis, at $90°$). So multiplying by $i$ rotates by $90°$ and scales by $1$ - a pure rotation.

Apply $i$ twice: rotate by $90°$, then $90°$ again - total $180°$. Rotating by $180°$ flips the sign. So $i \cdot i = -1$, which is $i^2 = -1$. The defining property of $i$ is just geometry.

Discomfort check. If $i$ is “imaginary,” why are we treating it as a rotation? Because the word “imaginary” is a historical accident - Descartes coined it dismissively to mean “fictitious.” Complex numbers are not imaginary in any philosophical sense. They are pairs of real numbers $(a, b)$ with a specific multiplication rule. That multiplication rule happens to correspond exactly to rotation-and-scaling in the plane. The geometric picture is not a metaphor for the algebra; it is the actual content of the algebra.

Section 2: Polar Form and Euler’s Formula

The geometric picture leads immediately to a cleaner notation. Any complex number $z$ with modulus $r$ and argument $\theta$ can be written as:

$$z = r(\cos\theta + i\sin\theta).$$

This is true from definitions: the point at angle $\theta$ and distance $r$ has coordinates $(r\cos\theta, r\sin\theta)$.

Now consider Taylor series. For any real number $x$:

$$e^x = 1 + x + \frac{x^2}{2!} + \frac{x^3}{3!} + \frac{x^4}{4!} + \cdots$$

$$\cos x = 1 - \frac{x^2}{2!} + \frac{x^4}{4!} - \frac{x^6}{6!} + \cdots$$

$$\sin x = x - \frac{x^3}{3!} + \frac{x^5}{5!} - \frac{x^7}{7!} + \cdots$$

Substitute $x = i\theta$ into the exponential series, using $i^2 = -1$, $i^3 = -i$, $i^4 = 1$, cycling with period 4:

$$e^{i\theta} = 1 + i\theta + \frac{(i\theta)^2}{2!} + \frac{(i\theta)^3}{3!} + \frac{(i\theta)^4}{4!} + \cdots$$

$$= 1 + i\theta - \frac{\theta^2}{2!} - \frac{i\theta^3}{3!} + \frac{\theta^4}{4!} + \frac{i\theta^5}{5!} - \cdots$$

Separate real and imaginary parts:

$$= \left(1 - \frac{\theta^2}{2!} + \frac{\theta^4}{4!} - \cdots\right) + i\left(\theta - \frac{\theta^3}{3!} + \frac{\theta^5}{5!} - \cdots\right) = \cos\theta + i\sin\theta.$$

This is Euler’s formula: $e^{i\theta} = \cos\theta + i\sin\theta$.

Every complex number with modulus $r$ and argument $\theta$ can now be written cleanly as:

$$z = re^{i\theta}.$$

This is polar form. It makes multiplication effortless: $z_1 z_2 = r_1 e^{i\theta_1} \cdot r_2 e^{i\theta_2} = r_1 r_2 e^{i(\theta_1 + \theta_2)}$. The geometric rule (moduli multiply, arguments add) is now the ordinary rule for multiplying exponentials.

Powers are equally clean: $z^n = r^n e^{in\theta}$. Raising to the $n$-th power raises the modulus to the $n$-th power and multiplies the argument by $n$.

At $\theta = \pi$: $e^{i\pi} = \cos\pi + i\sin\pi = -1$. Rearranging:

$$e^{i\pi} + 1 = 0.$$

This is Euler’s identity. It connects five fundamental constants - $e$, $i$, $\pi$, $1$, and $0$ - in a single equation. It is not a coincidence or a curiosity. It is the inevitable result of Euler’s formula evaluated at one specific angle. Its beauty comes from the fact that $\pi$ emerges naturally in the geometry of complex multiplication.

De Moivre’s theorem follows immediately: $(\cos\theta + i\sin\theta)^n = \cos(n\theta) + i\sin(n\theta)$. This is a powerful tool for expressing $\cos(n\theta)$ and $\sin(n\theta)$ in terms of $\cos\theta$ and $\sin\theta$ - expand the left side using the binomial theorem and separate real and imaginary parts.

Section 3: Complex Functions and What They Look Like

A complex function $f: \mathbb{C} \to \mathbb{C}$ takes a complex number and produces a complex number. It is a map from the plane to the plane.

Real functions $f: \mathbb{R} \to \mathbb{R}$ can be drawn as curves in the plane. Complex functions cannot be drawn this way - you would need four dimensions to draw input and output simultaneously. The standard approaches are: (1) draw the input plane and output plane side by side, showing where specific regions go, or (2) draw how the function distorts a grid.

Let us examine the key functions.

$f(z) = z^2$. In polar form: $f(re^{i\theta}) = r^2 e^{2i\theta}$. This squares the modulus and doubles the argument. The right half-plane ($\text{Re}(z) > 0$) maps to all of $\mathbb{C}$ minus the negative real axis. The circle $|z| = r$ maps to the circle $|w| = r^2$. Two points that are reflections of each other across the imaginary axis (same modulus, arguments $\theta$ and $\pi - \theta$) both map to the same output (modulus $r^2$, argument $2\theta$ and $2\pi - 2\theta$ - same after reducing). So $f(z) = z^2$ is two-to-one.

$f(z) = e^z$. Write $z = x + iy$: $e^z = e^x e^{iy} = e^x(\cos y + i\sin y)$. The modulus is $e^x$ and the argument is $y$. A vertical strip $\{a \leq \text{Re}(z) \leq b\}$ maps to the annulus $\{e^a \leq |w| \leq e^b\}$. A horizontal line $\text{Im}(z) = c$ maps to the ray at angle $c$. The exponential wraps vertical strips around the origin - it is periodic in the imaginary direction with period $2\pi$.

$f(z) = 1/z$. In polar form: $(1/r)e^{-i\theta}$. Modulus inverts; argument negates. Points inside the unit circle map outside; points outside map inside. The real line maps to itself; circles through the origin map to lines; circles not through the origin map to other circles. This map is called an inversion - it is one of the fundamental building blocks in conformal geometry.

Section 4: Holomorphic Functions

In real calculus, a function $f: \mathbb{R} \to \mathbb{R}$ is differentiable at $a$ if the limit:

$$f'(a) = \lim_{h \to 0} \frac{f(a+h) - f(a)}{h}$$

exists, where $h$ is a real number approaching $0$ from either side.

For complex functions, the definition looks the same: $f: \mathbb{C} \to \mathbb{C}$ is complex-differentiable (or holomorphic) at $z_0$ if:

$$f'(z_0) = \lim_{h \to 0} \frac{f(z_0 + h) - f(z_0)}{h}$$

exists, where $h$ is a complex number approaching $0$.

The definitions look identical. The difference is profound. When $h$ is real, there are two directions to approach $0$ - left and right. When $h$ is complex, there are infinitely many directions in the plane, and the limit must be the same regardless of which direction $h$ approaches $0$ from.

This is a vastly more stringent condition. Most functions that are smooth as maps from $\mathbb{R}^2$ to $\mathbb{R}^2$ are not complex-differentiable. The function $f(z) = \bar{z}$ (complex conjugate) is differentiable as a map $\mathbb{R}^2 \to \mathbb{R}^2$ but fails complex differentiability at every point.

Discomfort check. How can adding one dimension make differentiability so much harder? When $h$ is real, you need agreement from two directions. When $h$ is complex, you need agreement from every direction. Two constraints vs. infinitely many. The miracle is not that complex differentiability is rare - it is that any functions satisfy it at all, and that those that do turn out to have the extraordinary properties described in the rest of this post.

Examples of holomorphic functions.

  • $f(z) = z^n$: differentiable everywhere, $f'(z) = nz^{n-1}$.
  • $f(z) = e^z$: differentiable everywhere, $f'(z) = e^z$.
  • $f(z) = \sin z = (e^{iz} - e^{-iz})/(2i)$: differentiable everywhere.
  • $f(z) = 1/z$: differentiable everywhere except $z = 0$.

Examples that are NOT holomorphic.

  • $f(z) = \bar{z}$ (complex conjugate): not holomorphic anywhere.
  • $f(z) = |z|^2 = z\bar{z}$: holomorphic only at $z = 0$.
  • $f(z) = \text{Re}(z)$: not holomorphic anywhere.

Section 5: The Cauchy-Riemann Equations

Let $f(z) = u(x, y) + iv(x, y)$ where $z = x + iy$ and $u, v: \mathbb{R}^2 \to \mathbb{R}$ are the real and imaginary parts of $f$.

What does complex differentiability require of $u$ and $v$? Consider the limit along two specific directions.

Horizontal approach: $h = \Delta x$ (real, $\Delta y = 0$). The limit:

$$f'(z_0) = \lim_{\Delta x \to 0} \frac{u(x+\Delta x, y) - u(x,y)}{\Delta x} + i\frac{v(x+\Delta x, y) - v(x,y)}{\Delta x} = \frac{\partial u}{\partial x} + i\frac{\partial v}{\partial x}.$$

Vertical approach: $h = i\Delta y$ (imaginary, $\Delta x = 0$). The limit:

$$f'(z_0) = \lim_{\Delta y \to 0} \frac{u(x, y+\Delta y) - u(x,y)}{i\Delta y} + i\frac{v(x, y+\Delta y) - v(x,y)}{i\Delta y} = \frac{\partial v}{\partial y} - i\frac{\partial u}{\partial y}.$$

For both limits to agree, equate real and imaginary parts:

$$\frac{\partial u}{\partial x} = \frac{\partial v}{\partial y}, \qquad \frac{\partial u}{\partial y} = -\frac{\partial v}{\partial x}.$$

These are the Cauchy-Riemann equations. They are necessary for complex differentiability. Moreover: if $u$ and $v$ have continuous first partial derivatives and satisfy these equations at a point, then $f$ is holomorphic there.

What do they mean geometrically? The Jacobian of the map $(u, v)$ with respect to $(x, y)$ is:

$$J = \begin{pmatrix} \partial u/\partial x & \partial u/\partial y \\ \partial v/\partial x & \partial v/\partial y \end{pmatrix}.$$

The Cauchy-Riemann equations say $\partial u/\partial x = \partial v/\partial y$ and $\partial u/\partial y = -\partial v/\partial x$. Substituting $a = \partial u/\partial x$ and $b = \partial v/\partial x$:

$$J = \begin{pmatrix} a & -b \\ b & a \end{pmatrix}.$$

This is the matrix of a rotation-and-scaling by the complex number $a + bi$. A holomorphic function locally looks like a rotation-and-scaling - it is conformal (angle-preserving). It cannot shear, and it cannot stretch differently in different directions.

This explains why $f(z) = \bar{z}$ fails: conjugation is a reflection, which cannot be written as a rotation-and-scaling.

Discomfort check. The Cauchy-Riemann equations are two equations relating four partial derivatives. Why do two equations encode complex differentiability? Because they constrain the Jacobian from an arbitrary $2 \times 2$ matrix (four degrees of freedom) to a rotation-scaling matrix (two degrees of freedom, since $a$ and $b$ determine everything). The two equations eliminate shearing and non-uniform scaling. What remains is exactly the structure of complex multiplication.

Section 6: The Extraordinary Rigidity of Holomorphic Functions

Holomorphic functions have a property with no analog in real analysis: they are rigid. Knowing a holomorphic function on any open set - no matter how small - determines it everywhere.

Here is the precise statement. Suppose $f$ and $g$ are both holomorphic on a connected region $\Omega$, and $f(z) = g(z)$ for all $z$ in some small disk inside $\Omega$. Then $f(z) = g(z)$ for all $z \in \Omega$. The functions agree everywhere, forced by their agreement on just one small piece.

This is called the identity theorem or analytic continuation. There is no analog in real analysis: two smooth functions can agree on an entire interval and then diverge completely.

Why is complex analysis so rigid? Because holomorphic functions are analytic - they can be expanded as convergent power series around every point in their domain:

$$f(z) = \sum_{n=0}^{\infty} a_n (z - z_0)^n$$

where $a_n = f^{(n)}(z_0)/n!$. The coefficients are determined by the derivatives at $z_0$. The derivatives at $z_0$ are determined by the values of $f$ near $z_0$. So knowing $f$ on any small neighborhood of $z_0$ fixes all the $a_n$, which determines the power series, which determines $f$ everywhere the series converges.

Extend the power series to its full radius of convergence. Then start a new power series from a nearby point - still inside the convergence disk - and extend again. You can reach any point connected to $z_0$ by a path in $\Omega$, and the value at that point is uniquely determined. This process is analytic continuation.

A holomorphic function is like a crystal: once the seed is specified, the rest of the structure is forced. A smooth real function is like clay: you can reshape any part without affecting the rest.

This rigidity is not a limitation - it is power. It means that extending a holomorphic function to a larger domain, if it is possible, can be done in exactly one way.

Section 7: Cauchy’s Integral Theorem

Now we reach the central theorems of complex analysis. They rest on the Cauchy-Riemann equations combined with Green’s theorem.

Setup. Let $C$ be a closed curve in $\mathbb{C}$ - one that starts and ends at the same point. The complex line integral $\oint_C f(z) dz$ is defined by writing $dz = dx + i dy$ and expanding $f = u + iv$:

$$\oint_C f(z) dz = \oint_C (u dx - v dy) + i\oint_C (v dx + u dy).$$

Each piece is an ordinary real line integral.

Theorem (Cauchy’s Integral Theorem). If $f$ is holomorphic on and inside a simple closed curve $C$ (a closed curve that does not cross itself), then:

$$\oint_C f(z) dz = 0.$$

Proof sketch. Apply Green’s theorem to the first real integral: $\oint_C u dx - v dy = \iint_D (-\partial v/\partial x - \partial u/\partial y) dA$. By the second Cauchy-Riemann equation, $\partial u/\partial y = -\partial v/\partial x$, so the integrand is $-\partial v/\partial x + \partial v/\partial x = 0$. The double integral is zero. Similarly for the second integral. So $\oint_C f(z) dz = 0$.

The theorem says holomorphic functions have no “circulation.” If a fluid velocity field satisfied the Cauchy-Riemann equations, you could not build a whirlpool - any closed-loop integral of velocity would be zero. The condition of being holomorphic is exactly the condition of having no vorticity.

Path independence. Cauchy’s theorem implies that in a holomorphic region, the line integral $\int_\gamma f(z) dz$ depends only on the endpoints of $\gamma$, not on which path $\gamma$ takes between them. Any two paths from $z_1$ to $z_2$ staying inside the holomorphic region give the same integral - you can deform the path freely without changing the value.

Section 8: Cauchy’s Integral Formula

Cauchy’s theorem has a spectacular consequence. Let $C$ be a simple closed curve, $f$ holomorphic inside and on $C$, and $z_0$ any point strictly inside $C$. The function $f(z)/(z - z_0)$ has a singularity at $z_0$, so Cauchy’s theorem does not apply directly. But Cauchy’s theorem on a region with a small circle around $z_0$ removed gives:

$$f(z_0) = \frac{1}{2\pi i} \oint_C \frac{f(z)}{z - z_0} dz.$$

This is Cauchy’s integral formula. The value of $f$ at any interior point is completely determined by the values of $f$ on the boundary.

Read that again slowly. If you know $f$ on the circle $C$, you know $f$ at every interior point, without any additional information. This is the rigidity of Section 6 made into a formula. The boundary determines the interior.

This has immediate consequences.

Holomorphic functions have derivatives of all orders. Differentiate Cauchy’s formula $n$ times with respect to $z_0$:

$$f^{(n)}(z_0) = \frac{n!}{2\pi i} \oint_C \frac{f(z)}{(z - z_0)^{n+1}} dz.$$

Every derivative is given by a contour integral. Since the integrals are well-defined, the derivatives exist. A holomorphic function has a first derivative, a second derivative, a third derivative - in fact, derivatives of every order. There is no such result in real analysis: a real function can be differentiable once but not twice.

Liouville’s theorem. A bounded holomorphic function defined on the entire complex plane must be constant. Proof: from the formula for $f'(z_0)$ on a circle of radius $R$, one can show $|f'(z_0)| \leq M/R$ where $M = \sup|f|$. As $R \to \infty$, $|f'(z_0)| \to 0$. So $f'(z_0) = 0$ everywhere, meaning $f$ is constant.

Liouville’s theorem proves the Fundamental Theorem of Algebra: every polynomial of degree $\geq 1$ with complex coefficients has a complex root. The proof: if $p(z)$ has no root, then $1/p(z)$ is a bounded holomorphic function on $\mathbb{C}$ (it is bounded because $|p(z)| \to \infty$ as $|z| \to \infty$), so by Liouville’s theorem it is constant - but $1/p(z)$ is not constant, contradiction.

Section 9: Singularities and the Laurent Series

So far we have required $f$ to be holomorphic everywhere inside $C$. What happens at isolated singularities - points where $f$ is not defined?

Near an isolated singularity $z_0$, $f$ can be expanded in a Laurent series - a power series allowing negative powers:

$$f(z) = \cdots + \frac{a_{-2}}{(z-z_0)^2} + \frac{a_{-1}}{z - z_0} + a_0 + a_1(z-z_0) + a_2(z-z_0)^2 + \cdots$$

The terms with negative powers form the principal part. The type of singularity is determined by how many negative-power terms appear.

Removable singularity. The principal part is zero - no negative powers. The singularity is an artifact of the formula; $f$ can be defined at $z_0$ to make it holomorphic. Example: $\sin(z)/z$ at $z = 0$.

Pole of order $n$. The principal part has finitely many terms, the lowest being $a_{-n}/(z-z_0)^n$ with $a_{-n} \neq 0$. The function blows up at $z_0$, but in a controlled way: $|f(z)| \to \infty$ as $z \to z_0$. Examples: $1/z$ has a simple pole (order 1); $1/z^2$ has a pole of order 2.

Essential singularity. The principal part has infinitely many terms. Near an essential singularity, $f$ takes every complex value (with at most one exception) in any neighborhood of $z_0$. This is the Great Picard theorem. Example: $e^{1/z}$ at $z = 0$.

The residue at $z_0$ is the coefficient $a_{-1}$ in the Laurent expansion - the coefficient of $(z - z_0)^{-1}$.

Computing residues at simple poles. If $z_0$ is a simple pole:

$$\text{Res}(f, z_0) = \lim_{z \to z_0} (z - z_0) f(z).$$

If $f(z) = g(z)/h(z)$ where $h(z_0) = 0$, $h'(z_0) \neq 0$, and $g(z_0) \neq 0$:

$$\text{Res}(f, z_0) = \frac{g(z_0)}{h'(z_0)}.$$

Computing residues at poles of order $n$.

$$\text{Res}(f, z_0) = \frac{1}{(n-1)!} \lim_{z \to z_0} \frac{d^{n-1}}{dz^{n-1}} \left[(z-z_0)^n f(z)\right].$$

Discomfort check. How to compute residues at higher-order poles? For a simple pole, multiply by $(z - z_0)$ and take the limit. For a double pole, multiply by $(z-z_0)^2$, differentiate once, take the limit. For a triple pole, multiply by $(z-z_0)^3$, differentiate twice, divide by $2!$, take the limit. The pattern is systematic: multiply to clear the pole, differentiate $(n-1)$ times to isolate the $a_{-1}$ term, then take the limit.

Section 10: The Residue Theorem

Theorem (Residue Theorem). If $f$ is holomorphic inside and on a simple closed curve $C$ except at isolated singularities $z_1, z_2, \ldots, z_n$ inside $C$, then:

$$\oint_C f(z) dz = 2\pi i \sum_{k=1}^{n} \text{Res}(f, z_k).$$

The integral around $C$ is $2\pi i$ times the sum of residues of all singularities inside $C$.

The intuition: by Cauchy’s theorem, you can deform the contour $C$ down to small circles around each singularity. The integral around each small circle $\gamma_k$ is $2\pi i \cdot \text{Res}(f, z_k)$ - this follows from integrating the Laurent series term by term and using $\oint (z-z_k)^n dz = 0$ for $n \neq -1$ and $= 2\pi i$ for $n = -1$. Summing over all singularities gives the theorem.

The residue theorem reduces contour integrals to algebra. Instead of computing a complicated path integral directly, you: (1) identify all singularities inside $C$, (2) compute the residue at each, (3) sum and multiply by $2\pi i$.

Section 11: Real Integrals via Contour Integration

The residue theorem’s most powerful application is evaluating definite real integrals that resist standard calculus techniques.

Example 1. Compute $\displaystyle\int_{-\infty}^{\infty} \frac{1}{1 + x^2} dx$.

Form the complex function $f(z) = 1/(1+z^2) = 1/[(z-i)(z+i)]$. Integrate along the contour: real axis from $-R$ to $R$, then the upper semicircle of radius $R$ back to $-R$. The full contour integral equals $2\pi i \cdot \text{Res}(f, i)$ (only $z = i$ is in the upper half-plane).

$$\text{Res}(f, i) = \lim_{z \to i} (z - i)\frac{1}{(z-i)(z+i)} = \frac{1}{2i}.$$

Contour integral $= 2\pi i \cdot \frac{1}{2i} = \pi$.

As $R \to \infty$, the semicircle contribution vanishes (integrand $\sim 1/R^2$, arc length $\sim \pi R$, product $\to 0$). So the real-axis integral equals $\pi$:

$$\int_{-\infty}^{\infty} \frac{1}{1+x^2} dx = \pi.$$

Example 2. Compute $\displaystyle\int_{-\infty}^{\infty} \frac{\cos x}{1 + x^2} dx$.

No elementary antiderivative exists. Form $f(z) = e^{iz}/(1+z^2)$. On the upper semicircle, $|e^{iz}| = e^{-\text{Im}(z)} \leq 1$, so the semicircle integral still vanishes by Jordan’s lemma.

$$\text{Res}(f, i) = \lim_{z \to i}(z-i)\frac{e^{iz}}{(z-i)(z+i)} = \frac{e^{i \cdot i}}{2i} = \frac{e^{-1}}{2i}.$$

Contour integral $= 2\pi i \cdot e^{-1}/(2i) = \pi/e$.

Taking real parts: $\displaystyle\int_{-\infty}^{\infty} \frac{\cos x}{1+x^2} dx = \frac{\pi}{e}$.

The method is always the same: replace the real integral with a contour integral in the complex plane, close the contour with a curve on which the extra integral vanishes, apply the residue theorem.

Section 12: Connection to Fourier Analysis

The link between complex analysis and Fourier analysis is direct and essential.

The Laplace transform of a function $f(t)$ is:

$$F(s) = \int_0^{\infty} f(t) e^{-st} dt, \qquad s \in \mathbb{C}.$$

This is a complex function of $s$. The Fourier transform is the Laplace transform restricted to the imaginary axis $s = i\omega$:

$$\hat{f}(\omega) = F(i\omega) = \int_0^{\infty} f(t) e^{-i\omega t} dt.$$

The poles of $F(s)$ in the complex plane encode the behavior of $f(t)$. A pole at $s = \sigma + i\omega_0$ contributes a term proportional to $e^{\sigma t} e^{i\omega_0 t}$ to the inverse transform. If $\sigma < 0$ (pole in the left half-plane), the contribution decays - the system is stable. If $\sigma > 0$ (right half-plane), it grows - unstable. The imaginary part $\omega_0$ is the oscillation frequency.

This is why engineers draw pole-zero plots in the complex $s$-plane: the location of poles determines both the frequencies and the stability of the corresponding time-domain signal. A pole on the imaginary axis oscillates without decaying; a pole far to the left decays rapidly; a pole near the imaginary axis decays slowly, producing persistent oscillation.

The inverse Fourier transform and inverse Laplace transform are contour integrals. Evaluating them uses the residue theorem. The theory of Fourier analysis and the theory of complex analysis are not separate subjects - they are the same subject, viewed from different angles.

Section 13: Harmonic Functions

A function $u: \mathbb{R}^2 \to \mathbb{R}$ is harmonic if it satisfies Laplace’s equation:

$$\frac{\partial^2 u}{\partial x^2} + \frac{\partial^2 u}{\partial y^2} = 0.$$

Harmonic functions are everywhere in physics: the electrostatic potential in a charge-free region, the steady-state temperature distribution in a heat-conducting body, the velocity potential of an irrotational fluid flow. These are all governed by Laplace’s equation.

The connection to complex analysis: the real and imaginary parts of any holomorphic function are both harmonic.

Proof: if $f = u + iv$ is holomorphic, the Cauchy-Riemann equations give $u_x = v_y$ and $u_y = -v_x$. Differentiating the first by $x$ and the second by $y$:

$$u_{xx} = v_{yx} \qquad \text{and} \qquad u_{yy} = -v_{xy}.$$

Since $v_{xy} = v_{yx}$ (mixed partials are equal), $u_{xx} + u_{yy} = v_{yx} - v_{xy} = 0$. So $u$ satisfies Laplace’s equation. Similarly for $v$.

The converse is nearly true: given any harmonic $u$ on a simply connected domain, there exists a harmonic $v$ (its harmonic conjugate) such that $f = u + iv$ is holomorphic. Finding $v$ from $u$ uses the Cauchy-Riemann equations as a system of PDEs.

This gives a systematic way to solve boundary value problems for Laplace’s equation: find a holomorphic function whose real part satisfies the given boundary conditions. The toolkit of complex analysis (conformal maps, Cauchy’s formula) becomes a toolkit for solving Laplace’s equation - which is why complex analysis is so important in classical physics.

Discomfort check. If the real and imaginary parts of any holomorphic function are harmonic, does that mean all harmonic functions come from complex analysis? On simply connected domains, yes - every harmonic function is the real part of some holomorphic function. On domains with holes (like an annulus), this fails: $u(x,y) = \ln\sqrt{x^2+y^2}$ is harmonic on $\mathbb{C} \setminus \{0\}$ but has no harmonic conjugate on that domain (the issue is that going around the origin changes the conjugate by $2\pi$). This is the connection between the topology of the domain and complex analysis.

Section 14: What Holomorphic Functions Cannot Do

Collect the constraints. A holomorphic function on a connected open set:

  • Cannot be bounded unless it is constant (Liouville’s theorem).
  • Cannot have a local maximum of $|f(z)|$ at an interior point (maximum modulus principle: $|f(z)|$ attains its maximum on the boundary).
  • Cannot be real-valued unless it is constant (if $\text{Im}(f) = 0$ everywhere, the Cauchy-Riemann equations force the function to be constant).
  • Cannot vanish on an open set unless it is identically zero (identity theorem).
  • Cannot have two different holomorphic extensions - analytic continuation is unique.

These are not individual constraints - they are facets of a single underlying rigidity. The Cauchy-Riemann equations are simultaneously (1) a constraint on how the function varies, (2) a guarantee that integrals around closed curves are zero, and (3) a mechanism by which boundary data determines interior data.

The maximum modulus principle deserves emphasis. In real analysis, a continuous function on a closed bounded interval attains its maximum somewhere in the interior. For holomorphic functions: if $|f(z)|$ attains a local maximum at an interior point $z_0$, then $f$ is constant in a neighborhood of $z_0$, hence (by the identity theorem) constant everywhere. So $|f|$ can only achieve its global maximum on the boundary of the domain.

This is geometrically natural: a non-constant holomorphic function locally looks like a rotation-and-scaling. It maps small disks to small distorted regions. The center of a disk cannot be the furthest-from-origin point of its image - the image wraps around and goes further out somewhere on the boundary.

Section 15: Further Directions

The ideas developed here are the foundation of a vast field. Some directions to explore next:

The complex logarithm. Since $e^z$ is periodic with period $2\pi i$, its inverse is multi-valued. The complex logarithm is $\log z = \ln|z| + i\arg(z)$, but $\arg(z)$ is only defined up to multiples of $2\pi$. The “principal branch” $\text{Log}(z) = \ln|z| + i\text{Arg}(z)$ with $\text{Arg}(z) \in (-\pi, \pi]$ is a single-valued function - but it has a branch cut along the negative real axis. Cross the cut and you jump to a different branch. The logarithm and its branch structure are the entry point to Riemann surfaces.

Some other extensions:

Conformal mapping. Holomorphic bijections (one-to-one holomorphic functions with holomorphic inverses) preserve angles. Given a complicated domain (say, a region inside a curved boundary), you can map it conformally to a simple domain (the unit disk or the upper half-plane), solve whatever problem you have in the simple domain, then map back. This technique solves Laplace’s equation on complicated domains - modeling fluid flow, electrostatics, and heat conduction in geometry that would otherwise be intractable.

The Riemann mapping theorem. Any simply connected proper open subset of $\mathbb{C}$ can be conformally mapped to the unit disk. This means, from the perspective of complex analysis, all simply connected regions look the same.

Riemann surfaces. The function $\sqrt{z}$ is multi-valued: $\sqrt{1} = 1$ or $-1$. The logarithm $\log z$ has infinitely many values. On the complex plane, these are not functions. Riemann surfaces are geometric objects on which multi-valued functions become single-valued. The Riemann surface of $\sqrt{z}$ is a two-sheeted cover of $\mathbb{C}$, where going around the origin once takes you to the other sheet.

Analytic number theory. The Riemann zeta function $\zeta(s) = \sum_{n=1}^{\infty} n^{-s}$ converges for $\text{Re}(s) > 1$. By analytic continuation, it extends to almost all of $\mathbb{C}$ - this uses the rigidity of complex analysis. The Riemann hypothesis states that all nontrivial zeros of $\zeta$ lie on the line $\text{Re}(s) = 1/2$. Its truth (still unproven) would resolve the deepest questions about the distribution of prime numbers. Complex analysis is the only language in which this can even be stated.

Summary

Concept Content
Complex number $z = a + bi$ Point $(a,b)$ in the plane; modulus $= \sqrt{a^2+b^2}$, argument $=$ angle from real axis
Multiplication Moduli multiply; arguments add; multiplying by $i$ rotates $90°$
Polar form $z = re^{i\theta}$ Euler’s formula: $e^{i\theta} = \cos\theta + i\sin\theta$
Euler’s identity $e^{i\pi} + 1 = 0$
Holomorphic at $z_0$ $\lim_{h\to 0}(f(z_0+h)-f(z_0))/h$ exists for all complex $h \to 0$
Cauchy-Riemann $\partial u/\partial x = \partial v/\partial y$; $\partial u/\partial y = -\partial v/\partial x$; necessary and sufficient
Conformal Holomorphic maps locally preserve angles; Jacobian is rotation-scaling
Rigidity Holomorphic $f$ known on any open set $\Rightarrow$ determined everywhere
Cauchy’s theorem $\oint_C f dz = 0$ when $f$ holomorphic inside $C$
Cauchy’s formula $f(z_0) = \frac{1}{2\pi i}\oint_C \frac{f(z)}{z-z_0} dz$; interior from boundary
Infinite differentiability Holomorphic once $\Rightarrow$ differentiable all orders; each given by contour integral
Laurent series Power series with negative powers; residue $=$ coefficient of $(z-z_0)^{-1}$
Residue theorem $\oint_C f dz = 2\pi i \sum \text{Res}(f, z_k)$; integral from singularities
Real integrals Close contour in upper half-plane; residue theorem gives definite integral
Harmonic functions $u_{xx} + u_{yy} = 0$; real and imaginary parts of holomorphic functions
Maximum modulus principle $
Conformal mapping Holomorphic bijections preserve angles; maps complicated domains to simple ones
Analytic continuation Unique extension of holomorphic function beyond initial domain

The three theorems - Cauchy’s integral theorem, Cauchy’s integral formula, and the residue theorem - form a hierarchy. The first says holomorphic functions have zero circulation. The second says the value at any interior point is determined by boundary values. The third computes contour integrals through singularities from local algebraic data (residues). Each theorem is a consequence of the one before it, and all three ultimately rest on the Cauchy-Riemann equations and Green’s theorem.

If you understand why $\oint_C f(z) dz = 0$ for holomorphic $f$, you have the conceptual key. The rest is elaboration.

Working Through the Cauchy-Riemann Equations: Examples

It is worth checking a few functions explicitly to make the Cauchy-Riemann equations concrete.

Example 1: $f(z) = z^2$. Write $f(x+iy) = (x+iy)^2 = x^2 - y^2 + 2xyi$. So $u = x^2 - y^2$ and $v = 2xy$.

Check: $\partial u/\partial x = 2x$ and $\partial v/\partial y = 2x$. Equal. $\partial u/\partial y = -2y$ and $-\partial v/\partial x = -2y$. Equal. Cauchy-Riemann satisfied everywhere. So $z^2$ is holomorphic on all of $\mathbb{C}$, as expected.

Example 2: $f(z) = \bar{z}$. Write $f(x+iy) = x - iy$. So $u = x$ and $v = -y$.

Check: $\partial u/\partial x = 1$ and $\partial v/\partial y = -1$. Not equal. Cauchy-Riemann fails everywhere. Complex conjugation is not holomorphic, anywhere.

Example 3: $f(z) = |z|^2$. Write $f(x+iy) = x^2 + y^2$. So $u = x^2+y^2$ and $v = 0$.

Check: $\partial u/\partial x = 2x$ and $\partial v/\partial y = 0$. These are equal only when $x = 0$. $\partial u/\partial y = 2y$ and $-\partial v/\partial x = 0$. These are equal only when $y = 0$. Cauchy-Riemann is satisfied only at $z = 0$. So $|z|^2$ is differentiable at exactly one point. (It is differentiable at $z = 0$ because $f'(0) = 0$ can be verified directly.)

This last example is striking: a smooth function that is differentiable at exactly one point in the complex sense. No such phenomenon occurs in real analysis - if a function is differentiable at one real point, it is differentiable in a neighborhood.

Read next: