Analytic Functions and the Cauchy-Riemann Equations

A complex function $f (z) = u (x, y) + i v (x, y)$ is analytic (holomorphic) on an open set if it is complex-differentiable at every point there. Complex differentiability is dramatically stronger than real differentiability: it forces the Cauchy–Riemann equations $u_{x} = v_{y}$ and $u_{y} = - v_{x}$ , which (with continuity of partials) are equivalent to analyticity. Analyticity makes $u$ and $v$ harmonic, the function infinitely differentiable, and locally representable by a convergent power series. Used throughout fluid dynamics, electrostatics, signal processing, and number theory. Named for Augustin-Louis Cauchy and Georg Friedrich Bernhard Riemann (1820s–1850s).

A function of a complex variable is analytic at a point if the limit defining its derivative, $f^{'} (z_{0}) = lim_{h \to 0} \frac{f ( z _{0} + h ) - f ( z _{0} )}{h}$ , exists — where $h$ is a complex number free to approach $0$ from any direction. Because the same limit must be obtained along the real and imaginary axes, the real and imaginary parts $u, v$ are rigidly coupled by the Cauchy–Riemann equations. This is why analytic functions are so much rarer and so much better behaved than differentiable functions of two real variables. The surprising fact: complex-differentiability once on an open set implies differentiability infinitely often, plus a host of miracles (Cauchy's integral formula, the maximum principle, identity theorem) absent from real analysis.

The Core Insight

A complex derivative is a single complex number that must describe the local behaviour of $f$ regardless of the direction in which $h \to 0$ . Multiplication by a complex number is a rotation-plus-scaling. So $f$ is analytic exactly when its real Jacobian acts locally as a rotation-scaling — i.e. it is conformal (angle-preserving) wherever $f^{'} \neq = 0$ . The Cauchy–Riemann equations are precisely the algebraic condition that the $2 \times 2$ Jacobian has this special form.

Formal Statement

Complex Differentiability

Let $f$ be defined on an open set $Ω \subseteq C$ and let $z_{0} \in Ω$ . We say $f$ is complex-differentiable at $z_{0}$ if

f^{'} (z_{0}) = h \to 0 lim \frac{f ( z _{0} + h ) - f ( z _{0} )}{h}

exists as a finite complex number, where $h \in C$ , $h \neq = 0$ .

Here $z = x + i y$ with $x, y \in R$ and $i^{2} = - 1$ ; $h = Δ x + i Δ y$ ; $f (z) = u (x, y) + i v (x, y)$ with $u, v : Ω \to R$ . The notation $u_{x} = \partial u / \partial x$ , etc.

⟳The novelty over the real definition is that

h

ranges over a 2-D set; "the limit exists" silently demands all directional approaches agree.

Analytic / Holomorphic Function

f

is analytic (synonym: holomorphic) on an open set

Ω

if it is complex-differentiable at every point of

Ω

. A function analytic on all of

C

is called entire. "Analytic at a point

z_{0}

" means analytic on some open disc around

z_{0}

Cauchy–Riemann Equations

Suppose $f = u + i v$ is complex-differentiable at $z_{0} = x_{0} + i y_{0}$ . Then the partial derivatives of $u, v$ exist at $(x_{0}, y_{0})$ and satisfy

u_{x} = v_{y} and u_{y} = - v_{x} .

Moreover the derivative is then given by

f^{'} (z_{0}) = u_{x} + i v_{x} = v_{y} - i u_{y} .

Conversely, if $u, v$ have continuous first partials in a neighbourhood of $(x_{0}, y_{0})$ and satisfy $[2]$ there, then $f$ is complex-differentiable at $z_{0}$ (indeed analytic on that neighbourhood).

⟳Compute

[1]

once with

h = Δ x \to 0

(real) and once with

h = i Δ y \to 0

(imaginary) and equate the two complex values; the converse uses the real-differentiability total increment with the CR relations substituted in.

Wirtinger Form

Introduce the operators

\frac{\partial}{\partial z} = \frac{1}{2} (\frac{\partial}{\partial x} - i \frac{\partial}{\partial y}), \frac{\partial}{\partial z ˉ} = \frac{1}{2} (\frac{\partial}{\partial x} + i \frac{\partial}{\partial y}) .

Then the Cauchy–Riemann equations collapse to the single elegant condition

\frac{\partial f}{\partial z ˉ} = 0,

and when this holds $f^{'} (z) = \partial f / \partial z$ .

⟳Write

f = u + i v

, expand

\partial f / \partial \overset{z}{ˉ} = \frac{1}{2} [(u_{x} - v_{y}) + i (v_{x} + u_{y})]

; setting real and imaginary parts to zero reproduces

[2]

exactly.

Why It Holds

Necessity of the Cauchy–Riemann Equations

Step 1. Assume

f^{'} (z_{0})

exists, with value

a + ib

. Since the limit

[1]

is independent of how

h \to 0

, we may compute it along two convenient paths. Step 2 (real approach). Let

h = Δ x \in R

Δ x \to 0

. Then

\frac{f ( z _{0} + Δ x ) - f ( z _{0} )}{Δ x} = \frac{u ( x _{0} + Δ x , y _{0} ) - u ( x _{0} , y _{0} )}{Δ x} + i \frac{v ( x _{0} + Δ x , y _{0} ) - v ( x _{0} , y _{0} )}{Δ x} \to u_{x} + i v_{x} .

Hence

f^{'} (z_{0}) = u_{x} + i v_{x}

. Step 3 (imaginary approach). Let

h = i Δ y

Δ y \in R \to 0

. Then

1/ h = 1/ (i Δ y) = - i /Δ y

, so

\frac{f ( z _{0} + i Δ y ) - f ( z _{0} )}{i Δ y} = - i (\frac{u ( x _{0} , y _{0} + Δ y ) - u}{Δ y} + i \frac{v ( x _{0} , y _{0} + Δ y ) - v}{Δ y}) \to - i u_{y} + v_{y} .

Hence

f^{'} (z_{0}) = v_{y} - i u_{y}

. Step 4. The two expressions for

f^{'} (z_{0})

must be equal:

u_{x} + i v_{x} = v_{y} - i u_{y} .

Equating real parts:

u_{x} = v_{y}

. Equating imaginary parts:

v_{x} = - u_{y}

, i.e.

u_{y} = - v_{x}

. These are

[2]

, and either expression is

[3]

■

Sufficiency under Continuity of Partials

Step 1. Assume

u, v

have continuous partials near

(x_{0}, y_{0})

and satisfy

[2]

. Continuity of partials implies

u, v

are (real) differentiable, so

u (x_{0} + Δ x, y_{0} + Δ y) - u = u_{x} Δ x + u_{y} Δ y + ε_{1}, v (\dots) - v = v_{x} Δ x + v_{y} Δ y + ε_{2},

where

ε_{1}, ε_{2} = o (∣ h ∣)

with

h = Δ x + i Δ y

. Step 2. Combine:

f (z_{0} + h) - f (z_{0}) = (u_{x} + i v_{x}) Δ x + (u_{y} + i v_{y}) Δ y + (ε_{1} + i ε_{2}) .

Step 3. Use

[2]

to write

u_{y} = - v_{x}

v_{y} = u_{x}

. Then the

Δ y

coefficient is

u_{y} + i v_{y} = - v_{x} + i u_{x} = i (u_{x} + i v_{x})

. So

f (z_{0} + h) - f (z_{0}) = (u_{x} + i v_{x}) (Δ x + i Δ y) + o (∣ h ∣) = (u_{x} + i v_{x}) h + o (∣ h ∣) .

Step 4. Divide by

h

and let

h \to 0

: the quotient

\to u_{x} + i v_{x}

, a genuine complex limit. Hence

f^{'} (z_{0})

exists.

■

Continuity Hypothesis is Essential

The mere existence of partials satisfying $[2]$ at a single point does not guarantee complex differentiability. Sufficiency needs the partials to be continuous (or that $u, v$ are real-differentiable). See the counterexample below.

Augustin-Louis Cauchy | 1789–1857

Cauchy laid the foundations of complex analysis through his work on integration along paths in the complex plane (1825 memoir). He recognised that the value of an analytic function inside a contour is determined entirely by its boundary values — the seed of the integral formula. The directional-consistency conditions now bearing his and Riemann's names appeared in his investigations of complex integration and in earlier work by Euler and d'Alembert on fluid flow. Cauchy's rigour reshaped analysis: he is responsible for the modern $ε$ -style definition of limits underlying $[1]$ .

Building Intuition

Think of $f$ as a map of the plane to the plane. Its real Jacobian at $z_{0}$ is

J = (u_{x} v_{x} u_{y} v_{y}) .

For an arbitrary smooth map,

J

can be any

2 \times 2

matrix: stretching differently in different directions, shearing, reflecting. Multiplication by a single complex number

a + ib

, however, corresponds to the very special matrix

(a b - b a),

which is a uniform scaling by

a^{2} + b^{2}

combined with a rotation by

ar g (a + ib)

. Demanding

J

have this form forces

u_{x} = v_{y}

(

= a

) and

v_{x} = - u_{y}

(

= b

) — exactly the Cauchy–Riemann equations. So "analytic" = "the local linear approximation is a rotation-and-scaling, never a shear or reflection." Wherever

f^{'} \neq = 0

this means angles between curves are preserved: analytic maps are conformal.

The Cauchy–Riemann condition: any direction of approach yields the same derivative; the Jacobian acts as rotation-plus-scaling

Worked Examples

Verifying analyticity of $f(z)=z^2$

Given:

f (z) = z^{2}

. Write

z = x + i y

. Step 1 — separate parts.

z^{2} = (x + i y)^{2} = x^{2} - y^{2} + 2 i x y

, so

u (x, y) = x^{2} - y^{2}, v (x, y) = 2 x y .

Step 2 — partials.

u_{x} = 2 x, u_{y} = - 2 y, v_{x} = 2 y, v_{y} = 2 x .

Step 3 — check CR $[2]$ .

u_{x} = 2 x = v_{y}

✓ and

u_{y} = - 2 y = - v_{x}

✓. Both hold for all

(x, y)

, and the partials are continuous everywhere. Step 4 — derivative via $[3]$ .

f^{'} (z) = u_{x} + i v_{x} = 2 x + i 2 y = 2 (x + i y) = 2 z .

f (z) = z^{2} is entire, with f^{'} (z) = 2 z .

This matches the formal power-rule, as expected for any polynomial.

Conformal scaling factor of $f(z)=e^z$ at a point

Given:

f (z) = e^{z}

. Evaluate analyticity and the local scaling/rotation at

z_{0} = ln 2 + i \frac{π}{3}

. Step 1 — separate parts.

e^{x + i y} = e^{x} (cos y + i sin y)

, so

u = e^{x} cos y

v = e^{x} sin y

. Step 2 — partials.

u_{x} = e^{x} cos y, u_{y} = - e^{x} sin y, v_{x} = e^{x} sin y, v_{y} = e^{x} cos y .

Step 3 — CR check.

u_{x} = e^{x} cos y = v_{y}

✓;

u_{y} = - e^{x} sin y = - v_{x}

✓. Holds everywhere ⇒

e^{z}

is entire, with

f^{'} (z) = u_{x} + i v_{x} = e^{x} (cos y + i sin y) = e^{z} .

Step 4 — evaluate at $z_{0}$ .

f^{'} (z_{0}) = e^{z_{0}} = e^{l n 2} e^{iπ /3} = 2 (cos \frac{π}{3} + i sin \frac{π}{3}) = 2 (\frac{1}{2} + i \frac{3}{2}) = 1 + i 3 .

Step 5 — interpret. Local scaling factor

= ∣ f^{'} (z_{0}) ∣ = ∣1 + i 3 ∣ = 1 + 3 = 2

. Local rotation angle

= ar g (1 + i 3) = arctan (3) = \frac{π}{3} = 6 0^{\circ} .

Near z_{0}, e^{z} magnifies lengths by 2 and rotates by 6 0^{\circ} .

Common Misconceptions

Differentiable in $\mathbb{R}^2$ ≠ analytic

A map $(x, y) \mapsto (u, v)$ can be smooth as a real function yet fail to be complex-differentiable. The classic offender is $f (z) = \overset{z}{ˉ} = x - i y$ : here $u = x, v = - y$ , so $u_{x} = 1$ but $v_{y} = - 1$ , violating $[2]$ everywhere. $\overset{z}{ˉ}$ is infinitely real-differentiable yet nowhere analytic. Equivalently $\partial \overset{z}{ˉ} / \partial \overset{z}{ˉ} = 1 \neq = 0$ .

Analyticity is an open-set property

A function can be complex-differentiable at isolated points without being analytic anywhere. "Analytic at $z_{0}$ " requires differentiability on a whole neighbourhood, not just at $z_{0}$ .

CR equations hold at a point but $f$ is not differentiable there

Define

f (z) = ⎩ ⎨ ⎧ \frac{z ˉ ^{2}}{z}, 0, z \neq = 0 z = 0.

Writing

z = r e^{i θ}

\overset{z}{ˉ}^{2} / z = r e^{- 3 i θ}

. Along any straight ray

θ =

const, the difference quotient

\frac{f ( h ) - 0}{h} = \frac{h ˉ ^{2} / h}{h} = \frac{h ˉ ^{2}}{h ^{2}} = e^{- 2 i θ}

, which depends on $θ$ — different limits along different directions, so

f^{'} (0)

does not exist. Yet a direct computation along the axes shows the partials of

u, v

at the origin all exist and satisfy

u_{x} = v_{y}

u_{y} = - v_{x}

there. The CR equations are satisfied at the point, but

f

is not complex-differentiable, because real-differentiability (continuity of partials in a neighbourhood) fails. This is exactly why the converse theorem needs its continuity hypothesis.

$|z|^2$ — differentiable at one point only

f (z) = ∣ z ∣^{2} = x^{2} + y^{2}

has

u = x^{2} + y^{2}

v = 0

. CR demands

u_{x} = 2 x = v_{y} = 0

and

u_{y} = 2 y = - v_{x} = 0

, i.e.

x = y = 0

. So

f

is complex-differentiable only at $z = 0$ and analytic nowhere (no open set on which CR holds).

Connections

The Cauchy–Riemann equations imply, by differentiating $[2]$ again (when partials are $C^{2}$ , which analyticity guarantees), that $u_{xx} + u_{y y} = 0$ and $v_{xx} + v_{y y} = 0$ : both $u$ and $v$ are Harmonic Functions, with $v$ the harmonic conjugate of $u$ . This links analytic functions to Laplace's Equation and potential theory. The single complex equation $[5]$ , $\partial_{\overset{z}{ˉ}} f = 0$ , places analyticity inside the theory of the Wirtinger Derivatives and the $\overset{ˉ}{\partial}$ -operator. Analyticity is the gateway to Cauchy's Integral Theorem and Cauchy's Integral Formula, which in turn yield Taylor and Laurent Series representations and the Residue Theorem. The angle-preserving property ties the topic to Conformal Mappings.

Real-World Applications

Two-dimensional ideal fluid flow. The complex potential $Ω (z) = ϕ + i ψ$ packages velocity potential $ϕ$ and stream function $ψ$ ; analyticity is equivalent to the flow being irrotational and incompressible. Streamlines are level curves of $ψ$ .
Electrostatics & steady heat conduction. In planar problems, the electric potential is harmonic; analytic functions and conformal maps transform complicated boundary geometries (e.g. a wedge or slit) into simple ones (a half-plane).
Aerodynamics — Joukowski transform. The conformal map $w = z + 1/ z$ turns a circle into an airfoil shape, letting flow around a wing be computed from flow around a cylinder.
Signal processing & control theory. Transfer functions are analytic except at poles; the Residue Theorem evaluates inverse Laplace/Fourier transforms, and analyticity underlies the Kramers–Kronig relations linking the real and imaginary parts of a response (a CR-type Hilbert-transform relation).
Electrical engineering — impedance and the Smith chart. The bilinear (Möbius) map $Γ = (Z - Z_{0}) / (Z + Z_{0})$ is conformal, mapping impedance to reflection coefficient.
Number theory. The Riemann zeta function $ζ (s)$ is analytic on $C ∖ {1}$ ; analytic continuation and its zeros encode the distribution of primes.
Computational physics & finite elements. Conformal/quasi-conformal grid generation uses analytic mappings to mesh awkward 2-D domains smoothly.

Historical Development

1752 — Jean d'Alembert, studying fluid flow, writes down the equations $u_{x} = v_{y}$ , $u_{y} = - v_{x}$ in a hydrodynamic context.
1777 — Leonhard Euler uses the same relations in connection with complex integration and the evaluation of integrals.
1814–1825 — Augustin-Louis Cauchy develops complex contour integration; his 1825 memoir establishes path-independence of integrals of analytic functions, the foundation of the field.
1831 — Cauchy proves the integral formula and power-series expansion, showing analytic ⇒ infinitely differentiable.
1851 — Bernhard Riemann's doctoral thesis introduces the geometric viewpoint (Riemann surfaces, conformality) and treats $u_{x} = v_{y}$ , $u_{y} = - v_{x}$ as the defining conditions; the equations acquire his name alongside Cauchy's.
1850s–1870s — Karl Weierstrass builds complex analysis rigorously from convergent power series, defining "analytic" via series and proving equivalence with Cauchy's differentiability notion.
1900–1901 — Édouard Goursat removes the continuity-of-derivative hypothesis from Cauchy's theorem (the Cauchy–Goursat theorem), sharpening the logical foundations.
1927 — Wilhelm Wirtinger formalises the $\partial / \partial z$ , $\partial / \partial \overset{z}{ˉ}$ calculus, recasting CR as $\partial f / \partial \overset{z}{ˉ} = 0$ .
20th c. — The Looman–Menchoff theorem shows that continuity plus existence of partials satisfying CR (without assuming differentiability) already implies analyticity, refining the sufficiency conditions.

Subtleties & Limits

The local theory is complete, but global and boundary questions remain rich. Boundary behaviour: an analytic function on a disc need not extend continuously to the boundary, and characterising boundary values (the theory of $H^{p}$ Hardy spaces) is deep. Several complex variables: the naive analogue $\partial f / \partial \overset{z}{ˉ}_{j} = 0$ for each variable behaves utterly differently — Hartogs' phenomenon shows singularities cannot be isolated in $C^{n}$ for $n \geq 2$ , and which domains are "domains of holomorphy" (pseudoconvexity) is a central modern question. The $\overset{ˉ}{\partial}$ -problem: solving $\partial u / \partial \overset{z}{ˉ} = g$ with control of $u$ (Hörmander estimates) is an active analytic theme. Quasiconformal generalisations: relaxing CR to the Beltrami equation $f_{\overset{z}{ˉ}} = μ f_{z}$ underlies Teichmüller theory and remains technically demanding. Even the analytic-continuation behaviour governing the Riemann Hypothesis — zeros of $ζ (s)$ — is unresolved.

Analytic Functions and the Cauchy-Riemann Equations

The Core Insight

Formal Statement

Complex Differentiability

Analytic / Holomorphic Function

Cauchy–Riemann Equations

Wirtinger Form

Why It Holds

Necessity of the Cauchy–Riemann Equations

Sufficiency under Continuity of Partials

Continuity Hypothesis is Essential

Augustin-Louis Cauchy | 1789–1857

Building Intuition

The Cauchy–Riemann condition: any direction of approach yields the same derivative; the Jacobian acts as rotation-plus-scaling

Worked Examples

Verifying analyticity of $f(z)=z^2$

Conformal scaling factor of $f(z)=e^z$ at a point

Common Misconceptions

Differentiable in $\mathbb{R}^2$ ≠ analytic

Analyticity is an open-set property

CR equations hold at a point but $f$ is not differentiable there

$|z|^2$ — differentiable at one point only

Connections

Real-World Applications

Historical Development

Subtleties & Limits

See Also

Before this

Where this leads