Cauchy's Integral Theorem and Formula

  • Cauchy's Integral Theorem: ∮_C f(z)dz = 0 for holomorphic f on closed curve C; reveals rigidity of complex analytic functions.
  • Cauchy's Integral Formula: f(z₀) = (1/2πi)∮_C f(z)/(z-z₀)dz determines function values inside from boundary data alone.
  • Holomorphic functions are infinitely differentiable and analytic; complex differentiability is far stronger than real differentiability.
  • Foundational for residue theorem, maximum modulus principle, and computing difficult real integrals via complex analysis.
  • Modern rigorous formulation developed through 1800s by Riemann, Weierstrass, and Goursat (eliminated need for derivative continuity).

Cauchy's Integral Theorem and Formula

Overview

Cauchy's Integral Theorem and Cauchy's Integral Formula form the cornerstone of complex analysis, establishing a profound connection between the local behavior of holomorphic functions and their global properties through path integration. These results reveal that analytic functions possess a rigidity far beyond their real counterparts: knowing a function on a curve allows us to determine its values and all derivatives at interior points. This phenomenon has no real analog and demonstrates the special geometry of the complex plane. The importance of these theorems cannot be overstated. They provide the foundation for the residue theorem, allow computation of otherwise intractable integrals, prove that holomorphic functions are infinitely differentiable (indeed, analytic), and establish fundamental properties like the maximum modulus principle. The integral formula shows that complex differentiability is an extraordinarily strong condition—much stronger than real differentiability. Historically, Augustin-Louis Cauchy developed these ideas in the 1820s-1840s, though his original arguments lacked the rigor of modern standards. The theorem initially appeared with restrictive hypotheses, and it took decades of work by Riemann, Weierstrass, Goursat, and others to achieve the elegant modern formulations. Goursat's 1900 proof, which eliminated the need to assume continuity of the derivative, was particularly significant.

💡 The Core Insight

If a function is holomorphic on and inside a closed curve, then integrating it around that curve yields zero. Moreover, the value of the function at any interior point is completely determined by its values on the boundary through an explicit integral formula—the function is "rigid" and cannot be modified locally without global consequences.

Definition

Cauchy's Integral Theorem and Formula: Let U⊆CU \subseteq \mathbb{C} be an open set and f:U→Cf: U \to \mathbb{C} a holomorphic function.

Theorem: If γ\gamma is a closed curve in UU that is null-homotopic in UU (or equivalently, if UU is simply connected), then

∮γf(z) dz=0\oint_\gamma f(z) \, dz = 0

Formula: If additionally a∈Ua \in U lies in the interior of γ\gamma (with γ\gamma positively oriented), then

f(a)=12πi∮γf(z)z−a dzf(a) = \frac{1}{2\pi i} \oint_\gamma \frac{f(z)}{z - a} \, dz

More generally, for all n≥0n \geq 0,

f(n)(a)=n!2πi∮γf(z)(z−a)n+1 dzf^{(n)}(a) = \frac{n!}{2\pi i} \oint_\gamma \frac{f(z)}{(z - a)^{n+1}} \, dz

Core Theory

TheoremMain Result

Theorem 1 (Cauchy's Integral Theorem - Simply Connected Version): Let U⊆CU \subseteq \mathbb{C} be a simply connected domain and let f:U→Cf: U \to \mathbb{C} be holomorphic. Then for any closed piecewise-C1C^1 curve γ\gamma in UU,

∮γf(z) dz=0\oint_\gamma f(z) \, dz = 0

Theorem 2 (Cauchy's Integral Formula): Let U⊆CU \subseteq \mathbb{C} be an open set, f:U→Cf: U \to \mathbb{C} holomorphic, and γ\gamma a closed piecewise-C1C^1 curve in UU that is null-homotopic in UU. If a∈Ua \in U does not lie on γ\gamma, then

12πi∮γf(z)z−a dz=n(γ,a)⋅f(a)\frac{1}{2\pi i} \oint_\gamma \frac{f(z)}{z - a} \, dz = n(\gamma, a) \cdot f(a)

where n(γ,a)n(\gamma, a) is the winding number of γ\gamma around aa.

Theorem 3 (Generalized Cauchy Formula for Derivatives): Under the conditions of Theorem 2, if n(γ,a)=1n(\gamma, a) = 1,

f(n)(a)=n!2πi∮γf(z)(z−a)n+1 dzf^{(n)}(a) = \frac{n!}{2\pi i} \oint_\gamma \frac{f(z)}{(z - a)^{n+1}} \, dz

for all n∈N∪{0}n \in \mathbb{N} \cup \{0\}.

Proof

Proof of Theorem 1 (Goursat's approach): We first prove this for a triangle Δ\Delta, then extend to general curves. Step 1: The case of a triangle. Let TT be a closed triangular path in UU. We must show ∫Tf(z) dz=0\int_T f(z) \, dz = 0. Subdivide TT into four congruent triangular subpaths T1(1),T2(1),T3(1),T4(1)T^{(1)}_1, T^{(1)}_2, T^{(1)}_3, T^{(1)}_4 by connecting midpoints. By the orientation convention, the integrals over interior edges cancel, so

∫Tf(z) dz=∑j=14∫Tj(1)f(z) dz\int_T f(z) \, dz = \sum_{j=1}^4 \int_{T^{(1)}_j} f(z) \, dz

At least one subtriangle satisfies ∣∫Tj(1)f(z) dz∣≥14∣∫Tf(z) dz∣\left|\int_{T^{(1)}_j} f(z) \, dz\right| \geq \frac{1}{4}\left|\int_T f(z) \, dz\right|. Call this T(1)T^{(1)}. Iterate this process: construct a sequence of triangular paths T⊇T(1)⊇T(2)⊇⋯T \supseteq T^{(1)} \supseteq T^{(2)} \supseteq \cdots where each T(n+1)T^{(n+1)} is a sub-triangle of T(n)T^{(n)} with side length 2−n2^{-n} times that of TT, and

∣∫T(n)f(z) dz∣≥4−n∣∫Tf(z) dz∣\left|\int_{T^{(n)}} f(z) \, dz\right| \geq 4^{-n} \left|\int_T f(z) \, dz\right|

By the nested interval theorem, there exists a unique point a∈⋂n=1∞T(n)a \in \bigcap_{n=1}^\infty T^{(n)}. Since ff is holomorphic at aa, we have f(z)=f(a)+f′(a)(z−a)+ψ(z)(z−a)f(z) = f(a) + f'(a)(z-a) + \psi(z)(z-a) where lim⁡z→aψ(z)=0\lim_{z \to a} \psi(z) = 0. For the constant and linear terms, ∫T(n)[f(a)+f′(a)(z−a)] dz=0\int_{T^{(n)}} [f(a) + f'(a)(z-a)] \, dz = 0 (integration of polynomials around closed curves). Thus ∫T(n)f(z) dz=∫T(n)ψ(z)(z−a) dz\int_{T^{(n)}} f(z) \, dz = \int_{T^{(n)}} \psi(z)(z-a) \, dz. Let LnL_n be the perimeter of T(n)T^{(n)}. Then Ln=2−nLL_n = 2^{-n} L where LL is the perimeter of TT. For ϵ>0\epsilon > 0, choose δ>0\delta > 0 such that ∣z−a∣<δ  ⟹  ∣ψ(z)∣<ϵ|z - a| < \delta \implies |\psi(z)| < \epsilon. For nn large enough that the diameter of T(n)T^{(n)} is less than δ\delta, we have for all zz on T(n)T^{(n)}: ∣ψ(z)∣<ϵ|\psi(z)| < \epsilon and ∣z−a∣≤Ln|z - a| \leq L_n. Thus:

∣∫T(n)f(z) dz∣≤ϵLn⋅Ln=ϵLn2=ϵ⋅4−nL2\left|\int_{T^{(n)}} f(z) \, dz\right| \leq \epsilon L_n \cdot L_n = \epsilon L_n^2 = \epsilon \cdot 4^{-n} L^2

But also ∣∫T(n)f(z) dz∣≥4−n∣∫Tf(z) dz∣\left|\int_{T^{(n)}} f(z) \, dz\right| \geq 4^{-n} \left|\int_T f(z) \, dz\right|. Therefore ∣∫Tf(z) dz∣≤ϵL2\left|\int_T f(z) \, dz\right| \leq \epsilon L^2 for all ϵ>0\epsilon > 0, implying ∫Tf(z) dz=0\int_T f(z) \, dz = 0. Step 2: General closed curves in simply connected domains. Since UU is simply connected, any closed curve γ\gamma in UU bounds a region contained in UU. By Green's theorem considerations or triangulation arguments, we can approximate the integral over γ\gamma by sums of integrals over triangular paths, each of which vanishes by Step 1. Taking limits establishes the result for general piecewise-C1C^1 curves. (A more rigorous approach uses that in a simply connected domain, ff has a primitive function FF with F′=fF' = f, constructed via path integration. Then ∮γf(z) dz=F(γ(b))−F(γ(a))=0\oint_\gamma f(z) \, dz = F(\gamma(b)) - F(\gamma(a)) = 0 since γ\gamma is closed.) □

∎
Proof

Proof of Theorem 2 (Cauchy's Integral Formula): We may assume n(γ,a)=1n(\gamma, a) = 1 by linearity (the general case follows by considering connected components). Step 1: Consider the function g(z)=f(z)−f(a)z−ag(z) = \frac{f(z) - f(a)}{z - a} for z≠az \neq a. We can extend gg to be continuous at aa by setting g(a)=f′(a)g(a) = f'(a). This extended function is holomorphic on U∖{a}U \setminus \{a\}. Step 2: Let ϵ>0\epsilon > 0 be small enough that the circle Cϵ(a)={z:∣z−a∣=ϵ}C_\epsilon(a) = \{z : |z - a| = \epsilon\} lies in UU and in the interior of γ\gamma. By deformation of contours (using Cauchy's theorem in the region between γ\gamma and Cϵ(a)C_\epsilon(a)),

∮γf(z)z−a dz=∮Cϵ(a)f(z)z−a dz\oint_\gamma \frac{f(z)}{z - a} \, dz = \oint_{C_\epsilon(a)} \frac{f(z)}{z - a} \, dz

Step 3: Parametrize Cϵ(a)C_\epsilon(a) as z=a+ϵeiθz = a + \epsilon e^{i\theta}, θ∈[0,2π]\theta \in [0, 2\pi]. Then:

∮Cϵ(a)f(z)z−a dz=∫02πf(a+ϵeiθ)ϵeiθ⋅iϵeiθ dθ=i∫02πf(a+ϵeiθ) dθ\oint_{C_\epsilon(a)} \frac{f(z)}{z - a} \, dz = \int_0^{2\pi} \frac{f(a + \epsilon e^{i\theta})}{\epsilon e^{i\theta}} \cdot i\epsilon e^{i\theta} \, d\theta = i\int_0^{2\pi} f(a + \epsilon e^{i\theta}) \, d\theta

As ϵ→0\epsilon \to 0, by continuity of ff, this approaches i∫02πf(a) dθ=2πif(a)i \int_0^{2\pi} f(a) \, d\theta = 2\pi i f(a). But the integral is independent of ϵ\epsilon by Step 2, so

\oint_\gamma \frac{f(z)}{z - a} \, dz = 2\pi i f(a)$$ □
∎
Proof

Proof of Theorem 3 (Derivatives): We prove this by induction. The case n=0n = 0 is Theorem 2. Assume the formula holds for nn. We must show:

f(n+1)(a)=(n+1)!2πi∮γf(z)(z−a)n+2 dzf^{(n+1)}(a) = \frac{(n+1)!}{2\pi i} \oint_\gamma \frac{f(z)}{(z-a)^{n+2}} \, dz

Consider the difference quotient:

f(n)(a+h)−f(n)(a)h=n!2πih∮γf(z)[1(z−a−h)n+1−1(z−a)n+1]dz\frac{f^{(n)}(a+h) - f^{(n)}(a)}{h} = \frac{n!}{2\pi i h} \oint_\gamma f(z) \left[\frac{1}{(z-a-h)^{n+1}} - \frac{1}{(z-a)^{n+1}}\right] dz

Algebraically,

1(z−a−h)n+1−1(z−a)n+1=(z−a)n+1−(z−a−h)n+1(z−a−h)n+1(z−a)n+1\frac{1}{(z-a-h)^{n+1}} - \frac{1}{(z-a)^{n+1}} = \frac{(z-a)^{n+1} - (z-a-h)^{n+1}}{(z-a-h)^{n+1}(z-a)^{n+1}}

Using the identity An+1−Bn+1=(A−B)(An+An−1B+⋯+Bn)A^{n+1} - B^{n+1} = (A-B)(A^n + A^{n-1}B + \cdots + B^n) with A=z−aA = z-a and B=z−a−hB = z-a-h:

=h∑k=0n(z−a)k(z−a−h)n−k(z−a−h)n+1(z−a)n+1= \frac{h \sum_{k=0}^n (z-a)^k(z-a-h)^{n-k}}{(z-a-h)^{n+1}(z-a)^{n+1}}

As h→0h \to 0, this sum has n+1n+1 terms each approaching (z−a)n(z-a)^n, so the ratio approaches (n+1)(z−a)n+2\frac{(n+1)}{(z-a)^{n+2}}. By the dominated convergence theorem for path integrals (justified by uniform bounds on compact sets away from aa),

f^{(n+1)}(a) = \frac{n!}{2\pi i} \oint_\gamma \frac{f(z) \cdot (n+1)}{(z-a)^{n+2}} \, dz = \frac{(n+1)!}{2\pi i} \oint_\gamma \frac{f(z)}{(z-a)^{n+2}} \, dz$$ □
∎
⚠️ Common mistake

Common Misconception: The integral theorem does NOT require that ff be holomorphic on all of C\mathbb{C}, only on a region containing the curve and its interior. A critical error is forgetting that the curve must be null-homotopic in the domain of holomorphy—for example, f(z)=1/zf(z) = 1/z is holomorphic on C∖{0}\mathbb{C} \setminus \{0\}, and ∮∣z∣=11z dz=2πi≠0\oint_{|z|=1} \frac{1}{z} \, dz = 2\pi i \neq 0 because the unit circle is not null-homotopic in the punctured plane. The domain topology matters crucially.

mathematician

Augustin-Louis Cauchy (1789-1857) — Developed the integral theorem and formula in the 1820s-1840s, establishing the foundations of rigorous complex analysis and introducing the concept of residues.

visualization

Re Im γ a ε z-a f(a) = 1/(2πi) ∮ γ f(z)/(z-a) dz

Examples and Applications

Example1

Computing an integral using Cauchy's formula Evaluate ∮∣z∣=2ezz2−1 dz\oint_{|z|=2} \frac{e^z}{z^2 - 1} \, dz. Solution: Factor the denominator: z2−1=(z−1)(z+1)z^2 - 1 = (z-1)(z+1). The singularities are at z=1z = 1 and z=−1z = -1, both inside ∣z∣=2|z| = 2. Use partial fractions:

ezz2−1=ez2(z−1)−ez2(z+1)\frac{e^z}{z^2 - 1} = \frac{e^z}{2(z-1)} - \frac{e^z}{2(z+1)}

By Cauchy's integral formula, with f(z)=ezf(z) = e^z (holomorphic everywhere):

∮∣z∣=2ezz−1 dz=2πie1=2πie\oint_{|z|=2} \frac{e^z}{z-1} \, dz = 2\pi i e^1 = 2\pi i e
∮∣z∣=2ezz+1 dz=2πie−1\oint_{|z|=2} \frac{e^z}{z+1} \, dz = 2\pi i e^{-1}

Therefore:

∮∣z∣=2ezz2−1 dz=12(2πie)−12(2πie−1)=πi(e−e−1)=2πisinh⁡(1)\oint_{|z|=2} \frac{e^z}{z^2 - 1} \, dz = \frac{1}{2}(2\pi i e) - \frac{1}{2}(2\pi i e^{-1}) = \pi i(e - e^{-1}) = 2\pi i \sinh(1)
Example2

Computing a derivative using the generalized formula Find f′′(0)f''(0) where f(z)=11−zf(z) = \frac{1}{1-z} using Cauchy's formula with the contour ∣z∣=1/2|z| = 1/2. Solution: The function ff is holomorphic in ∣z∣<1|z| < 1. By the generalized Cauchy formula with n=2n = 2 and a=0a = 0:

f′′(0)=2!2πi∮∣z∣=1/2f(z)z3 dz=1πi∮∣z∣=1/21z3(1−z) dzf''(0) = \frac{2!}{2\pi i} \oint_{|z|=1/2} \frac{f(z)}{z^3} \, dz = \frac{1}{\pi i} \oint_{|z|=1/2} \frac{1}{z^3(1-z)} \, dz

Alternatively, we can compute directly. Since f(z)=∑n=0∞znf(z) = \sum_{n=0}^\infty z^n for ∣z∣<1|z| < 1 (geometric series), we have:

f′(z)=∑n=1∞nzn−1,f′′(z)=∑n=2∞n(n−1)zn−2f'(z) = \sum_{n=1}^\infty n z^{n-1}, \quad f''(z) = \sum_{n=2}^\infty n(n-1) z^{n-2}

Thus f′′(0)=2⋅1=2f''(0) = 2 \cdot 1 = 2. We can verify this with the integral formula:

∮∣z∣=1/21z3(1−z) dz=∮∣z∣=1/21z3∑n=0∞zn dz=∑n=0∞∮∣z∣=1/2zn−3 dz\oint_{|z|=1/2} \frac{1}{z^3(1-z)} \, dz = \oint_{|z|=1/2} \frac{1}{z^3} \sum_{n=0}^\infty z^n \, dz = \sum_{n=0}^\infty \oint_{|z|=1/2} z^{n-3} \, dz

By the Cauchy residue theorem or direct calculation, ∮∣z∣=1/2zn−3 dz=0\oint_{|z|=1/2} z^{n-3} \, dz = 0 unless n−3=−1n - 3 = -1, i.e., n=2n = 2, in which case the integral is 2πi2\pi i. Therefore the integral equals 2πi2\pi i, confirming f′′(0)=1πi⋅2πi=2f''(0) = \frac{1}{\pi i} \cdot 2\pi i = 2.

Counterexample

Why holomorphy everywhere in a region matters Consider f(z)=1zf(z) = \frac{1}{z} and the contour γ(t)=e2πit\gamma(t) = e^{2\pi i t} for t∈[0,1]t \in [0,1] (the unit circle). The function ff is holomorphic on C∖{0}\mathbb{C} \setminus \{0\}, and γ\gamma lies in this domain. However:

∮γ1z dz=∫011e2πit⋅2πie2πit dt=∫012πi dt=2πi≠0\oint_\gamma \frac{1}{z} \, dz = \int_0^1 \frac{1}{e^{2\pi i t}} \cdot 2\pi i e^{2\pi i t} \, dt = \int_0^1 2\pi i \, dt = 2\pi i \neq 0

This does not contradict Cauchy's theorem because γ\gamma is not null-homotopic in C∖{0}\mathbb{C} \setminus \{0\}—it winds once around the hole at the origin. The domain C∖{0}\mathbb{C} \setminus \{0\} is not simply connected. If we take a contour like ∣z∣=2|z| = 2 and f(z)=z2f(z) = z^2 (holomorphic everywhere), then Cauchy's theorem correctly gives ∮∣z∣=2z2 dz=0\oint_{|z|=2} z^2 \, dz = 0. The lesson: the topology of the domain relative to the contour is essential.

🧠 Intuition

Geometric and physical intuition Cauchy's integral theorem says that if you integrate a holomorphic function around a closed loop that doesn't "trap" any singularities, you get zero. Think of the function values as defining a vector field on the plane; holomorphy imposes such rigid constraints that the circulation around any contractible loop vanishes. The integral formula has a beautiful interpretation: the value of ff at a point aa is a weighted average of its values on a surrounding circle, weighted by the "kernel" 12πi(z−a)\frac{1}{2\pi i(z-a)}. This is the mean value property in complex analysis. The function cannot have local extrema in the interior (leading to the maximum modulus principle) because its value is always an average of nearby values. Physically, if ff represents a complex potential (e.g., in fluid dynamics or electrostatics), the integral theorem corresponds to conservation laws, and the integral formula shows that the potential at any point is determined by boundary data—a fundamental principle in potential theory. The increasing powers in the derivative formula show that higher derivatives are increasingly sensitive to the values of ff near aa, with the weighting kernel (z−a)−(n+1)(z-a)^{-(n+1)} giving more weight to points closer to aa.

Historical Development

Augustin-Louis Cauchy first stated his integral theorem in 1825, though his proof relied on the assumption that f′f' was continuous. This was a significant restriction that Cauchy himself was aware of, and the theorem appeared in various forms throughout his work in the 1820s-1840s. Cauchy's integral formula appeared in his 1831 work. The crucial breakthrough came with Édouard Goursat's 1900 proof, which eliminated the need to assume continuity of the derivative, showing that complex differentiability alone (the existence of f′(a)f'(a) at each point) was sufficient. This "Goursat's theorem" showed that holomorphic functions are automatically continuously differentiable, a remarkable rigidity property. Bernhard Riemann (1850s) developed a different approach through Riemann surfaces and what are now called conformal mappings, emphasizing geometric aspects. Karl Weierstrass (1860s-1870s) took a more analytical approach via power series, proving that holomorphic functions are analytic. The modern formulation in terms of homotopy and winding numbers emerged in the early 20th century with the development of algebraic topology. Henri Cartan and others in the 1950s generalized these ideas to several complex variables, where the situation becomes dramatically more complicated.

Connections to Other Areas

connection

Harmonic functions and Laplace's equation: If f=u+ivf = u + iv is holomorphic, then both uu and vv satisfy Laplace's equation ∇2ϕ=0\nabla^2 \phi = 0 (they are harmonic functions). Conversely, any harmonic function on a simply connected domain is the real part of some holomorphic function (its harmonic conjugate). Cauchy's formulas thus provide explicit integral representations for solutions to Laplace's equation, connecting complex analysis to potential theory, electrostatics, fluid dynamics, and heat equation theory. The Cauchy kernel 12πi(z−a)\frac{1}{2\pi i(z-a)} is the Green's function for the Laplace operator in two dimensions.

What You Need First

What This Opens Up

See Also

Residue Theorem · Maximum Modulus Principle · Morera's Theorem · Liouville's Theorem · Laurent Series · Conformal Mapping · Harmonic Functions · Winding Number