Gamma and Beta Functions

⚡ Key Takeaways

The Gamma function $\Gamma(z)$ extends the factorial to complex arguments via $\Gamma(n)=(n-1)!$ ; the Beta function $B(x,y)$ is a closely related two-variable integral satisfying $B(x,y)=\Gamma(x)\Gamma(y)/\Gamma(x+y)$ . They appear everywhere: probability distributions, special-function identities, fractional calculus, and physics integrals. Introduced by Euler (1729–1730), refined by Legendre, Gauss, and Weierstrass.

💡 The Core Insight

Factorials are defined by the recursion $n! = n\cdot(n-1)!$ . The Gamma integral satisfies the same recursion $\Gamma(z+1)=z\,\Gamma(z)$ but for continuous and complex $z$ . Demanding this recursion plus logarithmic convexity (Bohr–Mollerup) pins down a unique extension. The Beta function is simply what you get when two such integrals are multiplied and re-parameterized — the appearance of $\Gamma(x+y)$ in its denominator is the signature of this convolution structure.

DefinitionGamma Function (Euler integral of the second kind)

For $\operatorname{Re}(z)>0$ ,

1

\quad \Gamma(z)=\int_0^\infty t^{z-1}e^{-t}\,dt.

Derivation hint: convergence at $0$ needs $\operatorname{Re}(z)>0$ (integrand $\sim t^{z-1}$ ); convergence at $\infty$ is guaranteed by $e^{-t}$ for all $z$ . Here $z\in\mathbb{C}$ , $t$ is the real integration variable, $t^{z-1}=e^{(z-1)\ln t}$ uses the principal logarithm, and $e^{-t}$ provides exponential decay. The integral defines a holomorphic function on the right half-plane.

DefinitionBeta Function (Euler integral of the first kind)

For $\operatorname{Re}(x)>0,\ \operatorname{Re}(y)>0$ ,

2

\quad B(x,y)=\int_0^1 t^{x-1}(1-t)^{y-1}\,dt.

Derivation hint: convergence at $t=0$ requires $\operatorname{Re}(x)>0$ , at $t=1$ requires $\operatorname{Re}(y)>0$ . The variables $x,y\in\mathbb{C}$ are the two parameters; $t$ ranges over $[0,1]$ . The integrand is symmetric under $t\mapsto 1-t$ , which forces $B(x,y)=B(y,x)$ .

TheoremFundamental Properties

The Gamma function satisfies:

3

\quad \Gamma(z+1)=z\,\Gamma(z),\qquad \Gamma(n)=(n-1)!\ \ (n\in\mathbb{Z}^+).

Derivation hint: integrate $[1]$ by parts with $u=t^{z}$ , $dv=e^{-t}dt$ ; the boundary term vanishes for $\operatorname{Re}(z)>0$ .

4

\quad \Gamma(z)\,\Gamma(1-z)=\frac{\pi}{\sin(\pi z)}\qquad (z\notin\mathbb{Z}).

Derivation hint: reflection formula — evaluate $B(z,1-z)=\int_0^\infty \frac{u^{z-1}}{1+u}\,du=\pi/\sin(\pi z)$ by contour integration.

5

\quad B(x,y)=\frac{\Gamma(x)\,\Gamma(y)}{\Gamma(x+y)}.

Derivation hint: multiply two Gamma integrals, switch to polar/affine coordinates, and split the radial and angular parts (see proof below).

6

\quad \Gamma\!\left(\tfrac12\right)=\sqrt{\pi}.

Derivation hint: set $z=\tfrac12$ in $[4]$ , or evaluate the Gaussian integral after substituting $t=u^2$ in $[1]$ .

ProofRecursion

\Gamma(z+1)=z\Gamma(z)

Step 1. Start from the definition:

7

\quad \Gamma(z+1)=\int_0^\infty t^{z}e^{-t}\,dt.

Derivation hint: just replace $z$ by $z+1$ in $[1]$ .

Step 2. Integrate by parts with $u=t^{z}$ (so $du=z\,t^{z-1}dt$ ) and $dv=e^{-t}dt$ (so $v=-e^{-t}$ ):

8

\quad \Gamma(z+1)=\Big[-t^{z}e^{-t}\Big]_0^\infty+z\int_0^\infty t^{z-1}e^{-t}\,dt.

Derivation hint: standard $\int u\,dv = uv-\int v\,du$ .

Step 3. The boundary term vanishes: as $t\to\infty$ , $e^{-t}$ dominates; as $t\to 0^+$ , $t^z\to 0$ for $\operatorname{Re}(z)>0$ . Hence

9

\quad \Gamma(z+1)=z\,\Gamma(z).

Derivation hint: the surviving integral is exactly $\Gamma(z)$ from $[1]$ .

Step 4. Since $\Gamma(1)=\int_0^\infty e^{-t}dt=1$ , induction gives $\Gamma(n)=(n-1)!$ . $\blacksquare$

∎

ProofAnalytic continuation to

\mathbb{C}\setminus\{0,-1,-2,\dots\}

Step 1. Rearranging $[3]$ gives $\Gamma(z)=\Gamma(z+1)/z$ . The right side is defined for $\operatorname{Re}(z)>-1$ (except $z=0$ ). Derivation hint: $\Gamma(z+1)$ is holomorphic for $\operatorname{Re}(z+1)>0$ , i.e. $\operatorname{Re}(z)>-1$ .

Step 2. Iterating,

13

\quad \Gamma(z)=\frac{\Gamma(z+n+1)}{z(z+1)\cdots(z+n)}.

Derivation hint: apply $[3]$ repeatedly $n+1$ times.

Step 3. This is valid for $\operatorname{Re}(z)>-(n+1)$ except at $z=0,-1,\dots,-n$ , where simple poles occur. The residue at $z=-n$ is

14

\quad \operatorname{Res}_{z=-n}\Gamma(z)=\frac{(-1)^n}{n!}.

Derivation hint: in $[13]$ near $z=-n$ , only the factor $(z+n)$ vanishes; substitute $z=-n$ in the rest and use $\Gamma(1)=1$ . $\blacksquare$

∎

LE

Leonhard Euler

1707–1783

Euler discovered both functions while solving the interpolation problem posed by Goldbach and Daniel Bernoulli: find a smooth function passing through the factorial values. In a 1729 letter he gave an infinite-product form, and in 1730 the integral $[1]$ (then written with different conventions). His genius was recognizing that $n!$ is not an isolated arithmetic object but a sampled value of a single analytic curve — a viewpoint that seeded the entire theory of Special Functions and laid groundwork for Complex Analysis.

🧠 Intuition

Think of $\Gamma(z)$ as a weighted "area" machine: $\int_0^\infty t^{z-1}e^{-t}dt$ measures how the bump $e^{-t}$ is reshaped by the monomial weight $t^{z-1}$ . Larger $z$ pushes the weight to bigger $t$ , fattening the integral — exactly the growth captured by factorial. The recursion $\Gamma(z+1)=z\,\Gamma(z)$ is the continuous analogue of "to advance one step, multiply by the current value." Between integers, the curve is forced to dip: $\Gamma$ has a minimum near $z\approx 1.4616$ where $\Gamma\approx 0.8856$ — the factorial "interpolation" is not monotone on $(0,1)$ . The Beta function is best pictured as the shape factor of the region $\{u,v>0,\ u+v=\text{const}\}$ . When you slice the joint Gamma density along constant-sum lines, the angular distribution is exactly the Beta density — which is why Beta governs proportions and the Dirichlet Distribution.

Figure

The Gamma function on the real axis, showing poles at non-positive integers

ExampleEvaluating a Gaussian-type integral

Given: Compute $I=\int_0^\infty t^{5}e^{-t}\,dt$ and $J=\int_0^\infty x^{4}e^{-x^2}\,dx$ .

Step 1. For $I$ , match to $[1]$ : $t^{z-1}=t^5\Rightarrow z=6$ . So $I=\Gamma(6)=5!=120$ .

Step 2. For $J$ , substitute $u=x^2$ , $x=u^{1/2}$ , $dx=\tfrac12 u^{-1/2}du$ :

15

\quad J=\int_0^\infty u^{2}e^{-u}\cdot\tfrac12 u^{-1/2}\,du=\tfrac12\int_0^\infty u^{3/2}e^{-u}\,du=\tfrac12\,\Gamma\!\left(\tfrac52\right).

Derivation hint: $x^4=u^2$ and combine $u^{2}\cdot u^{-1/2}=u^{3/2}$ .

Step 3. Use the recursion: $\Gamma(\tfrac52)=\tfrac32\cdot\tfrac12\cdot\Gamma(\tfrac12)=\tfrac34\sqrt\pi$ . So $J=\tfrac12\cdot\tfrac34\sqrt\pi=\tfrac{3}{8}\sqrt\pi$ .

\boxed{I=120\ \text{(dimensionless)},\qquad J=\tfrac{3}{8}\sqrt{\pi}\approx 0.6647}

ExampleA Beta-function integral

Given: Compute $K=\int_0^{\pi/2}\sin^4\theta\,\cos^6\theta\,d\theta$ .

Step 1. Use the trigonometric form of Beta (from $t=\sin^2\theta$ ):

16

\quad \int_0^{\pi/2}\sin^{2a-1}\theta\,\cos^{2b-1}\theta\,d\theta=\tfrac12 B(a,b).

Derivation hint: substitute $t=\sin^2\theta$ , $dt=2\sin\theta\cos\theta\,d\theta$ , into $[2]$ .

Step 2. Match exponents: $2a-1=4\Rightarrow a=\tfrac52$ ; $2b-1=6\Rightarrow b=\tfrac72$ . Thus $K=\tfrac12 B(\tfrac52,\tfrac72)$ .

Step 3. Apply $[5]$ : $B(\tfrac52,\tfrac72)=\dfrac{\Gamma(5/2)\Gamma(7/2)}{\Gamma(6)}$ . With $\Gamma(\tfrac52)=\tfrac34\sqrt\pi$ , $\Gamma(\tfrac72)=\tfrac{15}{8}\sqrt\pi$ , $\Gamma(6)=120$ :

17

\quad B\!\left(\tfrac52,\tfrac72\right)=\frac{\tfrac34\sqrt\pi\cdot\tfrac{15}{8}\sqrt\pi}{120}=\frac{\tfrac{45}{32}\pi}{120}=\frac{45\pi}{3840}=\frac{3\pi}{256}.

Derivation hint: multiply the half-integer Gamma values, keep $\pi$ exact, simplify the fraction.

Step 4. Therefore $K=\tfrac12\cdot\tfrac{3\pi}{256}=\tfrac{3\pi}{512}$ .

\boxed{K=\dfrac{3\pi}{512}\approx 0.01841\ \text{(dimensionless)}}

⚠️ Common mistake

$\Gamma(n)=n!$ is FALSE. The shift is essential: $\Gamma(n)=(n-1)!$ . So $\Gamma(5)=24$ , not $120$ . The mismatch comes from the integrand $t^{z-1}$ , not $t^z$ . To get a true factorial extension with $\Pi(n)=n!$ , some authors use the Pi function $\Pi(z)=\Gamma(z+1)$ — but this is non-standard; always check conventions.

The Gamma integral $[1]$ is only the definition for $\operatorname{Re}(z)>0$ . For $\operatorname{Re}(z)\le 0$ the integral diverges; one must use the continuation $[13]$ . Plugging negative arguments directly into the integral is meaningless.

Counterexample

A tempting but wrong belief: " $\Gamma$ is defined everywhere because factorials of negative numbers should just be large." In fact $\Gamma$ has simple poles at $0,-1,-2,\dots$ , so quantities like $\Gamma(-2)$ are undefined ( $\infty$ ). Concretely, attempting $\Gamma(0)$ via the recursion: $\Gamma(0)=\Gamma(1)/0=1/0$ , undefined. And $\Gamma(-\tfrac12)$ does exist (it is $-2\sqrt\pi$ ) because $-\tfrac12$ is not an integer — so "negative argument" is not the issue; "non-positive integer" is. This subtlety routinely trips up students.

connection

The Gamma function is the hub of the special-function universe. It appears in Stirling's Approximation $\Gamma(z+1)\sim\sqrt{2\pi z}\,(z/e)^z$ , controls the normalization of the Gamma Distribution and Beta Distribution in probability, and defines the Riemann Zeta Function through its functional equation $\zeta(s)=\Gamma(1-s)\,\dots$ relation. The Beta function provides closed forms for volumes of $n$ -balls (linking to Multiple Integrals) and underlies the Dirichlet Distribution. Via the Mellin Transform, $\Gamma$ is literally the Mellin transform of $e^{-t}$ , tying it to Laplace and Fourier Transforms. The reflection formula $[4]$ is a gateway into Complex Analysis and contour integration.

🔬 Where this appears

⏳ History

- 1729–1730 — Euler. Solves the factorial-interpolation problem; gives both an infinite product and the integral form, establishing

\Gamma

and

B

.

- 1768 — Euler. Publishes the reflection-type and product relations connecting the two integrals.

- 1809 — Legendre. Introduces the notation

\Gamma

and the symbol; proves the duplication formula

\Gamma(z)\Gamma(z+\tfrac12)=2^{1-2z}\sqrt\pi\,\Gamma(2z)

.

- 1812 — Gauss. Gives the limit/product definition

\Gamma(z)=\lim_{n\to\infty}\frac{n!\,n^z}{z(z+1)\cdots(z+n)}

, valid for all non-pole

z

.

- 1854 — Weierstrass. Produces the canonical product

1/\Gamma(z)=z e^{\gamma z}\prod_n(1+z/n)e^{-z/n}

with the Euler–Mascheroni constant

\gamma

, displaying the zeros of

1/\Gamma

.

- 1856 — Riemann. Uses

\Gamma

in the functional equation of

\zeta(s)

, embedding it in analytic number theory.

- 1922 — Bohr & Mollerup. Prove that log-convexity plus

f(1)=1

and the recursion uniquely characterize

\Gamma

, finally justifying it as the factorial extension.

- 1931 — Artin. Publishes The Gamma Function, a definitive modern treatment streamlining the Bohr–Mollerup approach.

- 1970s — QFT era. Dimensional regularization ('t Hooft–Veltman) turns Gamma functions into everyday tools of particle physics.

open-problem

The Gamma function is never zero ( $1/\Gamma$ is entire), and its only singularities are the simple poles at non-positive integers — but the deep questions concern its arithmetic nature. Whether $\Gamma(1/3)$ , $\Gamma(1/4)$ are algebraically independent of $\pi$ involves transcendence theory; Chudnovsky proved $\Gamma(1/4)$ and $\pi$ are algebraically independent, while the status of $\Gamma(1/5)$ relative to known constants is more delicate. More broadly, period theory (Kontsevich–Zagier) seeks to classify exactly which Gamma values are "periods" and what relations they satisfy — a largely open program. On the analytic side, the asymptotic Stirling's Approximation is divergent (asymptotic, not convergent), and controlling its error terms in the complex plane near the negative real axis remains technically subtle.

Formal Statement

Why It Holds / Derivation

Building Intuition

Worked Examples

Common Misconceptions

Connections

Real-World Applications

Historical Development

Subtleties & Limits

See Also