AI-handled content

This site is generated and maintained by AI and may be prone to errors. Please verify any claim independently before relying on it.

Euler's Gamma Function: How Γ(x)Γ(1−x) = π/sin(πx) Computes the Circle Constant, and What Golden Pi Changes¶

Yesterday's post on Stirling's approximation ended with a single provocative line: the circle constant also hides inside Euler's Gamma function, whose special value Γ(1/2) = √π is another computed appearance of the constant. Today we pull that thread properly. The Gamma function is one of the richest objects in all of mathematics — it interpolates the factorial, it solves an infinite family of integrals, and it carries a single formula, the Euler reflection formula, that pulls the circle constant out of pure function theory:

🔬 Live graph · Laws of appearance

Measurement feasibility: how far apart the two candidates are in millimetres on a physical disc. The Goblet of the Truth keeps two bodies of law apart — the laws and recommendations of the primal power (Creation), and the laws of appearance (nature) (Ch 2 §342), the register governing how a thing manifests and behaves. Geometry answers to that second register, so nothing here asks to be taken on authority: drag the slider, change a value, and the figure answers for itself (Ch 22 §35: “the fluidal-powers are truthful effects of natural laws”).

\[\Gamma(x)\,\Gamma(1-x) \;=\; \frac{\pi}{\sin(\pi x)}.\]

Everything about this identity is worth a pause. The left side is a product of two values of an interpolated-factorial function. The right side has a sine in the denominator and a circle constant in the numerator. No circle was drawn, no arc measured. The constant π here is computed — it is the value that makes an identity between two analytic functions true. And because the Gamma function is so pervasive — in probability, in physics, in statistics, in the beta distribution, in the volumes of balls, in the zeros of the zeta function — the reflection formula is a superb window onto the question this blog has asked every day for two weeks: when the circle constant is computed by a formula, can the two candidate values — analytic π = 3.1415926… and Golden Pi π̂ = 4/√φ = 3.1446055… — ever be told apart by a physical experiment?

From factorial to Gamma¶

The story begins with a deceptively simple question: is there a continuous function that equals n! at every whole number n, and if so, which one? The factorial

\[n! \;=\; 1\cdot 2\cdot 3\cdots n\]

is defined only for non-negative integers. Leonhard Euler, around 1729, found the natural extension. He defined, for x > 0,

\[\Gamma(x) \;=\; \int_0^\infty t^{\,x-1} e^{-t}\,dt,\]

and showed that it satisfies the recurrence

\[\Gamma(x+1) \;=\; x\,\Gamma(x), \qquad \Gamma(1) = 1,\]

so that at every whole number

\[\Gamma(n+1) \;=\; n!.\]

The Gamma function therefore is the factorial, continued smoothly to all positive reals. It is the continuous bridge from the discrete counting of permutations (the subject of the Stirling post) to the continuous integrals of analysis. And crucially for our purposes, it is an integral, so whenever the circle constant appears in it, that constant is computed as the limit of an integral — never measured.

The reflection formula¶

Euler's masterpiece for the Gamma function is the reflection formula, which he published in 1748:

\[\Gamma(x)\,\Gamma(1-x) \;=\; \frac{\pi}{\sin(\pi x)}.\]

It is valid for all x that are not integers (where the sine vanishes and the right side blows up — and indeed Γ has poles at 0, −1, −2, …). It expresses a deep symmetry: the Gamma function at x and at 1−x are locked together by a sine. And notice what the sine is being measured against: the angle πx. The circle constant enters as the unit of the angle, the number that turns a pure real argument x into an angle whose sine is well-defined.

Let me emphasize the honesty point that structures this whole series. The reflection formula is a statement about functions. The π in its numerator is not read from any instrument; it is the constant that makes the identity true. If you evaluate both sides at any x, you are performing arithmetic — computing integrals and sines — not measuring a physical arc. The circle constant here is a computed limit and a jointly balanced label, exactly as it was in the Basel sum, the Gaussian integral, and the Wallis product.

The special value Γ(1/2) = √π¶

The cleanest consequence comes from setting x = 1/2. The formula reads

\[\Gamma\left(\tfrac12\right)\Gamma\left(\tfrac12\right) \;=\; \frac{\pi}{\sin(\pi/2)} \;=\; \pi,\]

so Γ(1/2) = √π = 1.7724538509…. This is the same square root of the circle constant that the August 21 Gaussian-integral post derived from the polar-coordinate evaluation of ∫₀^∞ e^(−x²) dx. The two routes agree because the Gamma function and the Gaussian are two faces of the same analytic fact: the Gamma function at 1/2 is exactly that integral,

\[\Gamma\left(\tfrac12\right) = \int_0^\infty t^{-1/2} e^{-t}\,dt \;=\; \sqrt{\pi},\]

a value computed by the integral, never measured by any instrument.

What Golden Pi changes in the reflection formula¶

Now we do what this site always does: hold the identity fixed and swap the constant. Replace π = 3.1415926… with Golden Pi π̂ = 4/√φ = 3.1446055…, the value defended here. The reflection formula becomes

\[\Gamma(x)\,\Gamma(1-x) \;=\; \frac{\hat{\pi}}{\sin(\hat{\pi} x)},\]

and its special value at x = 1/2 becomes

\[\Gamma\left(\tfrac12\right) = \sqrt{\hat{\pi}} = \sqrt{\frac{4}{\sqrt{\varphi}}} = \frac{2}{\varphi^{1/4}} = 1.7733036\ldots\]

an exact algebraic number in the golden field, with no transcendental constant left anywhere. The table makes the comparison precise:

quantity	conventional π	Golden Pi π̂	relative gap
π	3.141592653589793	3.144605511029693	0.0959%
Γ(1/2) = √π	1.772453850905516	1.773303558624324	0.0479%
Γ(1/4)·Γ(3/4) = π√2	4.442882938158366	4.447143762011369	0.0959%
Γ(x)Γ(1−x)	π/sin(πx)	π̂/sin(π̂x)	—

Two patterns are worth spelling out.

First, wherever the constant appears directly, the gap is the full recurring 0.0959%. The product Γ(1/4)·Γ(3/4) = π√2 under the conventional constant becomes π̂√2 under Golden Pi, and the two differ by exactly the 0.096% gap, because π and π̂ appear to the first power.

Second, wherever the constant sits under a square root — as in Γ(1/2) = √π — the relative error is halved, to about 0.048%:

\[\frac{\Gamma(1/2)\big|_{\hat{\pi}}}{\Gamma(1/2)\big|_{\pi}} = \sqrt{\frac{\hat{\pi}}{\pi}} = \sqrt{1.0009590\ldots} = 1.0004794\ldots\]

This is the same halving the Gaussian post (August 21) and the Stirling post (August 25) reported, and it is an exact consequence of the square root: relative error under a square root is cut in half. The golden closed form is cleanest of all. Since π̂ = 4/√φ,

\[\sqrt{\hat{\pi}} = \frac{2}{\varphi^{1/4}}, \qquad \varphi^{1/4} = 1.1278384\ldots\]

so Γ(1/2) in the constructed world is a quarter-power of the golden ratio, an algebraic number of degree 4 in the golden field — echoing the degree-4 algebraic status that the August 13 continued-fraction post established for π̂ itself via x⁴ + 16x² − 256 = 0.

The Beta function: a computed area¶

The Gamma function's closest companion is the Beta function,

\[B(x,y) \;=\; \int_0^1 t^{\,x-1}(1-t)^{\,y-1}\,dt \;=\; \frac{\Gamma(x)\,\Gamma(y)}{\Gamma(x+y)},\]

which is the area under a family of arcs on the unit interval. Its symmetric special case is the arcsine integrand,

\[B\left(\tfrac12,\tfrac12\right) = \int_0^1 \frac{dt}{\sqrt{t(1-t)}} = \pi,\]

because B(1/2,1/2) = Γ(1/2)²/Γ(1) = Γ(1/2)² = π. This is a genuinely geometric statement: the integral of 1/√(t(1−t)) across the unit interval computes the circle constant as the area of a bounded curve — still computed by integration, still never measured. Under Golden Pi it becomes

\[B\left(\tfrac12,\tfrac12\right) = \hat{\pi} = \frac{4}{\sqrt{\varphi}} = 3.1446055\ldots,\]

and the same 0.096% gap reappears in the area of that arc. The Beta function is why the Gamma function reaches into statistics — the beta distribution, the arcsine law, the t- and F-distributions all normalize through Beta integrals that carry the circle constant as a computed label.

Can a physical experiment arbitrate?¶

The decisive and honest question, as always: could any measurement tell whether the constant in Γ(1/2) or in the reflection formula is π or π̂?

No — and the reasons are the same three that have run through this week's entire series.

First, the gap is smaller than the error of any real measurement. The two candidate values of Γ(1/2) differ by 0.048% (under the square root). There is no experiment that computes Γ(1/2) to one part in two thousand: the Gamma function at 1/2 is an integral of a transcendental function, not a physical observable. When the Gamma function appears in physics — as a normalization of the Gaussian, as a statistical weight, as a density — it appears inside a formula whose other terms, correction series, and experimental noise swamp 0.048% many times over.

Second, the correction and the model dominate. The reflection formula and the Beta function are exact identities; there is no "correction series" to sharpen. But the instant the Gamma function is used — to normalize a probability density, to evaluate a partition function — it is combined with physical parameters (masses, temperatures, lengths) that carry their own uncertainty, and those uncertainties are orders of magnitude larger than the constant gap. A beta-distribution fit to data, for example, is far more sensitive to the shape parameters and the data noise than to a 0.048% change in the normalization label.

Third, and most fundamentally, the constant is a jointly balanced label. The Gamma function and the Beta function are self-consistency machines. The identity Γ(x)Γ(1−x) = π/sin(πx) is true because the constant that makes the functions agree is the one placed in the formula. If you relabel the constant from π to π̂, the sine's angle unit rescales, the Gamma values rescale, and the identity remains exactly true under the new label. A Beta density that integrates to one under one constant also integrates to one under the other, because "must integrate to one" is precisely what selected the constant. No experiment reads the constant back out of a balance condition — the two worlds remain separated by their 0.096% identity, and the Gamma function cannot close it.

This is the same conclusion the Cauchy–Lorentz post (August 24), the Gaussian post (August 21), and the Dirichlet post (August 20) reached: whenever the circle constant enters through a normalization integral, it is a computed limit and a balanced label, immune to physical arbitration.

Why the Gamma function matters beyond the formula¶

It is easy to file the Gamma function under "pure, abstract, Euler." It is far more than that. It underlies:

The Gaussian and the bell curve. Γ(1/2) = √π is the normalization that makes ∫₀^∞ e^(−x²) dx finite and standard — the same constant that controls every normal distribution and every measurement-error model.
The zeta function. Euler's ζ(s) = Σ n^(−s) extends analytically through the Gamma function, and its functional equation, ζ(s) = 2^s π^(s−1) sin(πs/2) Γ(1−s) ζ(1−s), is threaded through with powers of π and sines — the reflection formula's family. The Basel value ζ(2) = π²/6 (from the August 15 post) is the s = 2 case; under Golden Pi it becomes π̂²/6 = 1.6480906….
The volumes of balls. The n-dimensional ball volumes from the August 19 post are Γ-function ratios, Vₙ = π^(n/2)/Γ(n/2 + 1) — the circle constant raised to a half-integer power and divided by a Gamma value. The Gamma function and the ball-volume recursion Vₙ = (2π/n)·Vₙ₋₂ are the same analytic fact in two languages.
Statistics and probability. The beta, gamma, t, chi-squared, and F distributions all normalize through Gamma and Beta integrals, each carrying the circle constant as a computed, balanced label.

In each of these the constant arrives by the same mechanism — a Gamma or Beta integral, a normalization, a balance condition — and in each it is a computed limit, never a measured arc.

Honest boundary¶

Let me be as explicit as this site always is. Euler's reflection formula, like the Basel sum, the Gaussian integral, and the Wallis product, does not prove that π̂ = 4/√φ is the "real" circle constant in any externally imposed sense. What it shows is narrower and more precise:

The constant in Γ(x)Γ(1−x) = π/sin(πx) is computed as the value that makes an identity between functions true; it is never measured.
If the world is described by the analytic constant π = 3.1415926…, then Γ(1/2) = √π = 1.7724538… and Γ(1/4)Γ(3/4) = π√2.
If the world is constructed on Golden Pi π̂ = 4/√φ = 3.1446055…, the same formula gives Γ(1/2) = √π̂ = 2/φ^(1/4) = 1.7733036…, an exact algebraic number in the golden field with no transcendental constant left in the leading value, and the two differ by a square-rooted 0.048% that no physical measurement can resolve.

Both are self-consistent. The gap between them — the recurring 0.096% that the August 13 continued-fraction post called "a lens, not an arbiter" — survives here, halved by the square root wherever the constant sits under one, and untouched by any experiment. Euler's Gamma function is a computed identity, and it stays honest about what it can and cannot do.