Uniform Distribution

Modern computational systems implement random sampling using algorithms like the Mersenne Twister or PCG, generating pseudo-random integers mapped to floating-point numbers in the standard interval U ~ Uniform(0, 1). To sample from Uniform(a, b), we apply the linear map X = a + (b - a)*U. Because the transformation is strictly monotonic, probability density is preserved uniformly across the new width (b - a), guaranteeing that every subinterval of equal length within [a, b] has an identical probability of occurrence.

Machine learning weight initialization (e.g., Xavier/Glorot uniform initialization in PyTorch) draws initial neural network weights uniformly from [-1/sqrt(d), 1/sqrt(d)] to prevent vanishing and exploding gradients during forward passes.

John von Neumann introduced the first pseudo-random number generator (the middle-square method) in 1946 for Monte Carlo calculations at Los Alamos, recognizing the uniform distribution as the mother of all computational stochasticity.

'Scale and Shift' — multiply unit random U by span (b - a), then add baseline a: X = a + (b - a)*U.

Let X be the bus arrival time, where X ~ Uniform(10, 30). The total support width is b - a = 30 - 10 = 20 minutes. To satisfy the total probability axiom (total area = 1), the constant height must be f(x) = 1/(b - a) = 1/20 = 0.05 min^(-1). To find P(15 <= X <= 22), we evaluate the integral of the constant density over [15, 22]: P(15 <= X <= 22) = int_15²2 (1/20) dx = (22 - 15) / 20 = 7/20 = 0.35. Geometric area directly equals probability mass.

Airport ground operations and ride-hailing dispatch algorithms (e.g., Uber/Lyft pickup estimates) model arrival delays within guaranteed time windows to compute SLA violation probabilities.

Thomas Bayes and Pierre-Simon Laplace utilized the continuous uniform distribution in the 18th century as the 'principle of insufficient reason' (flat prior) when no prior knowledge favored one outcome over another.

'Target Width over Total Width' — for any uniform interval [c, d] inside [a, b], P = (d - c) / (b - a).

By definition, a uniform distribution assigns equal likelihood to equal subintervals. This implies that the density f(x) cannot fluctuate; it must be a constant value h across the support [a, b], and 0 elsewhere. By Kolmogorov's second axiom, the total probability across all possible real outcomes must equal 1: int_(-inf)^inf f(x) dx = int_a^b h dx = h * (b - a) = 1. Solving for h gives h = 1 / (b - a). If the support width doubles, the height must halve to preserve total probability.

Cryptographic hashing functions and random salt generators spread entropy uniformly across 256-bit key spaces to maximize uncertainty and ensure no individual cryptographic key bucket has higher collision probability.

Andrei Kolmogorov codified the normalization axiom int f(x)dx = 1 in 'Grundbegriffe der Wahrscheinlichkeitsrechnung' (1933), solidifying modern axiomatic probability.

'Unit Area Dictates Height' — width times height is 1, so height is always 1 over width: f(x) = 1/(b - a).

For any continuous random variable X ~ Uniform(a, b) and any width w > 0, consider two distinct intervals [x₁, x₁ + w] and [x₂, x₂ + w] lying completely within [a, b]. The probability of landing in the first is w / (b - a), and the probability of landing in the second is w / (b - a). Because the density function is horizontal and featureless, the probability depends solely on the interval length w, not its coordinate position. This translation invariance characterizes maximum ignorance or uniform spatial dispersion.

Round-robin load scheduling in distributed cloud architectures (e.g., Kubernetes pods) uses uniform random selection to ensure CPU traffic is evenly dispersed without creating hot spots.

Claude Shannon's 1948 information theory proved that among all continuous distributions bounded on [a, b], the Uniform distribution achieves maximum differential entropy (maximum uncertainty).

'Same Width, Same Odds' — if interval lengths match, their probabilities match.

By the continuous definition of expectation, E[X] = int_(-inf)^inf x f(x) dx. Substituting the uniform PDF f(x) = 1/(b - a) on [a, b]: E[X] = (1 / (b - a)) int_a^b x dx = (1 / (b - a)) * [x² / 2]_a^b = (b² - a²) / (2(b - a)). Factoring the numerator as a difference of squares: (b - a)(b + a) / (2(b - a)). Canceling (b - a) leaves E[X] = (a + b) / 2. For the bus arriving between 10 and 30 minutes, E[X] = (10 + 30) / 2 = 20 minutes.

Financial option pricing models (such as Monte Carlo strike simulations) use expected midpoint values to center payoff distributions when evaluating European binary contracts.

Pierre-Simon Laplace formalized continuous integration for centers of mass and probability expectations in 'Theorie analytique des probabilites' (1812), showing expectation is mathematically equivalent to the centroid of a physical laminar plate.

'Average of the Extremes' — the mean of Uniform(a, b) is simply the average of its boundaries: (a + b)/2.

Variance is defined as Var(X) = E[X²] - (E[X])². First compute the second moment: E[X²] = int_a^b x² (1 / (b - a)) dx = (1 / (b - a)) [x³ / 3]_a^b = (b³ - a³) / (3(b - a)). Factoring the difference of cubes: (b - a)(a² + ab + b²) / (3(b - a)) = (a² + ab + b²) / 3. Next, subtract the square of the mean: (E[X])² = ((a + b)/2)² = (a² + 2ab + b²) / 4. Finding a common denominator of 12: [4(a² + ab + b²) - 3(a² + 2ab + b²)] / 12 = (a² - 2ab + b²) / 12 = (b - a)² / 12. For the bus interval [10, 30], Var(X) = (30 - 10)² / 12 = 400 / 12 = 33.33 min², giving standard deviation sigma = sqrt(33.33) = 5.77 minutes.

Analog-to-digital converters (ADC) and digital signal processors introduce quantization noise modeled as Uniform(-Delta/2, Delta/2); the signal-to-noise ratio formula (SNR = 6.02 * N + 1.76 dB) is derived directly from this Delta² / 12 variance.

Abraham de Moivre established early moment calculations in 'The Doctrine of Chances' (1718), demonstrating how polynomial integrations systematically produce fractional factorial denominators.

'Spread Squared over Twelve' — the variance of Uniform(a, b) is always (width)² / 12.