Power Law & Pareto Distribution

Heavy tails, Zipf's law, and scale-free phenomena

Parameters

Real-world examples:

City sizes
α=1.16
Word freq
α=1.5
Earthquake
α=2.0
Wealth
α=2.5
Web links
α=3.0
Solar flares
α=1.8
Mean (finite if α>2):
Variance (finite if α>3):
Top 10% share: %
Fitted α (Hill):
Pareto / power law: P(X>x) = (x/x_min)^{-α+1}
PDF: p(x) = (α-1)/x_min · (x/x_min)^{-α}

Log-log plot: straight line with slope -α. Zipf's law: rank-frequency follows p∝rank^{-1} ⟺ α=2. Scale-free: no characteristic scale — the distribution looks the same at every scale.