Citation
A brief history of generative models for power law and lognormal distributions
TL;DR
Multiplicative processes generate log-normal distributions
Technical review of mechanisms generating power law and log-normal distributions, discussing when each applies. Essential for understanding that multiplicative processes typically generate log-normal distributions that approximate power laws only in upper tails.
Provides mathematical foundation for distinguishing between related heavy-tailed distributions and understanding their different implications.
Key Findings from Mitzenmacher (2004)
- Multiplicative processes generate log-normal distributions
- Log-normal approximates power law in upper tail
- Different mechanisms produce different distributional forms
- Historical development of generative models