A Langevin approach to the Log-Gauss-Pareto composite statistical structure

Iddo I. Eliazar, Morrel H. Cohen

Research output: Contribution to journalArticlepeer-review

23 Scopus citations

Abstract

The distribution of wealth in human populations displays a Log-Gauss-Pareto composite statistical structure: its density is Log-Gauss in its central body, and follows power-law decay in its tails. This composite statistical structure is further observed in other complex systems, and on a logarithmic scale it displays a Gauss-Exponential structure: its density is Gauss in its central body, and follows exponential decay in its tails. In this paper we establish an equilibrium Langevin explanation for this statistical phenomenon, and show that: (i) the stationary distributions of Langevin dynamics with sigmoidal force functions display a Gauss-Exponential composite statistical structure; (ii) the stationary distributions of geometric Langevin dynamics with sigmoidal force functions display a Log-Gauss-Pareto composite statistical structure. This equilibrium Langevin explanation is universal - as it is invariant with respect to the specific details of the sigmoidal force functions applied, and as it is invariant with respect to the specific statistics of the underlying noise.

Original languageEnglish (US)
Pages (from-to)5598-5610
Number of pages13
JournalPhysica A: Statistical Mechanics and its Applications
Volume391
Issue number22
DOIs
StatePublished - Nov 15 2012

All Science Journal Classification (ASJC) codes

  • Statistical and Nonlinear Physics
  • Statistics and Probability

Keywords

  • ''mild" randomness
  • ''wild" randomness
  • Exponential tails
  • Gauss distribution
  • Geometric Langevin dynamics
  • Geometric Ornstein-Uhlenbeck dynamics
  • Langevin dynamics
  • Laplace distribution
  • Linear forces
  • Log-Gauss distribution
  • Log-Laplace distribution
  • Ornstein-Uhlenbeck dynamics
  • Power-law tails
  • Sigmoidal forces
  • Universality

Fingerprint

Dive into the research topics of 'A Langevin approach to the Log-Gauss-Pareto composite statistical structure'. Together they form a unique fingerprint.

Cite this