English

Combinatorial Entropies and Statistics

Statistical Mechanics 2015-05-13 v2

Abstract

We examine the {combinatorial} or {probabilistic} definition ("Boltzmann's principle") of the entropy or cross-entropy function HlnWH \propto \ln \mathbb{W} or DlnPD \propto - \ln \mathbb{P}, where W\mathbb{W} is the statistical weight and P\mathbb{P} the probability of a given realization of a system. Extremisation of HH or DD, subject to any constraints, thus selects the "most probable" (MaxProb) realization. If the system is multinomial, DD converges asymptotically (for number of entities N\back\backN \back \to \back \infty) to the Kullback-Leibler cross-entropy DKLD_{KL}; for equiprobable categories in a system, HH converges to the Shannon entropy HShH_{Sh}. However, in many cases W\mathbb{W} or P\mathbb{P} is not multinomial and/or does not satisfy an asymptotic limit. Such systems cannot meaningfully be analysed with DKLD_{KL} or HShH_{Sh}, but can be analysed directly by MaxProb. This study reviews several examples, including (a) non-asymptotic systems; (b) systems with indistinguishable entities (quantum statistics); (c) systems with indistinguishable categories; (d) systems represented by urn models, such as "neither independent nor identically distributed" (ninid) sampling; and (e) systems representable in graphical form, such as decision trees and networks. Boltzmann's combinatorial definition of entropy is shown to be of greater importance for {"probabilistic inference"} than the axiomatic definition used in information theory.

Keywords

Cite

@article{arxiv.0902.3038,
  title  = {Combinatorial Entropies and Statistics},
  author = {Robert K. Niven},
  journal= {arXiv preprint arXiv:0902.3038},
  year   = {2015}
}

Comments

Invited contribution to the SigmaPhi 2008 Conference; accepted by EPJB volume 69 issue 3 June 2009

R2 v1 2026-06-21T12:12:44.770Z