Maximum Entropy Distribution with Constrained Mean

Pierre Dangauthier 1
1 E-MOTION - Geometry and Probability for Motion and Action
GRAVIR - IMAG - Graphisme, Vision et Robotique, Inria Grenoble - Rhône-Alpes
Abstract : This note presents the derivation of the maximum entropy distribution of a real variable on the unit segment, when its first moment is constrained. The functional form of the resulting distribution is an exponential with two Lagrange multipliers. We show that there is unique solution for those multipliers. However this solution has to be numerically approximated. As a special case, we find the uniform distribution when the constrained mean is centered.
Document type :
Preprints, Working Papers, ...
Liste complète des métadonnées

Cited literature [2 references]  Display  Hide  Download

https://hal.inria.fr/inria-00167289
Contributor : Rapport de Recherche Inria <>
Submitted on : Wednesday, September 5, 2007 - 2:35:27 PM
Last modification on : Friday, April 19, 2019 - 3:25:45 PM
Document(s) archivé(s) le : Friday, November 25, 2016 - 4:51:19 PM

Files

RR-6279.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : inria-00167289, version 2

Collections

Citation

Pierre Dangauthier. Maximum Entropy Distribution with Constrained Mean. 2007. ⟨inria-00167289v2⟩

Share

Metrics

Record views

252

Files downloads

295