Some properties of generalized Fisher information in the context of nonextensive thermostatistics
Jean-Fran\c{c}ois Bercher (LIGM)

TL;DR
This paper introduces extended forms of Fisher information within nonextensive thermostatistics, revealing their relationship with $q$-Gaussians and $q$-entropies, and establishing new variational and thermodynamic-like properties.
Contribution
It develops generalized Fisher information measures compatible with nonextensive statistics, linking them to $q$-Gaussians, $q$-entropies, and nonlinear heat equations, with new variational characterizations.
Findings
Generalized Fisher information minimized by $q$-Gaussians under constraints.
Derived a generalized de Bruijn identity involving $q$-entropies.
Showed convexity and bounds of the generalized Fisher information.
Abstract
We present two extended forms of Fisher information that fit well in the context of nonextensive thermostatistics. We show that there exists an interplay between these generalized Fisher information, the generalized -Gaussian distributions and the -entropies. The minimum of the generalized Fisher information among distributions with a fixed moment, or with a fixed -entropy is attained, in both cases, by a generalized -Gaussian distribution. This complements the fact that the -Gaussians maximize the -entropies subject to a moment constraint, and yields new variational characterizations of the generalized -Gaussians. We show that the generalized Fisher information naturally pop up in the expression of the time derivative of the -entropies, for distributions satisfying a certain nonlinear heat equation. This result includes as a particular case the classical de…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
