We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cond-mat.stat-mech

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Condensed Matter > Statistical Mechanics

Title: An intuition for physicists: information gain from experiments

Abstract: How much one has learned from an experiment is quantifiable by the information gain, also known as the Kullback-Leibler divergence. The narrowing of the posterior parameter distribution $P(\theta|D)$ compared with the prior parameter distribution $\pi(\theta)$, is quantified in units of bits, as: $ D_{\mathrm{KL}}(P|\pi)=\int\log_{2}\left(\frac{P(\theta|D)}{\pi(\theta)}\right)\,P(\theta|D)\,d\theta $. This research note gives an intuition what one bit of information gain means. It corresponds to a Gaussian shrinking its standard deviation by a factor of three.
Comments: Accepted to RNAAS; Corrected typos (Thanks to Tariq Yasin and Torsten En{\ss}lin)
Subjects: Statistical Mechanics (cond-mat.stat-mech); Instrumentation and Methods for Astrophysics (astro-ph.IM); Data Analysis, Statistics and Probability (physics.data-an)
Cite as: arXiv:2205.00009 [cond-mat.stat-mech]
  (or arXiv:2205.00009v3 [cond-mat.stat-mech] for this version)

Submission history

From: Johannes Buchner [view email]
[v1] Fri, 29 Apr 2022 18:00:01 GMT (67kb,D)
[v2] Wed, 4 May 2022 08:34:59 GMT (67kb,D)
[v3] Fri, 26 Aug 2022 14:54:24 GMT (67kb,D)

Link back to: arXiv, form interface, contact.