Information geometry of small diffusions

Tomonari Sei, Fumiyasu Komaki

Research output: Contribution to journalArticle

Abstract

Information geometrical quantities such as metric tensors and connection coefficients for small diffusion models are obtained. Asymptotic properties of bias-corrected estimators for small diffusion models are investigated from the viewpoint of information geometry. Several results analogous to those for independent and identically distributed (i.i.d.) models are obtained by using the asymptotic normality of the statistics appearing in asymptotic expansions. In contrast to the asymptotic theory for i.i.d.models, the geometrical quantities depend on the magnitude of noise.

Original languageEnglish
Pages (from-to)123-141
Number of pages19
JournalStatistical Inference for Stochastic Processes
Volume11
Issue number2
DOIs
Publication statusPublished - 2008 Jun 1

Keywords

  • Curved exponential family
  • Information geometry
  • Second-order asymptotic efficiency
  • Small diffusion models

ASJC Scopus subject areas

  • Statistics and Probability

Fingerprint Dive into the research topics of 'Information geometry of small diffusions'. Together they form a unique fingerprint.

  • Cite this