E-Values for Exponential Families: the General Case

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hao, Yunda, Grünwald, Peter
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912553820487680
author Hao, Yunda
Grünwald, Peter
author_facet Hao, Yunda
Grünwald, Peter
contents We analyze common types of e-variables and e-processes for composite exponential family nulls: the optimal e-variable based on the reverse information projection (RIPr), the conditional (COND) e-variable, and the universal inference (UI) and sequen\-tialized RIPr e-processes. We characterize the RIPr prior for simple and Bayes-mixture based alternatives, either precisely (for Gaussian nulls and alternatives) or in an approximate sense (general exponential families). We provide conditions under which the RIPr e-variable is (again exactly vs. approximately) equal to the COND e-variable. Based on these and other interrelations which we establish, we determine the e-power of the four e-statistics as a function of sample size, exactly for Gaussian and up to $o(1)$ in general. For $d$-dimensional null and alternative, the e-power of UI tends to be smaller by a term of $(d/2) \log n + O(1)$ than that of the COND e-variable, which is the clear winner.
format Preprint
id arxiv_https___arxiv_org_abs_2409_11134
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle E-Values for Exponential Families: the General Case
Hao, Yunda
Grünwald, Peter
Methodology
We analyze common types of e-variables and e-processes for composite exponential family nulls: the optimal e-variable based on the reverse information projection (RIPr), the conditional (COND) e-variable, and the universal inference (UI) and sequen\-tialized RIPr e-processes. We characterize the RIPr prior for simple and Bayes-mixture based alternatives, either precisely (for Gaussian nulls and alternatives) or in an approximate sense (general exponential families). We provide conditions under which the RIPr e-variable is (again exactly vs. approximately) equal to the COND e-variable. Based on these and other interrelations which we establish, we determine the e-power of the four e-statistics as a function of sample size, exactly for Gaussian and up to $o(1)$ in general. For $d$-dimensional null and alternative, the e-power of UI tends to be smaller by a term of $(d/2) \log n + O(1)$ than that of the COND e-variable, which is the clear winner.
title E-Values for Exponential Families: the General Case
topic Methodology
url https://arxiv.org/abs/2409.11134