2026-10-08T01:38:38Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34032105 2026-09-30T12:12:13Z category_29893 item_type_12 month_year_09_2026

Sharp Design Thresholds for Identifying Heterogeneous Bernoulli–Logit Sources under Unknown Additive Noise Akihiro Koide (24791941) Statistical theory identifiability Bernoulli sources logit model unknown additive noise finite projections discrete tomography Prony reconstruction <p dir="ltr">This preprint establishes identifiability thresholds for heterogeneous Bernoulli–logit models observed with unknown additive noise. It gives sharp bounds on the number of experimental settings, using either a bound on rate diversity or a bound on the number of components. It also classifies exceptional four-setting designs for common-rate models with at most seven components and treats additional nuisance distributions with zero-free entire characteristic functions.</p><p dir="ltr"><br></p><p dir="ltr">The deposit includes the manuscript PDF, LaTeX source, and Python verification scripts with reference results. The scripts use exact arithmetic to check finite identities and explicit examples. The general mathematical proofs appear in the manuscript.</p> 2026-09-30T12:12:13Z Text Preprint 10.6084/m9.figshare.34032105.v1 https://figshare.com/articles/preprint/Sharp_Design_Thresholds_for_Identifying_Heterogeneous_Bernoulli_Logit_Sources_under_Unknown_Additive_Noise/34032105 CC BY 4.0