Saved in:
Bibliographic Details
Main Author: Roif, Avishai
Format: Recurso digital
Language:
Published: Zenodo 2026
Subjects:
Online Access:https://doi.org/10.5281/zenodo.19828536
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866901215618531328
author Roif, Avishai
author_facet Roif, Avishai
contents <p class="MsoNormal">We introduce the B-test (ℬ), a new hypothesis test for real-world networks grounded in the Boundary Algebra (BA) framework. Building on the third paper of this series, which established that networks under BA constraints produce asymmetric normal degree distributions with skewness γ = cos(2π/k) determined by spectral class k ∈ {0, 2, 3, 4}, the B-test evaluates whether the observed skewness differences between network clusters are consistent with their theoretical spectral assignments. The test statistic is ℬ = Σ_{i<j} |γ*_i − γ*_j| · Z²_{ij}, where Z_{ij} is the standardized skewness difference and the weights |γ*_i − γ*_j| are fixed by the spectral classes. Under H₀, ℬ follows a weighted chi-squared distribution with analytically determined parameters (Satterthwaite approximation), requiring no bootstrap or permutation procedure. Simulation results demonstrate Type I error control at 0.035 and power exceeding 0.88 for deviations of Δγ ≥ 1.0, with power approaching 1.0 for larger deviations. The B-test provides the first ontologically grounded hypothesis testing framework for network science.</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_19828536
institution Zenodo
language
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle The B-Test: A Spectral Consistency Test for Real-World Networks Under Boundary Algebra
Roif, Avishai
B-test
Hypothesis Testing
Boundary Algebra
Spectral Consistency
Network Clusters
Satterthwaite Approximation
Subsampling Estimator
Skewness Difference
Ontological Distance
Network Science Methodology
Weighted Chi-Squared Distribution
Spectral Weights
<p class="MsoNormal">We introduce the B-test (ℬ), a new hypothesis test for real-world networks grounded in the Boundary Algebra (BA) framework. Building on the third paper of this series, which established that networks under BA constraints produce asymmetric normal degree distributions with skewness γ = cos(2π/k) determined by spectral class k ∈ {0, 2, 3, 4}, the B-test evaluates whether the observed skewness differences between network clusters are consistent with their theoretical spectral assignments. The test statistic is ℬ = Σ_{i<j} |γ*_i − γ*_j| · Z²_{ij}, where Z_{ij} is the standardized skewness difference and the weights |γ*_i − γ*_j| are fixed by the spectral classes. Under H₀, ℬ follows a weighted chi-squared distribution with analytically determined parameters (Satterthwaite approximation), requiring no bootstrap or permutation procedure. Simulation results demonstrate Type I error control at 0.035 and power exceeding 0.88 for deviations of Δγ ≥ 1.0, with power approaching 1.0 for larger deviations. The B-test provides the first ontologically grounded hypothesis testing framework for network science.</p>
title The B-Test: A Spectral Consistency Test for Real-World Networks Under Boundary Algebra
topic B-test
Hypothesis Testing
Boundary Algebra
Spectral Consistency
Network Clusters
Satterthwaite Approximation
Subsampling Estimator
Skewness Difference
Ontological Distance
Network Science Methodology
Weighted Chi-Squared Distribution
Spectral Weights
url https://doi.org/10.5281/zenodo.19828536