Internal Mathematical Consistency of the Correlated Ensemble Solver: Eight Self-Consistency Tests

Read full text Summarise with VIRAT
Authors: Oguike Ahaneku
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

A solver that claims to compute exact joint exceedance probabilities over large correlated variable sets must satisfy mathematical constraints that hold for any valid joint probability model, regardless of internal mechanism. These constraints provide a model-agnostic self-consistency framework that can validate or falsify a solver without reference to external ground truths or Monte Carlo comparison. Eight such tests are applied to the Correlated Ensemble Solver (CES). Seven tests are designed as necessary conditions; all seven pass. One test is an empirical characterisation of N-scaling behaviour rather than a necessary condition. The tests cover: positive semi-definiteness of correlation matrices, marginalisation consistency, permutation symmetry, convergence to the independence limit, monotonicity in correlation, exponential scaling, phase transition reproducibility, and the Bernoulli entropy inequality. Two structural properties are identified. First, the product rule for independent variables holds to 0.5% error at correlation 0.001, confirming correct implementation of the independence limit. Second, at high correlation (0.99) and N = 15, the ratio of joint to marginal probability is approximately 0.77, indicating that the CES implements a more complex dependence model than a simple Gaussian copula. The seven necessary-condition tests constitute a mathematical fingerprint consistent with a valid, internally coherent joint probability solver.

Fields of Study & Subjects

Financial Risk and Volatility Modeling Probability and Statistical Research Statistical Distribution Estimation and Applications

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
28 March 2026
Language
EN
Document type
Preprint