Adversarial Stress Testing of the Correlated Ensemble Solver: Hidden Fragility, Arbitrage Bounds, Phase Transition Precision, and Information Paradox
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
Four adversarial stress tests are applied to the Correlated Ensemble Solver (CES), designed to expose failure modes that standard validation cannot detect. Test 1 (Hidden Fragility) probes whether the CES can quantify systemic risk invisible to naive single-regime models; Test 2 (Arbitrage Bounds) verifies that no combination of CES queries violates logical probability bounds, including at the boundary of positive semi-definiteness; Test 3 (Phase Transition Precision) measures the smoothness and N-scaling of the critical probability curve at the cascade threshold; Test 4 (Information Paradox) verifies that moving variables from an uncorrelated to a correlated group produces the correct monotone entropy signature. All four tests pass. The most striking result is Test 1: a hidden regime-switching scenario with 1% crisis trigger probability produces a cascade probability of 5.8 x 10^-3, while the normal-regime probability of approximately 10^-170 is below Monte Carlo resolution at any practical sample size. This demonstrates the hidden fragility structure that single-regime models cannot quantify. The phase transition curves are smooth to second-derivative magnitude of 4 x 10^-5 across all N tested, with the transition sharpening monotonically with N—the fingerprint of collective threshold behaviour approaching a discontinuity in the large-N limit. No logical violations are detected across 15 near-singular correlation matrices and 15 subset bound checks. The entropy signature of Test 4 is confirmed: probability change increases monotonically as variables move from the uncorrelated to the correlated group, with effective correlation tracking the information density shift continuously.
Fields of Study & Subjects
Complex Systems and Time Series Analysis
Financial Risk and Volatility Modeling
Banking stability
regulation
efficiency
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 28 March 2026
- Language
- EN
- Document type
- Preprint