RQI-Audit: Statistical Certification of Random Number Generators Across Programming Languages
American Journal of Capstone Research · 2026
DOI: 10.70251/1318 ↗Kurz gefasst
Fachartikel in der Zeitschrift American Journal of Capstone Research, 2026. Verfasst von José Rodríguez Muñoz. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Random-number generators are commonly evaluated through separate statistical batteries, entropy summaries, collision measures, and serial-dependence diagnostics. These outputs are informative but difficult to compare across generator families, programming languages, and software builds. This study introduces the Randomness Quality Index (RQI), a calibrated 0-100 score designed to summarize multi-battery evidence while preserving diagnostic traceability, and evaluates RQI-Audit as an automated software-auditing workflow. A balanced factorial experiment crossed six generator families with six language environments: Python, C/C++, Java, JavaScript, Go, and Rust. For each of 36 generator-language cells, 177 replicate sequences of 5.64 x 10^6 unsigned 64-bit integers were generated under pinned runtime conditions, producing 6,372 replicate-level observations. Each sequence was evaluated using NIST SP 800-22 aggregate fail-count testing, TestU01 SmallCrush aggregate fail-count testing, entropy diagnostics, collision summaries, serial-dependence measures, and PractRand screening. Because RQI was bounded and upper-compressed, primary inference used rank-based and permutation-based methods. Generator family explained approximately 92% of observed RQI variation, while language environment and interaction terms each explained less than 2%. Controlled low-bit-bias and block-repetition corruptions produced high separability, with continuous -RQI reaching AUC values of 0.9993 and 0.9994, respectively. The results show that generator choice dominated language choice under the tested conditions and that RQI-Audit can support traceable regression monitoring without claiming formal cryptographic security.
Zitieren
APA
Muñoz, J. R. (2026). RQI-Audit: Statistical Certification of Random Number Generators Across Programming Languages. American Journal of Capstone Research. https://doi.org/10.70251/1318
Vancouver
Muñoz JR. RQI-Audit: Statistical Certification of Random Number Generators Across Programming Languages. American Journal of Capstone Research. 2026. doi:10.70251/1318
BibTeX
@article{Muoz2026,
author = {Muñoz, José Rodríguez},
title = {RQI-Audit: Statistical Certification of Random Number Generators Across Programming Languages},
journal = {American Journal of Capstone Research},
year = {2026},
doi = {10.70251/1318}
}
EUVIMED