Physics-informed phenomenology of safety invariance in learning-based control across thermal, electrical, physiological, and financial domains

Physics-informed phenomenology of safety invariance in learning-based control across thermal, electrical, physiological, and financial domains | Zenodo Skip to main Communities My dashboard Log in Sign up Zenodo is currently experiencing slowness and intermittent outages due to heavy automated traffic from bots and AI crawlers. We are aware of the problem, and our team is focused on stabilizing the service. Thank you for your patience. Published May 7, 2026 | Version 2 Preprint Open Physics-informed phenomenology of safety invariance in learning-based control across thermal, electrical, physiological, and financial domains Authors/Creators Serra-Taylor, Jose (Researcher) 1 Serra-Ramirez, Abbygail (Researcher) 2, 3 Irizarry-Borrero, Luis (Researcher) 2, 4 Show affiliations 1.

NVARIS, Puerto Rico 2. University of Central Florida 3.

Master of Science in Mechanical Engineering (MSME) Program 4.

Master in Robotics and Autonomous Systems (MS) Program Contributors Other: Perez-Torres, Solimar 1 Show affiliations 1.

NVARIS, Puerto Rico Description The safety-critical deployment of learning-based controllers requires reliable constraint satisfaction across heterogeneous dynamical regimes. One way to address this challenge within a physics-informed machine learning framework is to incorporate prior structural knowledge—such as forward-invariant feasibility constraints — directly into the control process. Here, we present a phenomenological study characterizing the macroscopic statistical signatures induced by a domain-agnostic constraint-enforcement layer across five safety-critical domains: thermal HVAC control, smart-grid frequency regulation, clinical CGM-based insulin safety, renewable microgrid management, and financial portfolio control. Across 842,511 historical samples spanning six orders of magnitude in timescale, we identify three reproducible signatures: (i) distributional truncation at feasibility boundaries, with no observed violations under enforcement; (ii) boundary redirection dynamics consistent with forward-invariant behavior; and (iii) systematic growth of intervention intensity as the safety boundary is approached. The zero-violation result corresponds to an exact 95% Clopper–Pearson upper bound of approximately 3.6 × 10⁻⁶ on the true violation probability. Prospective real-time validation in live market operations supports the persistence of these signatures under deployed conditions. Collectively, these observations support a physics-informed interpretation of enforced feasibility as boundary-limited probability flow compatible with reflected stochastic dynamics, while remaining agnostic to internal controller architectures, and suggest that physics-informed constraint enforcement provides an empirically grounded, domain-agnostic approach to safe learning-based control. This document is an updated and expanded version of the original preprint hosted on TechRxiv (DOI: 10.36227/techrxiv.177220350.08459501/v1). This version is being published on Zenodo to ensure continuous open access and citable stability during the TechRxiv platform migration. Files Serra_et_al_Physics-informed_phenomenology_of_safety_invariance.pdf Files (2.4 MB) Name Size Download all Serra_et_al_Physics-informed_phenomenology_of_safety_invariance.pdf md5:5fb2fd853149458de307ffbc791505cf 2.4 MB Preview Download Additional details Related works Is new version of Preprint: 10.36227/techrxiv.177220350.08459501/v1 (DOI) Dates Updated 2026-05-07 References Silver, D. et al. Mastering the game of Go with deep neural networks and tree search. Nature 529, 484–489 (2016). García, J. & Fernández, F. A Comprehensive Survey on Safe Reinforcement Learning. Journal of Machine Learning Research 16, 1437–1480 (2015). Schulman, J., Wolski, F., Dhariwal, P., Radford, A. & Klimov, O. Proximal Policy Optimization Algorithms. http://arxiv.org/abs/1707.06347 (2017). Achiam, J., Held, D., Tamar, A. & Abbeel, P. Constrained Policy Optimization. (2017). Berkenkamp, F., Turchetta, M., Schoellig, A. P. & Krause, A. Safe Model-based Reinforcement Learning with Stability Guarantees. http://arxiv.org/abs/1705.08551 (2017). Chow, Y., Nachum, O., Duenez-Guzman, E. & Ghavamzadeh, M. A Lyapunov-based Approach to Safe Reinforcement Learning. http://arxiv.org/abs/1805.07708 (2018). Ames, A. D. et al. Control Barrier Functions: Theory and Applications. in Control Barrier Functions: Theory and Applications (ed. EUCA) 3420–3431 (18th European Control Conference (ECC), Napoli, Italia, 2019). Zhao, W., He, T. & Liu, C. Model-free Safe Control for Zero-Violation Reinforcement Learning. in 5th Conference on Robot Learning (CoRL 2021) 784–793 (PMLR, London, UK, 2022). Clopper, C. J. & Pearson, E. S. the Use of Confidence or Fiducial Limits Illustrated in the Case of the Binomial. Biometrika 26, 404–413 (1934). Cinar, B. DiaData: An Integrated Large Dataset for Type 1 Diabetes and Hypoglycemia Research [Data set]. GitHub https://github.com/Beyza-Cinar/DiaData (2025). Clauset, A., Shalizi, C. R. & Newman, M. E. J. Power-law distributions in empirical data. SIAM Review vol. 51 661–703 at https://doi.org/10.1137/070710111 (2009). Risken, H. The Fokker-Planck Equation: Methods of Solution and Applications. (Springer, 1996). Python Developer (programmer3). Data Center Cold Source Control Dataset. Kaggle https://www.kaggle.com/datasets/programmer3/data-center-cold-source-control-dataset . Python Developer (programmer3). PMU Dataset for Power Line Fault Detection. Kaggle https://www.kaggle.com/datasets/programmer3/pmu-dataset-for-power-line-fault-detection . Python Developer (programmer3). Renewable Energy Microgrid Dataset. Kaggle https://www.kaggle.com/datasets/programmer3/renewable-energy-microgrid-dataset . Aroussi, R. yfinance: Download market data from Yahoo! Finance's API [Python library]. GitHub https://ranaroussi.github.io/yfinance/ (2026). Field, A. Discovering Statistics Using IBM SPSS Statistics. Statistics vol. 58 (Sage publications, 2018). Levine, S., Finn, C., Darrell, T. & Abbeel, P. End-to-end training of deep visuomotor policies. Journal of Machine Learning Research 17, 1-40 (2016). 225 Views 194 Downloads Show more details All versions This version Views Total views 225 225 Downloads Total downloads 194 194 Data volume Total data volume 559.4 MB 559.4 MB More info on how stats are collected.... Versions External resources Indexed in OpenAIRE Communities Keywords and subjects Keywords safety-critical control forward-invariant manifolds empirical scaling laws complex systems learning-based control Details DOI DOI Badge DOI 10.5281/zenodo.20073543 Markdown [](https://doi.org/10.5281/zenodo.20073543) reStructuredText .. image:: https://zenodo.org/badge/DOI/10.5281/zenodo.20073543.svg :target: https://doi.org/10.5281/zenodo.20073543 HTML <a href="https://doi.org/10.5281/zenodo.20073543"><img src="https://zenodo.org/badge/DOI/10.5281/zenodo.20073543.svg" alt="DOI"></a> Image URL https://zenodo.org/badge/DOI/10.5281/zenodo.20073543.svg Target URL https://doi.org/10.5281/zenodo.20073543 Resource type Preprint Publisher Zenodo Languages English Rights License Creative Commons Attribution Non Commercial Share Alike 4.0 International (CC-BY-NC-SA 4.0 No further description. Copyright 2026 Jose Serra-Taylor, Abbygail Serra-Ramirez, Luis Irizarry-Borrero. All rights reserved. Patent Pending. Citation Export Technical metadata Created May 7, 2026 Modified May 7, 2026 Jump up About About Policies Infrastructure Principles Projects Roadmap Contact Blog Blog Support Help FAQ Developers REST API OAI-PMH Contribute GitHub Donate Funded by Powered by CERN Data Centre & InvenioRDM Status Privacy policy Cookie policy Terms of Use This site uses cookies. Find out more on how we use cookies Accept all cookies Accept only essential cookies

Related documents

• Control as Safety (the Predictability Principle)#17568
• When technology awakens willingness but stalls action: the asymmetric psychological translation of managerial safety cognition.#317954
• Self-protective memory#251989
• Safety Invariance in Stochastic AI Moderation Pipelines: A Physics-Informed Empirical Study#174068
• La fiabilidad de la ciencia experimental#751933