Audit-Ready Refusal as Set-Invariance: Field-System Coupling under Finite-Horizon Viability

Read full text
Authors: Panagiotis Kalomoirakis
Cited by: 2

Abstract

What this isThis preprint introduces audit-ready refusal as a formal property of viable systems under finite capacity and time.Refusal is treated not as a heuristic or preference, but as a set-invariance constraint that preserves system viability under overload. Core contributionThe paper formalises a field–system coupling in which: a viability field defines an admissible region of operation (set invariance), and a system implements an explicit, typed refusal operator with traceable reasons. This coupling provides a minimal, implementation-agnostic bridge between thermodynamic viability laws and concrete system-level admission and routing decisions. Relation to prior workThe work builds on the Thermodynamic Right to Refusal (TRR), which establishes refusal as a necessity under overload, and on a companion manuscript that develops invariant-preserving maturation and internal governance required to realise this necessity in finite systems.Here, the focus is on the operational coupling that makes TRR invariance enforceable and auditable. Evidence and methodologyThe paper includes a reproducible bursty-flow benchmark with: proxy-style baselines, power-of-two choices routing, gate-only and gate+route ablations, and symmetry-breaking workloads (heterogeneous service rates). These experiments demonstrate when routing differences are identifiable, and clarify the trade-off between safety through refusal and throughput under governance constraints. Scope and positioningThis contribution is theoretical and systems-oriented. It provides formal semantics, invariance guarantees, and evidence methodology, without prescribing proprietary implementations or deployment-specific tuning. Intended audienceResearchers and practitioners in systems, control, networking, and compute governance are interested in overload safety, auditability, and finite-horizon viability.

Fields of Study & Subjects

Software System Performance and Reliability Formal Methods in Verification Reliability and Maintenance Optimization

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
18 January 2026
Document type
Preprint