Dataset: The Affective Thermodynamic Relationship — Empirical Evidence (Phases 1–4)

This dataset provides the complete empirical evidence supporting the study “The Affective Thermodynamic Relationship: An Architecture-Invariant Law of Interpretative Collapse in Large Language Models.” It includes all experimental data (Phases 1–4) and analysis pipelines:• Collapse probability (λ) across conflict levels (C = 1–5) • Temperature variance sweeps (σ² = 0.1–1.0) • GLMM, Arrhenius/Kramers, and Bayesian results • Bootstrap resampling distributions (n = 10,000) • Junk-Persona Prompt Injection (JPPI) outcomes and Affective Degradation Index (ADI) The dataset confirms an architecture-invariant thermodynamic law across GPT-4o, LLaMA-3, and Vendor-B systems.All materials comply with FAIR data principles and are released under CC-BY-4.0.