What does this research mean for the field?

The refinement of the Irreversibility Axiom in Constrained Generative Systems distinguishes three consumption types and classifies all possible asymptotic behaviors into four dynamic regimes, revealing a structural boundary between systems that must terminate or cycle deterministically and those that allow persistent internal choice. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

The research aims to refine the Irreversibility Axiom and classify asymptotic behaviors in constrained generative systems based on consumption types.

March 4, 2026Open Access

Consumption Types and Dynamic Regimes in Constrained Generative Systems: Refinement of the Irreversibility Axiom and SCC Classification of Asymptotic Behavior

Key Points

The research aims to refine the Irreversibility Axiom and classify asymptotic behaviors in constrained generative systems based on consumption types.
Distinction of three consumption types: node-consumptive, arc-consumptive, and trajectory-consumptive.
Classification of asymptotic behaviors into four dynamic regimes using SCC decomposition.
Proof of completeness and minimality of the classification.
Establishment of a forced finiteness theorem based on action alphabet derived from continuous groups.
Establishment of four distinct dynamic regimes: deterministic cycle, branched attractor, fixed point, and finite acyclic trajectory.
Identification of a structural boundary between terminating and non-terminating systems.
Saturation in branched attractor regime is asymptotic rather than terminal.
Verification of structural properties of sub-limit dynamics across all regimes.

Abstract

We refine the Irreversibility Axiom (Axiom 2) of the Constrained Generative Systems framework by distinguishing three consumption types: node-consumptive (T1), arc-consumptive (T2), and trajectory-consumptive (T3), according to what structural element is removed or accumulated by the consumption function. We then classify all possible asymptotic behaviors of trajectories in a CGS with finite state space into exactly four dynamic regimes — deterministic cycle (R1), branched attractor (R2), fixed point (R3), and finite acyclic trajectory (R4) — via the strongly connected component (SCC) decomposition of the transition graph. The classification is proved complete and minimal: no fifth regime exists, and no regime is reducible to another. We establish that regime R2 (branched attractor) is accessible only in trajectory-consumptive systems, where the consumption function records history without removing nodes or arcs from the transition graph. This result identifies a structural boundary between systems that must terminate or cycle deterministically (node- and arc-consumptive) and systems that admit persistent internal choice (trajectory-consumptive). We prove a forced finiteness theorem: in any CGS whose action alphabet is derived from a continuous group under a distinguishability threshold L > 0, the number of operationally distinct classes is bounded by ⌈1/L⌉, independently of the algebraic structure of the generating group. This extends Axiom 1 from systems with intrinsically finite alphabets to systems where finiteness is forced by the limit. We verify that the four structural properties of sub-limit dynamics hold under all four regimes, with regime-specific manifestations. In particular, saturation in R2 is asymptotic (convergence of the visit distribution) rather than terminal (exhaustion of capacity), and the efficiency paradox manifests as topological confinement rather than premature termination.

KI fragen

Bookmark

View Full Paper