First-Order Recoverability Collapse in Self-Referential Information Decoders: The Operating Loop of an AI System as a Driven Nonequilibrium Steady State

Pieter van Rooyen

Home/Archive/arXiv

arXiv · 2606.24861

First-Order Recoverability Collapse in Self-Referential Information Decoders: The Operating Loop of an AI System as a Driven Nonequilibrium Steady State

Pieter van Rooyen

View original source ↗Read source document ↗Export metadata ↓

Abstract

What kind of physical object is an artificial-intelligence system: a machine in the sense a refrigerator is -- dissipating free energy while holding an order imposed from outside -- or a dissipative structure in the sense a convection cell is -- an ordered state that exists only under throughput and loses stability past a critical drive? We argue the answer is split, as it is for the living cell and the star -- the trained artifact is a machine, quenched and storable; the operating loop is a dissipative structure in the informational sense -- and develop the framework in which the loop's classification becomes decidable. Modeling systems that couple inference to irreversible action as finite-capacity decoders under sustained informational driving, we characterize recoverable operation by a feasibility margin, local invertibility, and a stability diagnostic that diverges as capacity saturates. Making the feedback of uncertified output onto load explicit converts this continuous transition into a first-order one at mean-field level, sharpening in the fleet limit: lucid and collapsed states coexist in a cusp-organized bistable region with closed-form spinodals, collapse pre-empts the divergence, recovery is hysteretic, and for ungatedness alpha >= 1 load reduction alone cannot restore operation; reset restores only what is archived, making certification the operative stability lever. Cascades are subcritical branching with mean-field exponent 3/2 and a cutoff set by the grounded fraction of input. An instrumented real-workload pipeline experiment exhibits the collapse just below the spinodal computed from the measured service law, the backlog-delayed hysteretic recovery, and the cascade statistics. This supplies a statistical-mechanics account of the "metastable failures" documented in large-scale distributed systems, identifying recoverable dissipation as the stability criterion.

Explore related subjects

cond-mat.stat-mech cs.IT math.IT

Keep this discovery

BibTeX RIS

Pieter van Rooyen. 2026-06-23. First-Order Recoverability Collapse in Self-Referential Information Decoders: The Operating Loop of an AI System as a Driven Nonequilibrium Steady State. https://arxiv.org/abs/2606.24861

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Universal sampling of spin systems across quenched disorder

Statistical physics extracts macroscopic laws by averaging over the many microscopic degrees of freedom of a system. Disordered systems demand a second and far harder average, one over the quenched randomness itself. The classic analytical routes, the replica and cavity methods, become uncontrolled outside mean-field or tree-like limits, and conventional numerical algorithms like parallel tempering require expensive, independent equilibration for every disorder realization. In this work, we introduce a universal neural variational framework that amortizes inference across the disorder ensemble, eliminating both the need for per-instance Markov chain equilibration and the cost of retraining instance-specific variational ansatzes. Built on an encoder-decoder Transformer architecture, after training once, it produces an explicit approximation to the Boltzmann distribution given previously unseen disorder realizations without further optimization. We validate this framework on 2D Edwards-Anderson models, and apply it to the random-bond Ising model, successfully capturing the Binder cumulant crossings near the Nishimori multicritical point. These results shift the object of variational inference from the single instance to the disorder ensemble, opening a route to frustrated many-body systems where instance-by-instance computation is prohibitive.

cond-mat.stat-mech↗

arXiv2026

Information-Theoretic Characterization of Macroscopic Chaos Emerging from the Chemical Master Equation

Open chemical reaction networks exhibit stochastic concentration dynamics at finite system sizes, whereas their macroscopic limit is governed by deterministic rate equations that can display chaos. In this Letter, we show theoretically that a rate of information loss constructed from two-time mutual information recovers the Kolmogorov-Sinai entropy in the deterministic limit. We verify this result through numerical simulations of a Markov jump process for a three-species system involving seven reactions.

cond-mat.stat-mech↗

arXiv2026

Orientational order on non-orientable domains

We study the statistical properties of passive and active many-body systems with orientational degrees of freedom on non-orientable domains. By rephrasing topological constraints as non-local symmetry relations on an orientable double-cover, we show that non-orientability eliminates global rotational soft modes without acting like an external field. In a passive XY model, this results in topological caging, where orientational fluctuations that exhibit conventional diffusive behavior on a torus saturate on a Klein bottle to a finite value that we compute exactly in the thermodynamic limit. In models of active self-propelled particles with orientational degrees of freedom, topological caging persists despite continuously changing interaction neighborhoods. In an active Ising spin model, non-orientability enforces the coexistence of ordered anti-parallel domains with vanishing global polar order, a state that is absent on orientable domains.

cond-mat.stat-mech↗