Conservation as Grammar: No-Cloning and Unitarity as Structural Theorems of a Guarded Linear Calculus, with a Holographic Resolution Bound
Year: 2026
Type: preprint
Abstract: In standard quantum mechanics, unitarity is a postulate — forced, once probability preservation and reversibility are granted, by the symmetry structure of pure states — and no-cloning is a theorem about the formalism's linear maps. We exhibit a mechanized, sorted, guarded linear calculus in which both statements are *structural* theorems: properties of which well-formed processes exist at all, prior to any dynamics. No derivation duplicates a linear token (`no_contraction_content`, machine-checked in Lean 4), and in the guarded kernel no derivation consumes one into nothing (`no_vanishing_restricted`). The two laws carry different strength — cloning is forbidden by the grammar alone, even where discard and duplication remain available for other sorts — and the law's scope is pinned rather than settled: the cloning ban is the law of the linear types with no closed inhabitant, every preparable type provably clonable, the information carriers provably not, one boundary case open — while conservation of content requires the kernel's discard ban — an asymmetry that mirrors the physical one between a kinematic law and a dynamical one. The unrestricted calculus is not a harmless generalization: there, information loss is *derived* (an explicit construction is machine-checked) and the unguarded dynamics provably diverge — pay forever, deliver nothing. We give a translation manual into the language of observables, tensor networks, and structural holography; for the finitary fragment the dictionary is backed by a semantic bridge — a functor into the category of completely positive maps, under which the kernel/outside boundary is the no-discard/discard boundary and the loss process denotes the trace — while the remaining rows are readings: a dictionary, not an isomorphism, whose functor mirrors the laws without deriving them. Finally, if the guarded tick is *spatialized* — an assumption the calculus does not make — combining it with the Bekenstein–Hawking area law yields the bulk voxel scale of the spacetime-foam literature — $L \approx (l_p^2 R)^{1/3} \approx 5\times10^{-15}\,$m for the observable radius $R$, a scaling due to Károlyházy and to Ng — a projection, not a theorem. The exponent belongs to the entropy bound; the guard fixes no scale.
Author: Jaye Timothy Marshall
Language: English
Subfield: Software
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7233938602
