Introduction

Introduction

The Wilson Yang–Mills construction joins several kinds of mathematics: finite group-valued configurations and probability measures, effective actions and stationary variables, local coefficient spaces and equation quotients, continuum correlations, Hilbert-space fields and spectral readers. Recovering its conclusion requires these objects and operations to belong to one compatible construction. Equality of final formulas alone does not establish that differentiation, integration, calibration, quotienting and limiting operations preserve the information needed by the next stage.

This paper addresses that problem using Shadow Theory’s source–readout architecture and Source Lift protocol, originated by Jeremy Rodgers [3]. Rodgers directed the research through that framework and its worked examples; AI assisted its execution. A Source Lift constructs an admitted native support, its literal preparation and a realized sector containing the physical histories and auxiliary data. Typed readouts then recover the objects of the target derivation. Their sections establish reverse coverage, their fibres identify the distinctions removed, and their operation identities establish how the derivation passes through the realization. Reflective mathematics is the explicit accounting for these maps, domains, residuals and recovery claims. Section 2 introduces this framework before applying it.

The application is OpenAI’s underlying Lean Yang–Mills construction, available here through its supplied expanded mathematical translation [2]. The companion paper, A Wilson Yang–Mills Construction: A Natural-Language Mathematical Exposition, presents that construction on its own [1]. Both papers are included in the same Zenodo release, DOI 10.5281/zenodo.24173681, and each contains its essential mathematics. The companion emphasizes the construction itself. This paper develops the source–readout realization, the reflective analysis of its interfaces and the recovery proofs throughout the same derivation.

Mathematical contribution. The finite realization has an explicit section, a complete frame-fibre description and an exactly normalized measure pushforward. The family of histories is supported by an admitted fan, with a groupoid equivalence on each finite realized leaf. The reconstruction then proves frame-blindness of flags and budgets, ordered-derivative and activity-norm recovery, and quantitative recovery of an eliminated vertical discrepancy from the stationary residual. At the field stage it gives the curl/radial section and its gauge fibre, the calibrated coefficient reader, and the physical quotient defined by all spectator readings, with inclusions and derivatives descending to that quotient. The Hilbert-space, phase, product and collision-extension interfaces retain their actual domains and choices. These results explain precisely which operations are recovered, rather than treating every reduction as the same kind of inverse.

The full Yang–Mills calculations remain in the paper. They use Wilson’s lattice formulation [5], reflection positivity and reconstruction [11, 12, 13, 14, 15, 16], renormalization and cluster methods [17, 18, 19], and the conventional one-loop coefficient [20, 21]. The given physical inputs and the exact public conclusion are stated in Section 4; the assumption and dependency registers collect their scope. The earlier Einstein source–readout reconstruction [4] provides a model of exposition: define the primitives and maps, derive the operation, prove coverage, and identify the residual or information loss. Its gravitational results are not Yang–Mills premises.

Organization and reading conventions. The framework, public conclusion and analytic conventions fix the objects, target and normalizations. Native admission, the finite realization and the interface table then introduce the common witness. The construction follows dependency order: quadratic kernels; nonlinear geometry and budgets; marked integration and reset; stationary graphs, calibrated histories and convergence; labels, shell charges and physical detection; the full

inner continuum limit; phase and Wightman reconstruction; separated and nested products; collision extensions; Gaussian comparison; and the stress, trace and equation-sensor calculations. The common-family assembly collects their witnesses. Complete transport proofs and the assumption, amendment and correspondence registers are included as mathematical appendices.

At each stage the derivation is followed by its source–readout formulation: the realized record, reader, section or quotient, operation compatibility and recovery claim, with the relevant error or choice retained. Original equation tags are kept throughout. Input-line ranges refer to the preserved expanded source and support the complete coverage register. These references locate provenance; the local derivations supply the mathematics.