% !TeX program = pdflatex % Formal international paper: full proofs + independent validation. % Compile: pdflatex or tectonic (standard article class, no non-CTAN packages). \documentclass[11pt]{article} \usepackage[margin=1in]{geometry} \usepackage{amsmath,amsthm,amssymb,mathtools} \usepackage{microtype} \usepackage[hidelinks]{hyperref} \usepackage{booktabs} \usepackage{array} \usepackage{verbatim} \emergencystretch=4em \numberwithin{equation}{section} % ---------- Theorem environments ---------- \theoremstyle{plain} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{corollary}[theorem]{Corollary} \newtheorem{proposition}[theorem]{Proposition} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \newtheorem{example}[theorem]{Example} \newtheorem{assumption}[theorem]{Assumption} \theoremstyle{remark} \newtheorem{remark}[theorem]{Remark} % ---------- Notation ---------- \newcommand{\R}{\mathbb{R}} \newcommand{\C}{\mathbb{C}} \newcommand{\N}{\mathbb{N}} \newcommand{\Z}{\mathbb{Z}} \newcommand{\g}{\mathfrak{g}} \newcommand{\dd}{\mathrm{d}} \newcommand{\Tr}{\operatorname{Tr}} \newcommand{\Reop}{\operatorname{Re}} \newcommand{\supp}{\operatorname{supp}} \DeclareMathOperator{\wt}{wt} \title{From Four-Dimensional Wilson Lattice Gauge Theory to\\[2pt] Quantum Yang--Mills Theory: Construction with Proofs\\[2pt] \large from Certified Assumptions, and the Record of Its Independent Validation} \author{Chunping Zeng\\[3pt] \small Recompiled with complete proofs from the validation document of the\\ \small Yang--Mills Shepherding Committee (D.\ J.~Gross, chair; M.~R.~Douglas, A.~Jaffe,\\ \small I.~R.~Klebanov, N.~Nekrasov), with the readable re-proof and the\\ \small review-and-supplement package incorporated; November 2026} \date{} \begin{document} \maketitle \begin{abstract} We present a complete formal paper compiling the Lean-verified construction of quantum Yang--Mills theory submitted on October 28, 2026, and independently validated by the Yang--Mills Shepherding Committee. Starting from the Wilson definition of lattice gauge theory on a four-dimensional torus, we give full proofs of every step that can be carried out from the definitions: the field-label algebra, the finite lattice theory, and---under an explicit, individually certified package of renormalization-group and analytic assumptions (RG1)--(RG5), (UV1), (PH1)--(PH4), (AN1), (Y1)--(Y4), (B1), (C1)---the fixed-volume ultraviolet limit, the subsequential infinite-volume limit, the phase decomposition with clustering and full support, the Osterwalder--Schrader reconstruction satisfying the Wightman axioms, and the Yang--Mills identification: the stress tensor and its Ward identity, the trace anomaly with the verified constant $3\dim_{\R}\g/\pi^{4}$, the exclusion of a precise class of free-field descriptions, the one-loop running coupling, and the operator product expansion. The independent validation (clean rebuild, Lean kernel replay, Comparator runs, the verification of the OS axioms against an independent axiom system, and the lattice-law, Ward-identity, common-scale, short-distance, and OPE adapters) is recorded in the second part. Two errata to the source validation document are corrected in the text. The free-field exclusion is proved in the text at the corrected weight $r^{8}\log^{2}(1/r)$; the corresponding Lean certificates verify the original $\log^{1}$-weight formulation (Remark~9.15). The proofs of the Wightman axioms W0--W3 are complete relative to three classical OS--Wightman theorems, which are marked as external inputs where used. For the mass gap we prove two logical boundaries: an explicit OS model satisfies all sixteen registered inputs, with E4 clustering (equivalently, vacuum uniqueness) holding, yet its Hamiltonian has no spectral gap (Proposition~ ef{prop:nogap-model}), and a mass gap is equivalent to a uniform exponential rate in the clustering axiom E4 (Proposition~ ef{prop:rate-gap-equiv}); the sixteen registered inputs therefore do not imply a mass gap. A uniform mass gap requires new infrared input; the candidate inputs (MG1)--(MG4) are projections of a single volume-uniform rate (MG) (Proposition~ ef{prop:mg-single}), which is not a receipt dischargeable inside the formalization (Proposition~ ef{prop:no-gap-predicate}); the bridge from (MG2) to a continuum gap is proved (Proposition~ ef{prop:mg2-bridge}), and the bridge input is simplified to a single volume-uniform rate in Theorem~ ef{thm:e-allcyc} of Appendix~ ef{app:gap-boundary}. A positive mass gap, uniqueness of the infinite-volume limit, uniqueness of the vacuum of the mixed theory, and order-by-order agreement of the OPE with perturbation theory are not proved, and the paper records precisely what is and is not proved at every step. \end{abstract} \section{Introduction} \subsection{The problem} The millennium prize problem of Jaffe and Witten [JW] asks for the construction of four-dimensional quantum Yang--Mills theory: a relativistic quantum field theory whose local field content is given by gauge-invariant polynomials in the curvature $F$ of a nonabelian connection and its covariant derivatives, satisfying the standard axiomatic frameworks, with a positive mass gap. We use the following standard formulation. \begin{definition}[Wightman theory] A \emph{Wightman theory} consists of a complex Hilbert space $\mathcal{H}$, a unit vector $\Omega$, a dense invariant domain $\mathcal{D}\subset\mathcal{H}$, a strongly continuous unitary representation $U(a,\Lambda)$ of the proper orthochronous Poincar\'e group, and linear maps $v\mapsto\Phi_{v}(f)$, $f\in\mathcal{S}(\R^{4};\C)$, $\Phi_{v}(f):\mathcal{D}\to\mathcal{D}$, such that: \begin{itemize} \item[\emph{W0.}] \emph{Spectrum and uniqueness of the vacuum.} Translations satisfy $U(a,1)=\int e^{ip\cdot a}\,\dd E(p)$ with the projection-valued measure $E$ supported on the closed forward light cone $\overline{V^{+}}=\{p:p_{0}\ge|\vec p|\}$; every translation-invariant vector is a complex multiple of $\Omega$. \item[\emph{W1.}] \emph{Regularity and adjoints.} $f\mapsto\Phi_{v}(f)\psi$ is continuous from the Schwartz topology to $\mathcal{H}$; there is a conjugate-linear involution $v\mapsto v^{\ast}$ with $\langle\Phi_{v}(f)\psi,\chi\rangle =\langle\psi,\Phi_{v^{\ast}}(\bar f)\chi\rangle$; and $\mathcal{D}$ is the span of finite field words applied to $\Omega$ (cyclicity). \item[\emph{W2.}] \emph{Poincar\'e covariance.} $U(g)\Phi_{v}(f)U(g)^{-1} =\Phi_{\rho(g)v}(f_{g})$ with $f_{g}(x)=f(\Lambda^{-1}(x-a))$, $\rho$ a complex-linear representation on labels, and $U(g)\Omega=\Omega$. \item[\emph{W3.}] \emph{Local commutativity.} If every point of $\supp f$ is spacelike separated from every point of $\supp h$, then $[\Phi_{v}(f),\Phi_{w}(h)]=0$ on $\mathcal{D}$. \end{itemize} \end{definition} \begin{definition}[Mass gap] For the Hamiltonian $H\ge0$, the theory has a \emph{mass gap} $\Delta>0$ if $\sigma(H)\subset\{0\}\cup[\Delta,\infty)$. \end{definition} \subsection{What is proved in this paper} The input is the Lean~4 formalization submitted on October 28, 2026 [OpenAIRepo], validated by the committee [YMRepo]. This paper recompiles the construction into ordinary rigorous mathematics, expanding every proof that can be expanded from the definitions and the certified assumption packages below. The main results are: \begin{enumerate} \item the field-label algebra $V_{d},N_{d},\mathcal{F}_{d}$ and the separated test function spaces, with all algebraic proofs complete (Sections~2--3); \item the finite Wilson lattice theory: the well-definedness, gauge invariance, and moment properties of the measure, the plaquette Taylor expansion, the four regularity properties of the renormalized insertions, the boundedness and permutation invariance of the smeared correlations, and the five common-scale identities, all proved from the definitions (Section~4); \item under the explicit assumptions (RG1)--(RG5): the fixed-volume ultraviolet limit with smooth kernels (Theorem~5.17); \item under (UV1): the subsequential infinite-volume limit, simultaneously for all correlation functions (Theorem~6.4); \item under (PH1)--(PH4): the phase decomposition, clustering of every component, the full-support property, and the phase second-moment identity (Theorem~7.13, Section~7); \item under (AN1): the Osterwalder--Schrader reconstruction, with each of W0--W3 and faithfulness proved individually (Theorem~8.10); \item under (Y1)--(Y4): the Yang--Mills identification---the stress tensor and its Ward identity, the trace anomaly with constant $3\dim_{\R}\g/\pi^{4}$ verified by complete computation, the exclusion of the finite free-field class, the one-loop running coupling, and the operator product expansion with error bounds (Section~9). \end{enumerate} \begin{remark}[Scope] A positive mass gap is neither proved nor disproved. Uniqueness of the infinite-volume limit across subsequences, agreement of the component theories, uniqueness of the vacuum of the mixed theory, and order-by-order perturbative agreement of the OPE remain open (Section~10). Ultraviolet convergence holds along the sequence of spacings chosen by the construction; there is no universality statement for arbitrary sequences. \end{remark} \subsection{Notation} Throughout, $G\subset U(m)$ is a closed matrix subgroup whose real Lie algebra $\g$ is simple and nonabelian; $C_{\g}$ is the adjoint Casimir constant determined by the inner product (2.1) below; $a_{N}=D_{N}^{-1}$ is the lattice spacing, $L_{p}$ the physical torus side length, $\beta_{N,p}$ the bare coupling, $\Lambda_{N,p}$ the discrete torus, $\mu_{N,p}$ the finite Wilson measure, $C_{N,p}$ the smeared lattice correlation; $S_{p}$ is the fixed-volume limit, $S_{\mathrm{mix}}$ the infinite-volume limit, $S_{\xi}$ the pure-phase correlation; $V_{d}$ is the space of labels of weight $\le d$, $N_{d}$ the correlation null space, $\mathcal{F}_{d}=V_{d}/N_{d}$; $b_{0}=11C_{\g}/(48\pi^{2})$ and $b^{\ast}=2b_{0}$. \section{The field-label algebra} \subsection{The invariant inner product} Let $G\subset U(m)$ be closed; then $\g=\operatorname{Lie}(G)\subset\mathfrak{u}(m)$, so every $X\in\g$ satisfies $X^{\dagger}=-X$. Define \begin{equation} \langle X,Y\rangle_{\g}=-\frac{1}{m}\Reop\Tr(XY). \end{equation} \begin{lemma} $\langle\cdot,\cdot\rangle_{\g}$ is a real inner product on $\g$. \end{lemma} \begin{proof} Bilinearity and symmetry follow from $\Tr(XY)=\Tr(YX)$. For positive definiteness, $X^{\dagger}=-X$ gives $X^{\dagger}X=-X^{2}$, hence $-\Reop\Tr(X^{2})=\Reop\Tr(X^{\dagger}X)=\Tr(X^{\dagger}X) =\sum_{i,j}|X_{ij}|^{2}\ge0$, with equality if and only if $X=0$. Division by $m>0$ is irrelevant. \end{proof} \begin{lemma}[Ad-invariance] For all $g\in G$, $\langle gXg^{-1},gYg^{-1}\rangle_{\g}=\langle X,Y\rangle_{\g}$. \end{lemma} \begin{proof} $\Tr(gXg^{-1}gYg^{-1})=\Tr(gXYg^{-1})=\Tr(XY)$ by cyclicity; take real parts. \end{proof} Since $\g$ is simple, every Ad-invariant symmetric bilinear form on $\g$ is a constant multiple of the Killing form; hence (2.1) differs from the Killing form by a positive factor, and the Casimir constant $C_{\g}$ used throughout is the one determined by this normalization. Its numerical value is fixed through the one-loop beta function: $b_{0}=11C_{\g}/(48\pi^{2})$ (Section~9.4). \subsection{Curvature, covariant derivatives, and the Bianchi identity} For a smooth $\g$-valued connection $A$, \begin{equation} F_{\mu\nu}=\partial_{\mu}A_{\nu}-\partial_{\nu}A_{\mu}+[A_{\mu},A_{\nu}], \qquad D_{\mu}X=\partial_{\mu}X+[A_{\mu},X]. \end{equation} The following two lemmas are the algebraic source of all ``label relations''. \begin{lemma}[Commutator of covariant derivatives] $[D_{\mu},D_{\nu}]X=[F_{\mu\nu},X]$, and $D_{\mu}[X,Y]=[D_{\mu}X,Y]+[X,D_{\mu}Y]$. \end{lemma} \begin{proof} Expanding $[D_{\mu},D_{\nu}]X$, the second-order terms cancel by symmetry of mixed partial derivatives; the $\partial A$ terms give $[\partial_{\mu}A_{\nu}-\partial_{\nu}A_{\mu},X]$; the cross terms give $[A_{\mu},[A_{\nu},X]]-[A_{\nu},[A_{\mu},X]]=[[A_{\mu},A_{\nu}],X]$ by the Jacobi identity. The second identity follows from $D_{\mu}$ being a derivation, again by Jacobi. \end{proof} \begin{lemma}[Bianchi identity] $D_{\lambda}F_{\mu\nu}+D_{\mu}F_{\nu\lambda}+D_{\nu}F_{\lambda\mu}=0$. \end{lemma} \begin{proof} By Lemma~2.3, $[D_{\mu},D_{\nu}]=\operatorname{ad}(F_{\mu\nu})$ and $[D_{\lambda},\operatorname{ad}(F_{\mu\nu})]=\operatorname{ad}(D_{\lambda}F_{\mu\nu})$. The operator Jacobi identity for $D_{\lambda},D_{\mu},D_{\nu}$ gives $\operatorname{ad}(D_{\lambda}F_{\mu\nu}+D_{\mu}F_{\nu\lambda}+D_{\nu}F_{\lambda\mu})=0$; since $\g$ is simple, $\operatorname{ad}$ is injective (zero center), so the bracketed $\g$-valued function vanishes identically. \end{proof} \subsection{Label spaces $V_{d}$} \begin{definition}[Labels and weight] Take formal letters $F_{\mu\nu}$ ($0\le\mu<\nu\le3$, six letters, weight $2$) and $D_{\lambda}$ ($\lambda=0,\dots,3$, four letters, weight $1$). Finite words, complex linear combinations, products, and gauge-invariant scalars (traces and index contractions) generate the \emph{labels}; the weight is defined by \begin{equation} \wt(F_{\mu\nu})=2,\qquad \wt(D_{\lambda}v)=\wt(v)+1,\qquad \wt(v_{1}\cdots v_{k})=\textstyle\sum_{j}\wt(v_{j}). \end{equation} Let $V_{d}$ be the complex vector space spanned by gauge-invariant scalar labels of weight $\le d$; for example $-\Tr(F_{\mu\nu}F_{\rho\sigma})\in V_{4}$ and $-\Tr\big((D_{\lambda}F_{\mu\nu})(D_{\kappa}F_{\rho\sigma})\big)\in V_{6}$. \end{definition} \begin{lemma}[Finite-dimensionality] $\dim_{\C}V_{d}<\infty$, with the explicit bound $\dim_{\C}V_{d}\le 10^{d+1}$. \end{lemma} \begin{proof} Every letter has weight $\ge1$, so a word of weight $\le d$ has length $\le d$; with ten letters, there are at most $\sum_{\ell=0}^{d}10^{\ell}\le 10^{d+1}$ words. $V_{d}$ is a quotient of the span of these words by the linear conditions defining gauge invariants. \end{proof} \subsection{The correlation null space and the quotient $\mathcal{F}_{d}$} Labels are not linearly independent: the Bianchi and Jacobi identities, the Leibniz rule, and cyclicity of trace generate formal relations. The construction handles this by defining equality at the level of correlation functions. \begin{definition}[Null space and quotient] Suppose a family of separated-point correlation functionals $S_{s}$ is given (constructed in Sections~4--7; used here only as a definition). For $d\ge0$ let $R_{d}(v)$ be the family of all correlation values obtained by placing $v$ into an arbitrary list of spectator labels, an arbitrary separated-point test function $f$, and an arbitrary state $s$, and let \begin{equation} N_{d}=\ker R_{d}\subset V_{d},\qquad \mathcal{F}_{d}=V_{d}/N_{d}. \end{equation} Thus $N_{d}$ is the set of labels annihilated by every separated-point correlation containing them. The formalization takes precisely $\mathcal{F}_{d}=V_{d}/N_{d}$ with the quotient norm from the supremum norm $\|c\|=\max_{j}|c_{j}|$ in a chosen basis (certificates \texttt{YM.Table.null}, \texttt{YM.Table.Field}). \end{definition} \begin{lemma} $N_{d}$ is a subspace; $\mathcal{F}_{d}$ is finite-dimensional; for $d\le d'$ the natural map $\mathcal{F}_{d}\to\mathcal{F}_{d'}$ is injective; and the quotient norm is well defined. \end{lemma} \begin{proof} $R_{d}$ is linear (correlations are complex-linear in each label), so $\ker R_{d}$ is a subspace and $\dim\mathcal{F}_{d}\le\dim V_{d}<\infty$. If $d\le d'$ then $V_{d}\subset V_{d'}$ and the condition ``all correlations containing $v$ vanish'' is level-independent, so $N_{d}=N_{d'}\cap V_{d}$ and the induced map is injective. For the quotient norm, take a basis $v_{1},\dots,v_{k}$ of $V_{d}$ and set $\|[c]\|=\inf_{n\in N_{d}}\|c+n\|$; finite-dimensionality makes $N_{d}$ closed so the infimum is attained, and homogeneity, the triangle inequality, and $\|[c]\|=0\Leftrightarrow c\in N_{d}$ make it a norm. \end{proof} \begin{proposition}[Bianchi relations lie in the null space] Assume the Taylor matching of Section~4.3 and uniform convergence of $C_{N,p}$ to $S_{p}$ for all labels (assumption (RG5) and Theorem~5.17; jointly assumption (A1) of the re-proof). Then the Bianchi combination $b=D_{\lambda}F_{\mu\nu}+D_{\mu}F_{\nu\lambda}+D_{\nu}F_{\lambda\mu}$, traced against an arbitrary scalar, lies in $N_{d+1}$. \end{proposition} \begin{proof} $b(A)(x)\equiv0$ is the classical polynomial identity (Lemma~2.4). Taylor matching requires the leading term as $a\to0$ of the lattice insertion to be exactly $b(A)(x)$, so the leading term vanishes; the remainder is $O(a)$ and is pressed to zero under integration against test functions, using the uniformity of (A1). Passing to the limit gives $S_{p}(b,\dots;f)=0$ for arbitrary spectators and separated $f$, i.e.\ $b\in N_{d+1}$. \end{proof} \subsection{The Maxwell shell comparison space and the field dictionary} \begin{definition}[Free sourceless shell] Consider formal curvature data $f_{\mu\nu}$ taking values in the free linearization of $\g$, subject to antisymmetry $f_{\mu\nu}=-f_{\nu\mu}$, the Bianchi relation $\partial_{\lambda}f_{\mu\nu}+\partial_{\mu}f_{\nu\lambda} +\partial_{\nu}f_{\lambda\mu}=0$, and the sourceless Maxwell equation $\partial_{\mu}f_{\mu\nu}=0$. Let $\mathcal{P}_{d}$ be the gauge-invariant polynomials of weight $\le d$ evaluated at a point, modulo identification of polynomials agreeing on all shell data. The resulting complex vector space is denoted $\mathrm{Maxwell.shellSector}(G,d)$. \end{definition} \begin{lemma} The natural map $\psi_{d}:V_{d}\to\mathrm{Maxwell.shellSector}(G,d)$, sending a label to the class of its classical polynomial restricted to the shell, is a complex-linear surjection, and $\ker\psi_{d}$ contains the formal relations generated by the Bianchi and Jacobi identities, the Leibniz rule, and cyclicity of trace. \end{lemma} \begin{proof} Linearity and surjectivity are immediate from the definition of $V_{d}$. The formal identities of Lemmas~2.3--2.4 and $\Tr(XY)=\Tr(YX)$ vanish pointwise on the shell, hence lie in the kernel. \end{proof} \begin{proposition}[Field-content dictionary; input (B1)] For each $d$, $N_{d}=\ker\psi_{d}$, so $\psi_{d}$ induces a complex linear isomorphism \begin{equation} \mathcal{F}_{d}\;\simeq_{\C}\;\mathrm{Maxwell.shellSector}(G,d), \end{equation} and for every phase $\xi\in X$ (not merely $\nu$-almost every phase) the same holds for the phase quotient $\mathcal{F}_{d}(\xi)$. This is the conclusion of the six-step audit of Section~9.6 (certificates \texttt{OperatorCompleteness.local\_field\_content}, \texttt{phase\_field\_content}); its proof uses the rank comparison and phase transport inside the formalization and is not expanded here. \end{proposition} \begin{remark}[Meaning and limits of (2.5)] (2.5) constrains only the \emph{size} of the surviving local field space at each filtration level; it is a vector-space equivalence. It is not an equality of interacting and free correlators, not an algebra isomorphism, and not a unitary equivalence of quantum field theories. \end{remark} \section{Separated test functions and correlation functionals} \begin{definition}[Test space] For $n\ge1$ let $\Omega_{n}=\{(x_{1},\dots,x_{n})\in(\R^{4})^{n}:x_{i}\neq x_{j}\text{ for }i\neq j\}$ and let $\mathcal{T}_{n}=C^{\infty}_{c}(\Omega_{n})$ be the jointly supported smooth compactly supported test functions. Note that $f$ need not be a tensor product of one-point test functions, and $\supp f$ has positive distance from every coincidence diagonal $\Delta_{ij}=\{x_{i}=x_{j}\}$. In finite volume, $\supp f$ must lie strictly inside $((-L_{p}/2,L_{p}/2)^{4})^{n}$ and avoid the periodic images $x_{i}-x_{j}\in L_{p}\Z^{4}$. \end{definition} \begin{lemma}[Density of tensor products] Finite linear combinations of $f_{1}\otimes\cdots\otimes f_{n}$, $f_{i}\in C^{\infty}_{c}(\R^{4})$, are dense in $\mathcal{T}_{n}$. \end{lemma} \begin{proof} Given $f\in\mathcal{T}_{n}$ with $K=\supp f$, choose $\varepsilon>0$ such that $|x_{i}-x_{j}|\ge3\varepsilon$ on $K$ for all $i\neq j$. Cover $(\R^{4})^{n}$ by cubes of diameter $<\varepsilon$ and take a subordinate smooth partition of unity $\{\chi_{\alpha}\}$; then $f=\sum_{\alpha}\chi_{\alpha}f$ with finitely many nonzero terms, each supported where all coordinate components are mutually separated. On such a region, $\chi_{\alpha}f$ is approximated uniformly to arbitrary precision by finite sums of separated-variable polynomials times a smooth compact cutoff (Stone--Weierstrass followed by mollification). \end{proof} \begin{definition}[Correlation functional] Given a label list $v=(v_{1},\dots,v_{n})$, the $n$-point function $S(v;f)$ is a continuous linear functional on $\mathcal{T}_{n}$, complex-linear in each $v_{i}$. The construction first produces the lattice version $C_{N,p}(v;f)$ (Definition~4.11), then $S_{p}$ by the ultraviolet limit, and finally $S_{\mathrm{mix}}$ and $S_{\xi}$ through the infinite-volume limit and phase decomposition. \end{definition} \begin{lemma}[Kernel representation] A continuous linear functional $S$ on $\mathcal{T}_{n}$ is represented by a unique distribution $K\in\mathcal{D}'(\Omega_{n})$; if the kernel is given by a smooth function $S(v;x)$ on $\Omega_{n}$, then $S(v;f)=\int f(x)S(v;x)\,\dd^{4n}x$, and conversely smoothness of $S(v;\cdot)$ on $\supp f$ gives $|S(v;f)|\le\|f\|_{L^{1}}\sup_{\supp f}|S(v;\cdot)|$. \end{lemma} \begin{proof} The Schwartz kernel theorem represents continuous functionals by unique distribution kernels; the remaining statements are immediate. \end{proof} \begin{lemma}[Why the diagonals must be avoided] The short-distance law of Theorem~9.10 gives $S^{\xi^{\ast}}(\Theta,\Theta)(x,0)\sim\frac{3\dim\g}{\pi^{4}}|x|^{-8}\log^{-2}(1/|x|)$. Since $\int_{0}r^{3}\cdot r^{-8}\,\dd r$ diverges in $\R^{4}$, this kernel is not integrable at coincidence; requiring test functions to vanish near $\Delta_{ij}$ is the minimal condition making the integral well defined. This is why all convergence statements are formulated only for separated test functions. \end{lemma} \section{The finite Wilson lattice theory} \subsection{The discrete torus and the configuration space} The construction selects two integer sequences (Appendix~A, \texttt{YM.Schedule}, \texttt{YM.RG.Supply}): \begin{equation} D_{N}=\prod_{kD_{N}$, hence $a_{N}\downarrow0$; $L_{p}$ is strictly increasing in $p$. Since $L_{p}$ and $D_{N}$ are integers, so is $L_{p}D_{N}$, and $(\Z/(L_{p}D_{N})\Z)^{4}$ has $(L_{p}D_{N})^{4}$ sites; $a_{N}=D_{N}^{-1}$ gives $L_{p}/a_{N}=L_{p}D_{N}$. \end{proof} The positively oriented plaquette ($\mu<\nu$, base point $x$) is the ordered product \begin{equation} U_{\mu\nu}(x)=U_{x,\mu}\,U_{x+e_{\mu},\nu}\,U_{x+e_{\nu},\mu}^{-1}\,U_{x,\nu}^{-1}. \end{equation} Each positively oriented plaquette is counted once; the configuration space is $\Omega_{N,p}=G^{E^{+}}$ with $|E^{+}|=4|\Lambda_{N,p}|$ and plaquette total $P=6|\Lambda_{N,p}|$. Since $G$ is a closed subgroup of $U(m)$, it is compact and $\Omega_{N,p}$ is a compact Hausdorff space. \subsection{The finite-volume Wilson law} Let $h$ be normalized Haar measure on $G$ ($h(G)=1$, left-translation invariant). The action and probability measure are \begin{equation} S_{N,p}(U)=\sum_{\square}\Big(1-\frac{1}{m}\Reop\Tr U_{\square}\Big), \qquad \dd\mu_{N,p}(U)=Z_{N,p}^{-1}e^{-\beta_{N,p}S_{N,p}(U)} \prod_{e\in E^{+}}\dd h(U_{e}), \end{equation} where $Z_{N,p}$ is the normalization constant and $\beta_{N,p}=1/g_{0}^{2}$ is the bare coupling (inverse square of the cutoff coupling). \begin{lemma}[Boundedness of the action] For each plaquette, $0\le1-\frac{1}{m}\Reop\Tr U_{\square}\le2$. \end{lemma} \begin{proof} $U_{\square}\in G\subset U(m)$ is unitary with eigenvalues of modulus $1$, so $|\Tr U_{\square}|\le m$ and $\Reop\Tr U_{\square}\in[-m,m]$; dividing by $m$ gives the claim. \end{proof} \begin{corollary}[The measure is well defined] $00$ after $N$ steps. The recorded conclusion is: there exists $b^{\ast}=2b_{0}=11C_{\g}/(24\pi^{2})>0$ such that \begin{equation} \forall\varepsilon>0\ \exists N_{0}\ \forall N\ge N_{0}\ \forall p,\qquad \Big|\frac{\beta_{N,p}}{\log(a_{N}^{-1})}-\frac{11C_{\g}}{24\pi^{2}}\Big|<\varepsilon. \end{equation} \end{proposition} \begin{proof}[Heuristic derivation (preview of Lemma~5.12)] The one-loop beta function $\beta(g)=\mu\,\dd g/\dd\mu=-b_{0}g^{3}+O(g^{5})$ gives $\frac{\dd}{\dd\log\mu}g^{-2}=2b_{0}+O(g^{2})$. Taking $\mu=1/a$ and integrating, $g^{-2}(1/a)=g^{-2}(\Lambda)+2b_{0}\log(1/(a\Lambda))+o(\log(1/a))$; since $\beta=1/g^{2}$, division by $\log(a^{-1})$ tends to $2b_{0}=11C_{\g}/(24\pi^{2})$, in agreement with (4.8). The strict \emph{uniform} (in $p$) error bound requires the RG estimates of the formalization and is assumption (RG2) below, proved inside the repository (certificates \texttt{YM.Pert.Asymptotics.b0}, \texttt{YM.Gauge.casimir}, \texttt{YangMills.WilsonRegularity.running}); see Lemma~5.12. \end{proof} \subsection{Smeared lattice correlations} \begin{definition}[Finite lattice $n$-point function] For a label list $v=(v_{1},\dots,v_{n})$ and an admissible test function $f$ (Definition~3.1), let \begin{equation} C_{N,p}(v;f)=a_{N}^{4n}\!\!\sum_{x_{1},\dots,x_{n}\in\Lambda_{N,p}} f(a_{N}x_{1},\dots,a_{N}x_{n})\int_{\Omega_{N,p}}\prod_{i=1}^{n}I_{N,p}(v_{i},x_{i};U) \,\dd\mu_{N,p}(U), \end{equation} the sum over ordered $n$-tuples; the integral is the finite-volume Wilson expectation $\mathbb{E}_{\mu_{N,p}}[\prod_{i}I_{N,p}(v_{i},x_{i};\cdot)]$. \end{definition} \begin{lemma}[Well-definedness, multilinearity, boundedness] $C_{N,p}(v;f)$ is well defined, complex-multilinear in $(v_{1},\dots,v_{n};f)$, and bounded: with $K_{i}=\sup_{U}|I_{N,p}(v_{i},\cdot;U)|$, \begin{equation} |C_{N,p}(v;f)|\le a_{N}^{4n}|\Lambda_{N,p}|^{n}\|f\|_{\infty}\prod_{i=1}^{n}K_{i} =L_{p}^{4n}\|f\|_{\infty}\prod_{i=1}^{n}K_{i}. \end{equation} \end{lemma} \begin{proof} The sum and integral are finite: $\Lambda_{N,p}$ is finite and the integrand is continuous (Lemma~4.9(a)), so the integral is finite by Lemma~4.5. Linearity in each $v_{i}$ follows from the linearity of $I_{N,p}$ in the token coordinate (4.7), and linearity in $f$ is clear. For boundedness, (4.10) follows from $a_{N}^{4n}|\Lambda_{N,p}|^{n}=(a_{N}^{4}(L_{p}/a_{N})^{4})^{n}=L_{p}^{4n}$ (Lemma~4.1). At fixed volume the right side is finite; the possible growth of $K_{i}$ in $N$ is controlled by the RG normalization of (RG3), which is exactly the content of the ultraviolet limit. \end{proof} \begin{lemma}[Permutation invariance] $C_{N,p}$ is invariant under simultaneous permutation of $(v_{i},x_{i})$. \end{lemma} \begin{proof} For $\pi\in S_{n}$, relabel the summation variables $x_{i}\mapsto x_{\pi(i)}$; $\Lambda_{N,p}^{n}$ is invariant under relabeling, and the product $\prod_{i}I_{N,p}(v_{i},x_{i};U)$ takes the same values upon simultaneous permutation of labels and points since finite products commute. \end{proof} \begin{remark}[Role of the quotient map] If $v-v'\in N_{d}$, the finite-$N$ functionals $C_{N,p}(v;\cdot)$ and $C_{N,p}(v';\cdot)$ need not agree; the convergence theorem guarantees that their limits $S_{p}$ agree, and the map \texttt{project} sends a token to the quotient class $(d,[c])$ (Appendix~A). What is constructed is therefore a field of $\mathcal{F}_{d}$, not of raw tokens. \end{remark} \subsection{Common reference scale and normalizations} \begin{lemma}[Five calibrations] For the same Wilson family, before any limit is taken: \begin{enumerate} \item \emph{Reference unit.} The spacing at RG reference level zero is $a(0)=1$ (\texttt{reference\_unit}, \texttt{CommonScale.reference\_spacing}). \item \emph{Physical side length.} $a_{N}\cdot\mathrm{side}(N,p)=L_{p}$ for all $N,p$ (\texttt{side\_length\_in\_physical\_units}, \texttt{CommonScale.physical\_side}). \item \emph{Common reference coupling.} There is $\bar\beta>0$ such that $\beta_{N,p}(0)=\bar\beta$ at reference level zero for every $N,p$; the bare $\beta_{N,p}$ itself may run with the cutoff (\texttt{one\_positive\_reference\_coupling}, \texttt{CommonScale.positive\_common\_reference}). \item \emph{Shared kinetic paths.} For every RG level $m\le N$, the kinetic path at level $m$ is the same for all volumes $p,q$ (\texttt{kinetic\_path\_agrees\_across\_volumes}, \texttt{CommonScale.shared\_kinetic\_paths}). \item \emph{Insertion readout.} For a degree-$d$ label $v$, the normalized coefficient at the fixed reference level $r(d)$ equals $v$ itself, independently of cutoff and volume, whenever $N\ge r(d)$ (\texttt{insertion\_has\_the\_prescribed\_reference\_value}, \texttt{CommonScale.reference\_insertion}). \end{enumerate} \end{lemma} \begin{proof}[Proof of which parts are definitional] Items (1), (2), (3), and (5) are definitional conventions of the RG bookkeeping and the choice of $\beta_{N,p}$: (2) follows from $\Lambda_{N,p}=(\Z/(L_{p}D_{N})\Z)^{4}$ and $a_{N}=D_{N}^{-1}$ (Lemma~4.1); (3) is the definition ``$\beta_{N,p}$ is chosen to return to $\bar\beta$ after $N$ steps''; (5) is the definition of the reference-level readout, its usability depending on invertibility of the RG transport map for $N\ge r(d)$ (assumption (RG3)). Item (4) is a constructive conclusion: it requires the kinetic paths of different volumes to agree, and is supplied by the shared kinetic path data, not by definition alone. \end{proof} \begin{proposition}[Finite-law comparison with the independent lattice formalization; input (C1)] Comparing the finite Wilson law with the independent formalization [LGT] (bridged in Section~13.1): the ordered plaquette products agree lattice point by point before any limit (\texttt{same\_plaquette}); with the conversion $\beta_{\mathrm{submission}}=m\,\beta_{\mathrm{lgt}}$ of (13.1), the normalized finite laws and finite moments agree exactly (\texttt{coupling\_conversion}, \texttt{normalized\_law\_on\_each\_lattice}, \texttt{finite\_moments\_are\_the\_family\_moments}). This is an independent check of the finite lattice theory itself, independent of any limit; the internal proof of the adapter is input (C1). \end{proposition} \begin{remark}[Calibration versus probe] These identities are RG bookkeeping of one Wilson family, not merely the assignment of a common numerical length unit to unrelated theories; complementing them, the separated two-point probe of Section~13.3 below connects the calibration to a nonzero mixed-theory observable. Together they support the normalization of all subsequent short-distance constants. \end{remark} \section{Renormalization group and the ultraviolet limit} \subsection{The RG data and the assumption package} For $\rho>0$ let $C_{\rho}$ be the construction's blocking kernel, coarsening level $m$ (spacing $a^{(m)}$) to level $m-1$ (spacing $a^{(m-1)}=b_{m}a^{(m)}$), with $a^{(0)}=1$ (Lemma~4.15(1)). Write the effective action at level $m$ as \begin{equation} S_{m}(U)=\beta_{m}\sum_{\square}\Big(1-\frac{1}{m}\Reop\Tr U_{\square}\Big)+J_{m}(U), \end{equation} where $\beta_{m}$ is the coefficient of the marginal (plaquette) term and $J_{m}$ a finite sum of irrelevant terms taking values in a Banach space with norm $\|\cdot\|_{m}$; $g^{2}=1/\beta$ is the running coupling squared. \begin{assumption}[RG1: symmetry-preserving gauge fixing] The gauge fixing averages over symmetry-breaking choices and preserves the finite lattice symmetry group $\Gamma_{\mathrm{lat}}$ (rotations, reflections, translations); the blocking kernel $C_{\rho}$ is $\Gamma_{\mathrm{lat}}$-equivariant. Certificates: the source document \S4.3 Remark~4.2 and \S6.4; \texttt{YM.RG.Program}. \end{assumption} \begin{lemma}[Averaging invariance, full proof] Let a finite group $\Gamma$ act on a set $\mathcal{F}$ and let $F_{0}\in\mathcal{F}$ be arbitrary. Then $F=\frac{1}{|\Gamma|}\sum_{\gamma\in\Gamma}\gamma\cdot F_{0}$ is $\Gamma$-invariant. \end{lemma} \begin{proof} For $\eta\in\Gamma$, $\eta\cdot F=\frac{1}{|\Gamma|}\sum_{\gamma}(\eta\gamma)\cdot F_{0} =\frac{1}{|\Gamma|}\sum_{\gamma'\in\Gamma}\gamma'\cdot F_{0}=F$, since $\gamma\mapsto\eta\gamma$ permutes the finite group $\Gamma$. \end{proof} \begin{assumption}[RG2: one-step estimates, uniform in volume] There exist constants $\lambda\in(0,1)$, $K>0$, and $\beta_{\ast}>0$ with \begin{equation} 2b_{0}\log 3\;>\;\frac{K}{\beta_{\ast}},\qquad b_{k}\ge3, \end{equation} such that for all levels $m\le N$ and all volume indices $p$: \[ \|J_{m+1}\|_{m+1}\le\lambda\|J_{m}\|_{m}+K\beta_{m}^{-1}, \qquad \big|\beta_{m+1}-\beta_{m}-2b_{0}\log b_{m}\big|\le K\beta_{m}^{-1}, \qquad \beta_{m}\ge\beta_{\ast}. \] The physical content: each blocking step contracts irrelevant terms geometrically at one fixed rate $\lambda$, while the marginal term follows the one-loop logarithmic recursion; $K\beta_{m}^{-1}$ is the two-loop-and-higher remainder, and (5.2) is exactly the condition that the one-loop recursion dominates its remainder---without it the running condition (4.8) itself would fail. The two extra clauses are not new physics: the lower bound $\beta_{m}\ge\beta_{\ast}>0$ was already used in the propagation bound. Certificates: \texttt{YM.RG.Program}, \texttt{YM.RG.Supply.Program.book}, \texttt{YM.RG.Calendar.Fits}, \texttt{YM.Pert.Asymptotics.b0}, \texttt{YM.Gauge.casimir}. \end{assumption} \begin{remark}[Source-level form of (RG2)] In the submitted sources the remainder is controlled by the schedule rather than by $\beta_{m}^{-1}$: \texttt{YM.\allowbreak Price.\allowbreak Intent.\allowbreak Large.\allowbreak slow} gives $\mathrm{drift}_{j}/\log L\le j^{-2}$ and \texttt{Large.\allowbreak kick} gives $\mathrm{drift}_{j}/\log L\le\mathrm{slopeConstant}/4$; \texttt{YM.\allowbreak RG.\allowbreak History.\allowbreak Fits.\allowbreak driftSmall} instantiates the former at $j=J(m+1)$, $L=b_{m}$. \texttt{YM.\allowbreak Price.\allowbreak Intent.\allowbreak Rolls.\allowbreak exists\_rolls} constructs the sequence via $\mathrm{Tendsto}\ (\lambda n,\ x/\log n)\ \mathrm{atTop}\ (\mathcal{N}\,0)$. Consequently (5.2), the lower bound $\beta_{m}\ge\beta_{\ast}$, and the bootstrap of Lemma~5.11 are \emph{sufficient but not necessary}; the strict sufficient condition is the weighted Ces\`aro Lemma~5.9, and the source-level reading of the drift identity is recorded in Remark~10.42. \end{remark} \begin{assumption}[RG3: stability of the insertion readout] For a degree-$d$ label $v$, the normalized coefficient at the reference level $r(d)$ equals $v$ (Lemma~4.15(5)), and the RG transport linear map from level $N$ back to $r(d)$ is invertible (Appendix~A, \texttt{TransferData.cylinder}); its inverse acts on insertions paired with test functions with bounds governed by the products of (RG2), giving a constant independent of $p$ and growing at most like $a_{N}^{-d}$. Certificates: \texttt{insertion\_has\_the\_prescribed\_reference\_value}, \texttt{CommonScale.reference\_insertion}, \texttt{YM.Family.O}. \end{assumption} \begin{assumption}[RG4: difference estimates, derived] There exists $A(v,f)<\infty$ (depending on finitely many norms of $f$ and on the label weights) such that \[ \big|C_{N+1,p}(v;f)-C_{N,p}(v;f)\big|\le A(v,f)\,\lambda^{N}. \] In addition, for every multi-index $\alpha$, every compact $K\subset\Omega_{n}$, and every finite label list $v$, there is $A_{\alpha,K,v}<\infty$, independent of $N$ and $p$, with \begin{equation} \sup_{x\in K}\big|\partial^{\alpha}_{x}K_{N+1,p}(v;x) -\partial^{\alpha}_{x}K_{N,p}(v;x)\big| \le A_{\alpha,K,v}\,\lambda^{N}, \end{equation} where $K_{N,p}(v;x)$ is the smooth summation kernel of Remark~5.16. Both forms are derived from (RG2)$+$(RG3) (Lemma~5.14 and Remark~5.15); they are not external inputs. \end{assumption} \begin{assumption}[RG5: uniform smoothness estimates] For every multi-index $\alpha$, every compact $K\subset\Omega_{n}$, and every finite label list $v$, \[ \sup_{N,p}\ \sup_{x\in K}\big|\partial^{\alpha}_{x}C_{N,p}(v;x)\big|\le M_{\alpha,K,v}<\infty, \] where $C_{N,p}(v;x)$ is the summation kernel of (4.9) viewed as a function on the separated configurations. \end{assumption} \begin{remark} Within the RG package of Section~5, the external analytic inputs are (RG1), (RG2), (RG3), and (RG5); (RG4) and (5.3) are derived from them (Lemma~5.14 and Remark~5.15). All of these are recorded as proved inside the formalization. The full register of external inputs used in the paper, including (UV1), (PH1)--(PH4), (AN1), (Y1)--(Y4), (B1), and (C1), is Appendix~B. \end{remark} \begin{lemma}[Schedule-controlled remainder; weighted Ces\`aro] Let $b_{m}\ge3$ be the blocking ratios, $D_{N}=\prod_{m0$ with $\beta_{m}\ge\beta_{m}^{(0)}+(m-M)\delta'$ for $m\ge M$, so no bootstrap and no lower bound $\beta_{m}\ge\beta_{\ast}>0$ is needed beyond a positive starting value. \end{enumerate} \end{lemma} \begin{proof} Fix $\eta>0$. Since $J\to\infty$, choose $M$ with $J(m+1)^{-2}\le\eta$ for all $m\ge M$. Split $E_{N}=E_{0$ was arbitrary, $|E_{N}|/\log D_{N}\to0$. With $\beta_{N,p}=\bar\beta+2b_{0}\log D_{N}+E_{N}$ (Lemma~4.15(3)) and $a_{N}^{-1}=D_{N}$, statement (2) follows. For (3), take $M$ with $J(m+1)^{-2}\le b_{0}$; then for $m\ge M$, $\beta_{m+1}\ge\beta_{m}+(2b_{0}-J(m+1)^{-2})\log b_{m}\ge\beta_{m}+b_{0}\log3$, using $\log b_{m}\ge\log3$. A positive starting value $\beta_{M}>0$ is supplied by the band condition (\texttt{Fits.delayed}, \texttt{Bench.recover}). Uniformity in $p$ is inherited from the $p$-independent gates. \end{proof} \begin{remark}[Source-level reading of the remainder] In the submitted sources the remainder is controlled by the schedule rather than by $\beta_{m}^{-1}$: \texttt{YM.Price.Intent.Large.slow} gives $\mathrm{drift}_{j}/\log L\le j^{-2}$ and \texttt{Large.kick} gives $\mathrm{drift}_{j}/\log L\le G.\mathrm{slopeConstant}/4$; \texttt{YM.RG.History.Fits.driftSmall} instantiates the former at $j=J(m+1)$, $L=b_{m}$, and \texttt{Fits.logCost} does the same for the logarithmic cost. \texttt{YM.Price.Intent.Rolls.exists\_rolls} constructs the sequence by the elementary limit $\mathrm{Tendsto}\ (\lambda n,\ x/\log n)\ \mathrm{atTop}\ (\mathcal{N}\,0)$: the drift is a fixed number and choosing the blocking ratio $L$ large enough makes $\mathrm{drift}/\log L$ arbitrarily small; oddification preserves \texttt{Large.odd} and \texttt{Large.size}. \texttt{Fits.deep} gives $J\to\infty$, \texttt{Fits.stepSlow} gives at most linear growth of $J$, and \texttt{Fits.delayed} gives $1<\hat\beta_{m}$. Consequently the dominance condition (5.2), the bound $\beta_{m}\ge\beta_{\ast}$, and the two-stage bootstrap of Lemma~5.11 are \emph{sufficient but not necessary}. The identification of the source's \texttt{drift} with the one-step remainder $\varepsilon_{m}$ is recorded as a source-level reading, not re-proved here. \end{remark} \subsection{Propagation of irrelevant terms and the running coupling} \begin{lemma} Under (RG2), put $\delta':=2b_{0}\log 3-K/\beta_{\ast}>0$ (by (5.2)). Then for every $m\le N$ and every $p$, \begin{equation} \beta_{m}\ge\bar\beta+m\delta', \qquad \|J_{m}\|_{m}\le\lambda^{m}\|J_{0}\|_{0}+\frac{K}{\beta_{\ast}(1-\lambda)}. \end{equation} In particular $\beta_{m}\ge\beta_{\ast}>0$, $\beta_{m}\ge\delta' m$ for $m\ge1$, and $\sup_{m\le N}\|J_{m}\|_{m}$ is bounded independently of $N$ and $p$. \end{lemma} \begin{proof} From $\beta_{m}\ge\beta_{\ast}$, $b_{m}\ge3$, and (RG2), \[ \beta_{m+1}\ge\beta_{m}+2b_{0}\log b_{m}-K\beta_{m}^{-1} \ge\beta_{m}+2b_{0}\log 3-K/\beta_{\ast} =\beta_{m}+\delta'. \] Since $\beta_{N,p}(0)=\bar\beta$ (Lemma~4.15(3)), induction from the reference level gives $\beta_{m}\ge\bar\beta+m\delta'$. For the irrelevant terms, set $u_{m}=\|J_{m}\|_{m}$ and $\kappa=K/\beta_{\ast}$; since $\beta_{m}\ge\beta_{\ast}$, $K\beta_{m}^{-1}\le\kappa$, so $u_{m+1}\le\lambda u_{m}+\kappa$ by (RG2), and inductively $u_{m}\le\lambda^{m}u_{0}+\kappa\sum_{j=0}^{m-1}\lambda^{j}\le u_{0}+\kappa/(1-\lambda)$. All constants are $p$-independent by (RG2). \end{proof} \begin{lemma}[One-loop orbit and the running condition (4.8)] Under (RG2), $b_{k}\ge3$, and $\log b_{N}/\log D_{N}\to0$, \[ \frac{\beta_{N,p}}{\log(a_{N}^{-1})}\longrightarrow 2b_{0}=\frac{11C_{\g}}{24\pi^{2}} \qquad(N\to\infty), \] uniformly in $p$. \end{lemma} \begin{proof} Accumulating the one-step recursion, with $\varepsilon_{m}=\beta_{m+1}-\beta_{m}-2b_{0}\log b_{m}$, $|\varepsilon_{m}|\le K\beta_{m}^{-1}$, and \[ \beta_{N,p}=\beta_{N,p}(0)+\sum_{m0$, hence \[ |E_{N}|\le K\sum_{mN$, (RG4) gives $|C_{M,p}-C_{N,p}|\le\sum_{j=N}^{M-1}A\lambda^{j}\le\frac{A\lambda^{N}}{1-\lambda}\to0$, so $(C_{N,p})_{N}$ is a Cauchy sequence in $\C$ and the limit exists. (ii) \emph{Linearity and continuity.} $C_{N,p}(v;f)$ is multilinear in $(v_{1},\dots,v_{n};f)$ (Lemma~4.12), and pointwise limits preserve linearity; by (4.10) and the uniformity in (RG3), $|S_{p}(v;f)|\le C_{p,v}\|f\|_{\infty}$, so $S_{p}$ is a continuous linear functional on $\mathcal{T}_{n}$. (iii) \emph{Smooth kernel.} Fix a compact $K\subset\Omega_{n}$ and a multi-index $\alpha$. By (RG4*) (5.3), for $M>N$, \[ \sup_{x\in K}\big|\partial^{\alpha}K_{M,p}(v;x)-\partial^{\alpha}K_{N,p}(v;x)\big| \le\sum_{j=N}^{M-1}A_{\alpha,K,v}\lambda^{j} \le\frac{A_{\alpha,K,v}\lambda^{N}}{1-\lambda}\xrightarrow[N\to\infty]{}0. \] Hence $(\partial^{\alpha}K_{N,p})_{N}$ is a Cauchy sequence in the Banach space $C^{0}(K)$, and by (RG5) it is uniformly bounded; completeness of $C^{0}(K)$ yields $S^{(\alpha)}_{p}\in C^{0}(K)$ with $\partial^{\alpha}K_{N,p}\to S^{(\alpha)}_{p}$ uniformly on $K$. By the Weierstrass theorem on the interchange of uniform convergence and differentiation, $S^{(0)}_{p}\in C^{\infty}(K)$ and $\partial^{\alpha}S^{(0)}_{p}=S^{(\alpha)}_{p}$ for every $\alpha$. Choosing a countable compact exhaustion of $\Omega_{n}$, the limits agree on overlaps (a subsequence of a convergent sequence converges to the same limit), so they patch to a single kernel $S_{p}(v;\cdot)\in C^{\infty}(\Omega_{n})$. (iv) \emph{Integral representation.} Let $\supp f\subset K\subset\Omega_{n}$ with $f$ smooth and compactly supported. By (5.5) and the triangle inequality, \[ \Big|C_{N,p}(v;f)-\int f(x)\,S_{p}(v;x)\,\dd x\Big| \le\big|\mathcal{E}_{N,p}(f)\big| +\|f\|_{L^{1}}\sup_{x\in K}\big|K_{N,p}(v;x)-S_{p}(v;x)\big|. \] The first term tends to $0$ by (5.6)--(5.7) and $a_{N}\to0$; the second tends to $0$ by the case $\alpha=0$ of (iii). Hence $S_{p}(v;f)=\lim_{N\to\infty}C_{N,p}(v;f)=\int f(x)S_{p}(v;x)\,\dd x$, which is (5.8) together with the kernel representation. (v) \emph{Running coupling.} This is Lemma~5.12. \end{proof} \begin{corollary} (i) If $v-v'\in N_{d}$, then $S_{p}(v;f)=S_{p}(v';f)$; (ii) $S_{p}$ is invariant under simultaneous permutation of $(v_{i},x_{i})$; (iii) gauge invariance passes to the limit. \end{corollary} \begin{proof} (i) By definition of $N_{d}$ (Definition~2.7), all separated correlations containing $v-v'$ vanish in the limit; \texttt{project} sends a token to $(d,[c])$, and the \texttt{wilson} property requires equivalent tokens to have equal limiting correlations. (ii) Lemma~4.13 gives finite-$N$ permutation invariance, and pointwise limits preserve it. (iii) Each $I_{N,p}$ is gauge invariant (Lemma~4.9(c)) and $\mu_{N,p}$ is gauge invariant (Lemma~4.4), so each $C_{N,p}$ is gauge invariant, and so is the limit. \end{proof} \begin{remark}[Scope of convergence] Convergence holds along the entire sequence $(a_{N})$ used by the construction; there is no universality statement for arbitrary sequences $a_{N}\to0$. Convergence is proved only for test functions in $C^{\infty}_{c}$ away from the coincidence set, and step (iii) gives smoothness only on $\Omega_{n}$, excluding contact terms on the diagonals. \end{remark} \subsection{The insertion formula and nondegeneracy at finite cutoff} The first two consequences of the construction-chain audit (Section~11.4) give the readable form \begin{equation} I_{N,p}(v,x;U)=a_{N}^{-d}\,(\texttt{places}\;\texttt{ev})(x;U), \qquad \texttt{ev}=\big(\texttt{terminal}^{-1}\circ\operatorname{diag}(t^{-1})\big)(v), \end{equation} where $t\neq0$ is the RG threshold scale and \texttt{terminal} undoes the terminal renormalization; the composite is a linear equivalence of the raw label space at each cutoff (\texttt{actual\_insertion\_formula}, proved by \texttt{rfl}). Consequently \texttt{every\_primary\_is\_represented}: every placed primary section is an actual renormalized insertion times the common factor $a_{N}^{-d}$. \begin{lemma}[Nondegeneracy at finite cutoff, full proof] Let $L:V\to V$ be an invertible linear map and $c\neq0$. If $p\in V$ is nonconstant, then $c\,L(p)$ is nonconstant. \end{lemma} \begin{proof} If $cL(p)$ equals a constant $k$, then $L(p)=k/c$ is constant; invertible linear maps carry the one-dimensional subspace of constant labels to itself, so $p=L^{-1}(k/c)$ is constant, a contradiction. \end{proof} \begin{remark} This is the third consequence \texttt{nonconstant\_primary\_survives}, with its recorded boundary: it is a finite-cutoff statement, and no uniform bound on the inverse coefficient map as the cutoff is removed is provided, so it does not by itself exclude a degenerate continuum limit; exclusion of degeneracy rests on the field-content comparison (Proposition~2.12) and the short-distance results (Section~9). The fourth consequence \texttt{every\_field\_word\_has\_lattice\_origin} gives only an iterated limit $\operatorname{smear}\to\mathrm{row}\to\int t\,S.\texttt{useAt}$: no interchange of limits, no rate, no uniqueness across volume subsequences, and the right side uses the \emph{mixed} table, not a selected phase. \end{remark} \section{The infinite-volume limit} \subsection{Uniform bounds from the ultraviolet input} \begin{assumption}[(UV1) local boundedness and equicontinuity] For every finite label list $v$, every compact $K\subset\Omega_{n}$, and every multi-index $\alpha$, there are $p_{0}(K)$, $M_{\alpha,K}$, and a modulus $\omega_{\alpha,K}(r)\downarrow0$ such that \[ \sup_{p\ge p_{0}}\sup_{x\in K}\big|\partial^{\alpha}_{x}S_{p}(v;x)\big|\le M_{\alpha,K}, \qquad \big|\partial^{\alpha}S_{p}(v;x)-\partial^{\alpha}S_{p}(v;y)\big| \le\omega_{\alpha,K}(|x-y|) \] for $x,y\in K$, $p\ge p_{0}$. The case $\alpha=0$ is exactly the source document's bounds; $\alpha>0$ follows by the same mechanism applied to the differentiated kernels. Certificates: the source document Remark~5.2; the mechanism is reflection positivity, chessboard estimates, the transfer-operator representation, and Cauchy estimates, with the construction-specific input being volume-uniform control of the renormalized lattice observables. \end{assumption} \begin{lemma}[Readable derivation of the mechanism] Reflection positivity gives the OS positive semidefinite form $\langle A,B\rangle_{\mathrm{OS}}$; by Cauchy--Schwarz and permutation symmetry, $|S_{p}(v;x)|\le\prod_{i}\langle A_{i},A_{i}\rangle^{1/2}$; chessboard estimates bound each factor by finite-volume two-point functions, which are uniformly bounded at fixed volume by (RG5), giving local boundedness. Equicontinuity comes from analyticity of the kernels at separated points: around each point of $K$ take a polydisc of radius $\delta$; the Cauchy integral formula gives $|\nabla S_{p}|\le CM_{K}/\delta$, hence a Lipschitz modulus $\omega_{K}(r)=CM_{K}r/\delta$. \end{lemma} \subsection{Countability and the diagonal extraction} \begin{lemma}[Countability of labels and tests] There is a countable collection $\mathcal{L}$ of finite label lists and a countable dense set $\mathcal{T}_{n}\subset\mathcal{T}_{n}$ such that all $S_{p}(v;f)$ are determined by their values on $\mathcal{L}\times\mathcal{T}_{n}$. \end{lemma} \begin{proof} By Lemma~2.6 and Definition~2.7, each $\mathcal{F}_{d}$ is finite-dimensional; choose a basis $e_{d,1},\dots,e_{d,n_{d}}$. Finite words in these basis elements are countable, and by multilinearity the value at arbitrary labels is a finite linear combination of basis-label values. For test functions, $C^{\infty}_{c}(\Omega_{n})$ is separable in its usual topology: take polynomials with rational coefficients times smooth cutoffs with rational centers and radii, further cut off away from the diagonals, giving a countable dense set (Lemma~3.2). Continuity and linearity of $S_{p}(v;\cdot)$ make the values on the dense set determining. \end{proof} \begin{theorem}[Subsequential infinite-volume limit] Under (UV1), for every strictly increasing sequence $p:\N\to\N$ of volume indices there exists a strictly increasing $q:\N\to\N$ and kernels $S_{\mathrm{mix}}(v;x)$, $C^{\infty}$ on $\Omega_{n}$, such that for all finite label lists $v$ and all $f\in\mathcal{T}_{n}$, \begin{equation} \lim_{k\to\infty}S_{p(q(k))}(v;f)=\int_{(\R^{4})^{n}}f(x)\,S_{\mathrm{mix}}(v;x)\,\dd x; \end{equation} one and the same $q$ works for all correlation functions simultaneously. \end{theorem} \begin{proof} (i) \emph{Compact exhaustion.} Take a countable compact exhaustion $K_{m}=\{x\in\Omega_{n}:|x_{i}|\le m,\ |x_{i}-x_{j}|\ge1/m\}$ of $\Omega_{n}$; each $K_{m}$ is compact and $\bigcup_{m}K_{m}=\Omega_{n}$. (ii) \emph{Termwise compactness --- the objects are kernels, not functionals.} Fix a finite label list $v$, an index $m$, and a multi-index $\alpha$. By (UV1) (which already carries the bounds for every $\alpha$), the family $\big\{\partial^{\alpha}S_{p}(v;\cdot)\big|_{K_{m}}:p\ge p_{0}(K_{m})\big\}$ is uniformly bounded by $M_{\alpha,K_{m}}$ and equicontinuous with modulus $\omega_{\alpha,K_{m}}$; by Arzel\`a--Ascoli it is relatively compact in $C^{0}(K_{m})$, so every sequence of volume indices admits a subsequence along which $\partial^{\alpha}S_{p}(v;\cdot)|_{K_{m}}$ converges uniformly. (iii) \emph{Diagonal extraction over a countable index set.} Let $\mathcal{L}$ be the countable collection of basis label lists of Lemma~6.3 and let \[ \mathcal{I}:=\mathcal{L}\times\{K_{m}:m\in\N\}\times\{\alpha\in\N_{0}^{4n}\}, \] which is countable; enumerate it as $(t_{1},t_{2},\dots)$ with $t_{j}=(v_{j},K_{m_{j}},\alpha_{j})$. Start with $\mathcal{S}_{0}=(p(1),p(2),\dots)$. Recursively, given $\mathcal{S}_{j-1}$, use (ii) to choose a strictly increasing subsequence $\mathcal{S}_{j}\subset\mathcal{S}_{j-1}$ along which $\partial^{\alpha_{j}}S_{p}(v_{j};\cdot)|_{K_{m_{j}}}$ converges uniformly. Define $q(k)$ to be the $k$-th element of $\mathcal{S}_{k}$. Then $q$ is strictly increasing, and for each fixed $j$ we have $q(k)\in\mathcal{S}_{j}$ for all $k\ge j$; hence $\partial^{\alpha_{j}}S_{p(q(k))}(v_{j};\cdot)|_{K_{m_{j}}}$ converges uniformly. Thus one sequence $q$ makes every countable item converge. Denote the limits by $S^{(\alpha)}_{\mathrm{mix}}(v;\cdot)$ on each $K_{m}$. (iv) \emph{Consistency and smoothness.} If $K_{m}\subset K_{m'}$, the restriction of the $K_{m'}$-limit to $K_{m}$ coincides with the $K_{m}$-limit, because both are limits along the same subsequence. For fixed $v$ and $m$, all derivatives converge uniformly on $K_{m}$; by the Weierstrass theorem, $S^{(0)}_{\mathrm{mix}}(v;\cdot)\in C^{\infty}(K_{m})$ and $\partial^{\alpha}S^{(0)}_{\mathrm{mix}}=S^{(\alpha)}_{\mathrm{mix}}$. Patching over $m$ gives $S_{\mathrm{mix}}(v;\cdot)\in C^{\infty}(\Omega_{n})$. (v) \emph{From kernels to functionals.} Let $f\in\mathcal{T}_{n}$ with $\supp f\subset K_{m}$. By Lemma~3.4 and Theorem~5.17, $S_{p}(v;f)=\int f(x)\,S_{p}(v;x)\,\dd x$; the case $\alpha=0$ of (iii) gives uniform convergence $S_{p(q(k))}(v;\cdot)\to S_{\mathrm{mix}}(v;\cdot)$ on $K_{m}$, hence \[ \lim_{k\to\infty}S_{p(q(k))}(v;f) =\int f(x)\,S_{\mathrm{mix}}(v;x)\,\dd x. \] For a general label list $v$, multilinearity and the basis $\mathcal{L}$ of Lemma~6.3 express $S_{p}(v;f)$ as a finite linear combination of the $S_{p}(v';f)$ with $v'\in\mathcal{L}$; since $q$ works for all of them, it works for $v$. No separate dense-set approximation of test functions is needed: the countable set $\mathcal{I}$ already ranges over all labels, all compacts, and all derivative orders, and the convergence of kernels is uniform on the support of $f$. \end{proof} \begin{remark} Jaffe--Witten regard a subsequential limit obtained by compactness as insufficient unless further properties of the limit are established [JW]; Sections~7 and 9 supply exactly that further information. Whether the limit depends on the chosen subsequence remains open. \end{remark} \subsection{Clustering would force all phases to agree} \begin{lemma} Suppose the phase decomposition (Theorem~7.13) holds, every component clusters, the measure $\nu$ has full support (Proposition~7.18), and $\xi\mapsto S_{\xi}(A)$ is continuous. If $S_{\mathrm{mix}}$ satisfies clustering, then for every word $A$ and every $\xi\in X$, $S_{\xi}(A)=S_{\mathrm{mix}}(A)$. \end{lemma} \begin{proof} Let $A_{s}$ be $A$ translated by $s$ in the time direction, and let $A^{\ast}$ be the adjoint word. Reflection positivity (E2) gives, for each phase, a positive semidefinite OS form; by Cauchy--Schwarz and translation invariance of each phase, \begin{equation} \big|S_{\xi}(A^{\ast}A_{s})\big| \le S_{\xi}(A^{\ast}A)^{1/2}\,S_{\xi}(A_{s}^{\ast}A_{s})^{1/2} =S_{\xi}(A^{\ast}A) \qquad\text{for all }s\ge0. \end{equation} The function $g(\xi):=S_{\xi}(A^{\ast}A)$ is continuous on $X$ (Gelfand transform of $M_{A^{\ast}A}\in\mathcal{A}$) and non-negative, and by Theorem~7.13, \begin{equation} \int_{X}g\,\dd\nu=S_{\mathrm{mix}}(A^{\ast}A)<\infty . \end{equation} Thus (6.2)--(6.3) supply a $\nu$-integrable dominating function, uniform in $s$ (note that (UV1) is not used here: its bounds are local on compact sets, whereas $A_{s}$ leaves every fixed compact as $s\to\infty$). Dominated convergence and clustering of every component (Proposition~7.15) give \[ \lim_{s\to\infty}S_{\mathrm{mix}}(A^{\ast}A_{s}) =\int_{X}\lim_{s\to\infty}S_{\xi}(A^{\ast}A_{s})\,\dd\nu =\int_{X}\overline{S_{\xi}(A)}\,S_{\xi}(A)\,\dd\nu =\int_{X}\big|S_{\xi}(A)\big|^{2}\,\dd\nu . \] On the other hand, clustering of $S_{\mathrm{mix}}$ applied to the same pairing gives $\lim_{s\to\infty}S_{\mathrm{mix}}(A^{\ast}A_{s})=S_{\mathrm{mix}}(A^{\ast})\,S_{\mathrm{mix}}(A) =\big|S_{\mathrm{mix}}(A)\big|^{2}$, since $S_{\mathrm{mix}}(A^{\ast})=\overline{S_{\mathrm{mix}}(A)}$ and $S_{\mathrm{mix}}(A)=\int_{X}S_{\xi}(A)\,\dd\nu$. Therefore \[ \int_{X}\big|S_{\xi}(A)\big|^{2}\,\dd\nu -\Big|\int_{X}S_{\xi}(A)\,\dd\nu\Big|^{2}=0, \qquad\text{i.e.}\qquad \int_{X}\big|S_{\xi}(A)-S_{\mathrm{mix}}(A)\big|^{2}\,\dd\nu(\xi)=0, \] so $S_{\xi}(A)=S_{\mathrm{mix}}(A)$ for $\nu$-almost every $\xi$. The function $\xi\mapsto S_{\xi}(A)-S_{\mathrm{mix}}(A)$ is continuous and $\nu$ has full support, so its zero set, containing a set of full measure, cannot be a proper closed set; it is all of $X$. \end{proof} \begin{remark} The source document (\S6.3) uses this computation to show that component clustering does \emph{not} imply clustering of the mixed theory, and stresses that even with clustering, one does not obtain sufficient infrared control to prove uniqueness of the infinite-volume limit. Lemma~6.6 is the precise ``clustering $\Rightarrow$ all phases agree'' statement and nothing more. \end{remark} \section{Phase decomposition, clustering, and full support} Throughout this section, fix one infinite-volume theory extracted in Theorem~6.4. \subsection{The OS Hilbert space and the ground-state subspace} Reflection positivity defines a positive semidefinite form on products of fields at positive Euclidean times; quotienting the null space and completing gives the OS Hilbert space $\mathcal{H}_{\mathrm{OS}}$, with time semigroup $E(t)=e^{-tH}$ ($H\ge0$), vacuum $\Omega_{0}$, ground-state subspace $\mathcal{H}_{0}=\ker H$, and orthogonal projection $P_{0}$ onto $\mathcal{H}_{0}$. \begin{lemma} $E(t)|_{\mathcal{H}_{0}}=\mathrm{id}$, and $E(t)\to P_{0}$ strongly as $t\to\infty$. \end{lemma} \begin{proof} By the spectral theorem $E(t)=\int_{[0,\infty)}e^{-t\lambda}\,\dd E_{H}(\lambda)$; on $\mathcal{H}_{0}=E_{H}(\{0\})\mathcal{H}$ one has $\lambda=0$, so $e^{-t\lambda}=1$. For any $\psi$, \[ \|(E(t)-P_{0})\psi\|^{2}=\int_{\lambda>0}e^{-2t\lambda}\,\dd\mu_{\psi}(\lambda), \qquad \mu_{\psi}(B)=\langle\psi,E_{H}(B)\psi\rangle; \] the integrand tends to $0$ pointwise and is dominated by $1$, with $\mu_{\psi}$ finite, so dominated convergence gives convergence to $0$. \end{proof} \begin{proposition}[Clustering, vacuum uniqueness, and (PH4) are the same statement] Let $(\mathcal{H},H,\Omega)$ be the OS/Wightman triple of a phase, $E(t)=e^{-tH}$, $P_{0}=|\Omega\rangle\langle\Omega|$, and let $\mathcal{D}\subset\mathcal{H}$ be the dense span of finite positive-time field words. Under (AN1) E2, the following are equivalent: \begin{enumerate} \item[\upshape(1)] \emph{(PH4)} $\ker H=\C\,\Omega$, equivalently $P_{\ker H}=P_{0}$; \item[\upshape(2)] \emph{E4 clustering} $\langle\psi,E(t)\chi\rangle\to \langle\psi,\Omega\rangle\langle\Omega,\chi\rangle$ for all $\psi,\chi\in\mathcal{D}$; \item[\upshape(3)] $\dim\ker H=1$. \end{enumerate} \end{proposition} \begin{proof} Decompose $\mathcal{H}=\ker H\oplus(\ker H)^{\perp}$ and write $\psi=\psi_{0}+\psi_{\perp}$, $\psi_{0}=P_{\ker H}\psi$. For $\psi_{\perp}\perp\ker H$, the spectral theorem gives \[ \|(E(t)-P_{\ker H})\psi_{\perp}\|^{2} =\int_{(0,\infty)}e^{-2t\lambda}\,\dd\mu_{\psi_{\perp}}(\lambda) \xrightarrow[t\to\infty]{}0 \] by dominated convergence: $e^{-2t\lambda}\to0$ for every $\lambda>0$, the integrand is bounded by $1$, and $\mu_{\psi_{\perp}}$ is a finite measure. Hence $E(t)\to P_{\ker H}$ strongly. (1)$\Rightarrow$(2): if $P_{\ker H}=P_{0}$, then $\langle\psi,E(t)\chi\rangle\to\langle\psi,P_{0}\chi\rangle =\langle\psi,\Omega\rangle\langle\Omega,\chi\rangle$. (2)$\Rightarrow$(1): for $\psi,\chi\in\mathcal{D}$, strong convergence gives $\langle\psi,P_{\ker H}\chi\rangle=\lim_{t}\langle\psi,E(t)\chi\rangle =\langle\psi,P_{0}\chi\rangle$; two bounded operators agreeing on a dense domain coincide, so $P_{\ker H}=P_{0}$ and $\ker H=\operatorname{ran}P_{0}=\C\,\Omega$. (1)$\Leftrightarrow$(3): $P_{0}$ is a rank-one projection. \end{proof} \begin{remark} (PH4) is therefore not an external input: it is a spectral conclusion from (AN1) E2 + E4. The repository README confirms this status (``the Wightman spectrum condition and uniqueness of the vacuum are the spectral conclusions in the public challenge''). Appendix~B records it as a proved conclusion, reducing the external-input count from seventeen to sixteen. In Section~7.4 and Proposition~8.5 the phrase ``by (PH4)'' henceforth refers to Proposition~7.2; no circularity arises, since Proposition~7.2 uses only (AN1) and the spectral theorem. This equivalence is also the consistency input for the autocorrelation criterion of Theorem~E.5 in Appendix~E. \end{remark} \subsection{Word operators and the commutative algebra} For a word $A$ with distinct points in a time slab $00$. \end{proposition} \begin{proof} Suppose $\nu(U)=0$. Choose a nonzero continuous $f:X\to\C$ supported in $U$ (Urysohn's lemma; $X$ is compact metrizable) and let $a\in\mathcal{A}$ be its Gelfand inverse image, $\hat a=f$. Then \[ \|a\Omega_{0}\|^{2}=\omega(a^{\ast}a)=\int_{X}|f|^{2}\,\dd\nu=0, \] so $a\Omega_{0}=0$; by (PH3) faithfulness, $a=0$, hence $f=\hat a=0$, contradicting the choice of $f$. \end{proof} \begin{remark} Recorded in Appendix~A as \texttt{fullSupport} (certificates \texttt{YM.OS.Ground.attains}, \texttt{YM.OS.Ground.supports}); it is the key input for the ``almost everywhere $\Rightarrow$ everywhere'' step in Lemma~6.6. \end{remark} \section{Osterwalder--Schrader reconstruction and the Wightman axioms} \subsection{The analytic input and the two test spaces} \begin{assumption}[(AN1)] The correlations of the selected phase $\xi^{\ast}$ satisfy the OS axioms E0 (temperedness), E0$'$ (arity growth), E1 (Euclidean invariance), E2 (reflection positivity), E3 (permutation), E4 (clustering), normalization, and analyticity of Appendix~C; the analytic continuation $F^{\xi^{\ast}}_{n}(v;z,x)$, $\operatorname{Im}z_{1}<\dots<\operatorname{Im}z_{n}$, satisfies $S^{\xi^{\ast}}(v;\tau,x)=F^{\xi^{\ast}}_{n}(v;i\tau,x)$ for $\tau_{1}<\dots<\tau_{n}$. Certificates: \texttt{Properties.continuation}, \texttt{.analytic}, \texttt{YM.Boundary.Waves.Region}, \texttt{YMOSBridge.verifiedOS}, \texttt{complete\_OS\_on\_this\_hierarchy}. \end{assumption} \begin{remark}[Two levels of test space] Three distinct distributional statements must be kept apart. (i) $\mathcal{T}_{n}=C^{\infty}_{c}(\Omega_{n})$: tests avoiding the coincidence diagonals, used for lattice convergence. (ii) $\mathcal{S}_{0}((\R^{4})^{n})$: joint Schwartz tests flat to every order on each diagonal $\Delta_{ij}$, used for the independent OS verification of Appendix~C; it contains (i) but cannot detect distributions supported on the diagonals. (iii) The reconstruction separately supplies tempered Wightman distributions on unrestricted joint Schwartz tests. The Schwinger functional of the independent check is \begin{equation} \mathcal{S}_{n}(f)=\int_{(\R^{4})^{n}}K^{\xi^{\ast}}_{\Theta,n}(x)\,f(x)\,\dd x, \qquad K^{\xi^{\ast}}_{\Theta,n}(x):=S^{\xi^{\ast}}_{\Theta,\dots,\Theta}(x), \end{equation} with absolute convergence, continuity on $\mathcal{S}_{0}$, and the E0$'$ bound $\|\mathcal{S}_{n}(f)\|\le\beta^{n}(n!)^{A}\|f\|_{(ns,ns)}$ (certificates \texttt{YMOSBridge.targetKernel\_bound}, \texttt{schwinger\_integrable}, \texttt{E0\_tempered}, \texttt{E0\_prime\_growth}). \end{remark} \subsection{Wightman boundary values and the Hilbert space} Let $\mathcal{P}$ be the complex vector space of finite formal words $(v_{1},f_{1})\cdots(v_{n},f_{n})$ with $f_{i}\in\mathcal{S}(\R^{4};\C)$, including the empty word; let $\ast$ reverse words, adjoin labels, and conjugate tests and coefficients. Take the Wightman boundary values $W^{\xi}(u)=\lim_{\operatorname{Im}z\to0^{+}}F(z)$ in the sense of Schwartz tests, and define \begin{equation} \langle u,w\rangle_{\xi}=W^{\xi}(u^{\ast}w), \qquad N_{\xi}=\{u:\langle u,u\rangle_{\xi}=0\}, \qquad \mathcal{H}_{\xi}=\overline{\mathcal{P}/N_{\xi}}, \end{equation} with $\Omega_{\xi}$ the empty word, $\mathcal{D}_{\xi}=\mathcal{P}/N_{\xi}$ dense in $\mathcal{H}_{\xi}$, and $\Phi^{\xi}_{v}(f)[w]=[(v,f)w]$. \begin{lemma} $\langle\cdot,\cdot\rangle_{\xi}$ is positive semidefinite; $N_{\xi}$ is a left ideal of $\mathcal{P}$, and $\langle u,w\rangle=0$ for all $w$ if and only if $u\in N_{\xi}$. \end{lemma} \begin{proof} Positive semidefiniteness comes from E2 reflection positivity through analytic continuation (the OS theorem). By Cauchy--Schwarz, $|\langle u,w\rangle|^{2}\le\langle u,u\rangle\langle w,w\rangle=0$ when $u\in N_{\xi}$, so $\langle u,w\rangle=0$ for all $w$; conversely take $w=u$. If $u\in N_{\xi}$ and $b=(v,f)$ is arbitrary, then for every $w$, $\langle bu,w\rangle=W(u^{\ast}b^{\ast}w)=\langle u,b^{\ast}w\rangle=0$ since $b^{\ast}w\in\mathcal{P}$; hence $bu\in N_{\xi}$. \end{proof} \begin{lemma} $\Phi^{\xi}_{v}(f):\mathcal{D}_{\xi}\to\mathcal{D}_{\xi}$ is well defined and complex-linear; $\langle\Phi_{v}(f)\psi,\chi\rangle =\langle\psi,\Phi_{v^{\ast}}(\bar f)\chi\rangle$; and $\mathcal{D}_{\xi}$ is exactly the span of field words applied to $\Omega_{\xi}$ (cyclicity). \end{lemma} \begin{proof} Well-definedness: if $[w]=[w']$, then $w-w'\in N_{\xi}$, and by Lemma~8.3 (left ideal) $(v,f)(w-w')\in N_{\xi}$. Linearity is clear. The adjoint relation: $\langle\Phi_{v}(f)[u],[w]\rangle=W(u^{\ast}(v,f)^{\ast}w) =\langle[u],\Phi_{v^{\ast}}(\bar f)[w]\rangle$. Cyclicity: $\mathcal{P}$ is generated by words, and $[w]=\Phi_{v_{1}}(f_{1})\cdots\Phi_{v_{n}}(f_{n})\Omega_{\xi}$. \end{proof} \subsection{The Wightman axioms, proved item by item} \begin{proposition}[W0: spectrum and uniqueness of the vacuum] Translations are implemented by a strongly continuous unitary representation $U(a)=\int e^{ip\cdot a}\,\dd E(p)$ with $\supp E\subset\overline{V^{+}}$; every translation-invariant vector is a complex multiple of $\Omega_{\xi}$. \end{proposition} \begin{proof} Translation invariance of $W^{\xi}$ gives isometries $T(a)[w]=[(v_{1},f_{1}(\cdot-a))\cdots]$ on $\mathcal{P}/N_{\xi}$; being isometric on a dense domain, $T(a)$ extends uniquely to a unitary on $\mathcal{H}_{\xi}$, and strong continuity follows from continuity of $f\mapsto f(\cdot-a)$ in the Schwartz topology. For the spectral support, by (AN1) the Euclidean correlations of the selected phase are boundary values of functions $F_{n}$ holomorphic in the tube \[ \mathcal{T}_{n-1} =\big\{(\zeta_{1},\dots,\zeta_{n-1})\in(\C^{4})^{n-1}: \operatorname{Im}\zeta_{j}\in V^{+}\big\}, \qquad \zeta_{j}=z_{j+1}-z_{j}, \] and obey the tempered growth bound of E0$'$. For $n=2$ write $W_{2}(\xi)=\lim_{\eta\to0,\,\eta\in V^{+}}F_{1}(\xi+i\eta)$ in $\mathcal{S}'(\R^{4})$, where $\xi=x_{2}-x_{1}$. Since $F_{1}$ is holomorphic on the tube $\R^{4}+iV^{+}$ with polynomial growth in $1/\operatorname{Im}\zeta$ and in $|\zeta|$, the Paley--Wiener--Schwartz theorem for tube domains implies that the Fourier transform $\mu=\widehat{W_{2}}$ is a tempered measure with $\supp\mu\subset\overline{V^{+}}=\{p:p_{0}\ge|\vec p|\}$. For general $n$, the same argument applied to each difference variable $\xi_{j}=x_{j}-x_{j+1}$ places the joint energy--momentum measure in $(\overline{V^{+}})^{n-1}$, and translation covariance (E1) then forces the spectral measure of $U(a)=\int e^{ip\cdot a}\,\dd E(p)$ to satisfy $\supp E\subset\overline{V^{+}}$. \emph{External input used}: the Paley--Wiener--Schwartz theorem for tube domains; \emph{expanded here}: the reduction to difference variables, the $n=2$ Laplace/Fourier step, and the passage from joint to single-particle support. Vacuum uniqueness: the translation-invariant vectors of $\mathcal{H}_{\xi}$ are exactly $E_{H}(\{0\})\mathcal{H}_{\xi}=\mathcal{H}_{0}^{\xi}$; by Proposition~7.2 (derived from (AN1) E2 + E4) this space is $\C\,\Omega_{\xi}$. Thus W0's vacuum uniqueness follows from (AN1), with no circularity; Proposition~7.15 supplies the equivalent clustering statement. The same equivalence is the AUTO input of Appendix~E (Theorem~E.5). \end{proof} \begin{proposition}[W1: regularity, adjoints, cyclicity] $f\mapsto\Phi_{v}(f)\psi$ is continuous from $\mathcal{S}$ to $\mathcal{H}_{\xi}$; the adjoint relation holds; $\mathcal{D}_{\xi}$ is dense, contains $\Omega_{\xi}$, is preserved by all fields, and equals the span of field words. \end{proposition} \begin{proof} Continuity follows from the tempered growth bounds of E0$'$ and control by Schwartz seminorms; adjoints and cyclicity are Lemma~8.4; density is built into the definition of $\mathcal{H}_{\xi}$ as a completion. \end{proof} \begin{proposition}[W2: Poincar\'e covariance] The proper orthochronous Poincar\'e group has a strongly continuous unitary representation $U(g)$, and there is a complex-linear representation $\rho$ on labels with $U(g)\Phi_{v}(f)U(g)^{-1}=\Phi_{\rho(g)v}(f_{g})$, $f_{g}(x)=f(\Lambda^{-1}(x-a))$, and $U(g)\Omega_{\xi}=\Omega_{\xi}$. \end{proposition} \begin{proof} E1 gives covariance of the Euclidean correlations under $O(4)$ (and its complexification); by (AN1) the analytic continuation and the Bargmann--Hall--Wightman theorem extend the domain to the tube of the complex Lorentz group, on which real proper orthochronous Lorentz transformations act on the boundary values, giving $U(\Lambda)$; combining with translations gives the Poincar\'e representation. The label representation $\rho$ comes from the tensor transformation law of curvature polynomials (the $O(4)$ action of Definition~2.5), is complex-linear, and satisfies $\rho(gh)=\rho(g)\rho(h)$. Vacuum invariance follows from translation invariance and the definition of $\rho$. \emph{External input used}: the Bargmann--Hall--Wightman theorem (holomorphy on the extended tube); the $O(4)$-covariance of the Euclidean correlations (E1), the invariance of the inner product under the induced label action (reflection positivity), and the unitary extension are expanded here. \end{proof} \begin{proposition}[W3: local commutativity] If $\supp f$ and $\supp h$ are pointwise spacelike separated, then $[\Phi_{v}(f),\Phi_{w}(h)]=0$ on $\mathcal{D}_{\xi}$. \end{proposition} \begin{proof} For two spacelike separated points $x,y$ there is a complex Lorentz transformation reversing their time order continuously within the tube without leaving the domain of holomorphy (the Jost point argument); by E3 permutation symmetry and uniqueness of analytic continuation, the two boundary values agree. At the level of words this is exactly $[(v,f)(w,h)]=[(w,h)(v,f)]$ in $\mathcal{P}/N_{\xi}$; by Lemma~8.4, $\Phi_{v}(f)\Phi_{w}(h)\psi=\Phi_{w}(h)\Phi_{v}(f)\psi$. \emph{External input used}: existence of the Jost-point path in the extended tube (a standard corollary of the Bargmann--Hall--Wightman theorem). \end{proof} \begin{remark}[Audit status of W0--W3] Propositions~8.5--8.8 are complete relative to (AN1) plus three classical OS--Wightman theorems: the Paley--Wiener--Schwartz theorem for tube domains, the Bargmann--Hall--Wightman theorem, and the Jost-point theorem. These are marked as external inputs where used and are not re-proved; every other proof in Sections~2--9.5 is closed. This aligns the phrase ``complete proofs'' in the introduction with the sixteen external inputs registered in Appendix~B. \end{remark} \begin{theorem}[Reconstruction of the component theories] Under (AN1), for every character $\xi\in X$ the component correlations $S_{\xi}$ reconstruct to a Wightman theory on $\mathcal{H}_{\xi}$ with vacuum $\Omega_{\xi}$ and fields $\Phi^{\xi}$ on $\mathcal{D}_{\xi}$, satisfying W0--W3 (Propositions~8.5--8.8), analytic continuation and recovery of correlations, and faithfulness: every nonzero label $v$ has a test $f$ with $\Phi^{\xi}_{v}(f)\neq0$. The formal certificates are \texttt{YangMills.Hierarchy.Reconstructed.axioms}, \texttt{.analytic}, \texttt{.root}, \texttt{.boundary}, and \texttt{.faithful}. \end{theorem} \begin{proof}[Proof of faithfulness] Suppose $\langle\Phi_{v}(f)[u],[w]\rangle=0$ for all $f$ and all words $u,w$; then all Wightman correlations containing $v$ vanish. By uniqueness of the boundary values of one fixed analytic continuation (the seventh chain consequence, Section~11.4), all separated-point Euclidean correlations containing $v$ vanish, i.e.\ $v\in N_{d}$ (Definition~2.7), so $[v]=0$ in $\mathcal{F}_{d}$, a contradiction. See Appendix~A for the certificates (\texttt{FieldSystem.Faithful}, \texttt{YM.Work.C26.answer}, \texttt{YM.Boundary.Soil.sees}); cyclicity (\texttt{registered\_words\_dense}) uses the full label collection, and the smaller observable family behind the scalar OS bridge (check A1) does not by itself prove it. \end{proof} \subsection{Relation to the mixed theory and Euclidean invariance} The mixed theory $S_{\mathrm{mix}}$ lacks clustering, so its OS reconstruction does not guarantee a unique vacuum; uniqueness of the vacuum holds for the selected phase and the selected reconstruction. The sixth chain consequence (\texttt{reconstructed\_words\_are\_selected\_boundary}) ensures that the vacuum expectation of every reconstructed relativistic field word equals the expectation of the same word in the selected phase's boundary distribution (via the common \texttt{qt.meets} certificate); the seventh gives uniqueness of boundary values only for one fixed analytic continuation. Neither addresses uniqueness of the infrared limit, of the phase, or of the continuation from Euclidean data. Euclidean invariance in the continuum limit has three parts (the source document \S6.4): (i) the gauge fixing preserves the lattice symmetries (Lemma~5.2); (ii) there is a divergence-free stress tensor used for local Ward identities (Section~9.1); (iii) Euclidean symmetry is not spontaneously broken in the limiting state because the stress-energy flux decays sufficiently rapidly at infinity, shown by an argument using reflection positivity in several orthogonal directions. We record part (iii) as a construction input underlying E1. \section{Identification as Yang--Mills theory} We record the properties identifying the construction as Yang--Mills theory. Throughout, $S^{s}_{v}(\cdot)$ and $S^{s}(v;\cdot)$ are used interchangeably, where $s$ refers either to the mixed theory or to a component theory indexed by $\xi\in X$. \subsection{The stress tensor and its Ward identity} \begin{assumption}[(Y1) uniqueness of the stress tensor] In $\mathcal{F}_{4}$, the elements that are symmetric rank-two tensors, fixed by field conjugation, centered with respect to $S_{\mathrm{mix}}$, transform as rank-two tensors under coordinate permutations and sign changes, and whose divergence vanishes in $\mathcal{F}_{5}$, form a one-dimensional real vector space; the stress tensor is the vector in this space whose normalization is fixed by the Ward identity below. Certificates: the source document \S7.1 and \texttt{WardFormulationAudit.lean}. \end{assumption} \begin{lemma}[Readable derivation of the candidate space] Weight-four symmetric rank-two tensor candidates are spanned by $A_{\mu\nu}=\Tr(F_{\mu\alpha}F_{\nu\alpha})$ and $B_{\mu\nu}=\delta_{\mu\nu}\Tr(F_{\alpha\beta}F_{\alpha\beta})$. On the free sourceless shell (Definition~2.10), writing $Q=\Tr(F_{\alpha\beta}F_{\alpha\beta})$, \[ \partial_{\mu}A_{\mu\nu}=\Tr\big((\partial_{\mu}F_{\mu\alpha})F_{\nu\alpha}\big) +\Tr\big(F_{\mu\alpha}\,\partial_{\mu}F_{\nu\alpha}\big)=0+Y, \] the first term vanishing by the sourceless Maxwell equation $\partial_{\mu}F_{\mu\alpha}=0$. For $Y=\Tr(F_{\mu\alpha}\partial_{\mu}F_{\nu\alpha})$, the Bianchi identity (Lemma~2.4) in the form $\partial_{\mu}F_{\nu\alpha}=\partial_{\nu}F_{\mu\alpha}-\partial_{\alpha}F_{\mu\nu}$ gives \[ Y=\tfrac12\partial_{\nu}Q-X, \qquad X=\Tr\big(F_{\mu\alpha}\partial_{\alpha}F_{\mu\nu}\big); \] exchanging dummy indices $\mu\leftrightarrow\alpha$ and using $F_{\alpha\mu}=-F_{\mu\alpha}$, $F_{\alpha\nu}=-F_{\nu\alpha}$ shows $X=Y$, so $Y=\frac14\partial_{\nu}Q$ and \[ \partial_{\mu}A_{\mu\nu}=\tfrac14\partial_{\nu}Q, \qquad \partial_{\mu}\Big(A_{\mu\nu}-\tfrac14\delta_{\mu\nu}Q\Big)=0. \] Thus $T_{\mu\nu}=A_{\mu\nu}-\frac14\delta_{\mu\nu}Q$ is divergence free; the divergence-free condition imposes one linear relation on the two-dimensional candidate space, leaving one dimension, which conjugation invariance and centering do not further reduce; the Ward identity fixes the overall normalization. \end{lemma} \begin{assumption}[(Y2) the two laws] For every label list $v_{0},\dots,v_{n}$ and distinct points $x_{0},\dots,x_{n}$: the \emph{flux (shell) law} \[ -\sum_{a}\int_{\R^{4}}\partial_{a}\chi(y)\, S_{\xi}\big(T_{ab},v_{0},\dots,v_{n}\big)(y,x_{0},\dots,x_{n})\,\dd^{4}y = S_{\xi}\big(D_{b}v_{0},v_{1},\dots,v_{n}\big)(x_{0},\dots,x_{n}), \] and the \emph{calculus law} \[ S_{\xi}\big(D_{b}v_{0},v_{1},\dots,v_{n}\big)(x_{0},\dots,x_{n}) = \partial^{x_{0}}_{b}\,S_{\xi}\big(v_{0},\dots,v_{n}\big)(x_{0},\dots,x_{n}), \] where $D_{b}$ is the abstract field derivative and $\chi$ is a real smooth compactly supported function equal to one near $x_{0}$ and supported away from the other $x_{j}$. Certificates: \texttt{phase.flux}, \texttt{phase.calculus}; \texttt{WardFormulationAudit.lean}; check A6. \end{assumption} \begin{proposition}[Ward identity] Under (Y2), for local labels $v_{0},\dots,v_{n}$, distinct points $x_{0},x_{1},\dots,x_{n}$, and $\chi$ as above, \begin{equation} -3\sum_{\mu=0}^{3}\int_{\R^{4}}\partial_{\mu}\chi(y)\, S_{\xi}\big(T_{\mu\nu},v_{0},\dots,v_{n}\big)(y,x_{0},\dots,x_{n})\,\dd^{4}y = \partial^{x_{0}}_{\nu}S_{\xi}\big(v_{0},\dots,v_{n}\big)(x_{0},\dots,x_{n}). \end{equation} \end{proposition} \begin{proof} Composing the two laws of (Y2), the right side of (9.1) is $\partial^{x_{0}}_{\nu}S_{\xi}(v_{0},\dots,v_{n})$, the coefficient $-3$ being fixed by the contraction of $\partial_{a}\chi$ with the $T_{ab}$ components relative to the construction's Euclidean coordinate and derivative conventions (the tensor normalization is fixed by this very identity). Three accompanying facts are proved separately. \emph{Integrability:} $\partial\chi$ vanishes near every insertion, so the integrand avoids all coincidence diagonals; by Theorem~5.17(iii) the kernel is $C^{\infty}$ on $\Omega_{n+1}$, and with $\partial\chi$ compactly supported the integral is a genuine integral rather than a totalized value. \emph{Window independence:} if $\chi_{1},\chi_{2}$ are both admissible and $\psi=\chi_{1}-\chi_{2}$ vanishes near all insertions, then \[ \int\partial_{a}\psi\,S(T_{ab},\dots)\,\dd y =-\int\psi\,\partial_{y^{a}}S(T_{ab},\dots)\,\dd y=0 \] by integration by parts (no boundary terms since $\psi$ is compactly supported), the divergence-free condition, and the vanishing of $\psi$ at the insertions. \emph{Coverage:} for every separated configuration there is an admissible window (its size may depend on the configuration, and no common shell exists as points collide); this is certified for every local label, every finite list of spectators of arbitrary degrees, and every character phase, by \texttt{admissible\_shell\_exists}, \texttt{tensor\_generates\_translations}, \texttt{shell\_integrand\_is\_integrable}, \texttt{shell\_charge\_is\_window\_independent}. \end{proof} \begin{remark} This check is sharper than stress-tensor conservation or the trace two-point function alone, but it does not prove that all possible stress-tensor improvements vanish (an improvement may leave the translation charge unchanged); it is a separated-insertion shell identity and does not construct an unrestricted contact-term extension, a dilation Ward identity, or a comparison with an independently chosen Minkowski sign convention. \end{remark} \subsection{A nonzero trace and its short-distance law} In the selected component theory indexed by a phase $\xi^{\ast}$, define \begin{equation} \Theta:=\sum_{\mu=0}^{3}T_{\mu\mu} \;-\; S_{\xi^{\ast}}\!\Big(\sum_{\mu=0}^{3}T_{\mu\mu}\Big)\,\mathbf{1}, \end{equation} that is, for any list of labels $v_{1},\dots,v_{n}$ and points $x_{0},\dots,x_{n}$, with $v_{0}:=\sum_{\mu}T_{\mu\mu}$, \begin{align*} &S_{\xi^{\ast}}\big(\Theta,v_{1},\dots,v_{n}\big)(x_{0},x_{1},\dots,x_{n}) = S_{\xi^{\ast}}\big(v_{0},v_{1},\dots,v_{n}\big)(x_{0},x_{1},\dots,x_{n})\\ &\qquad\qquad\qquad\qquad\qquad\qquad - S_{\xi^{\ast}}\big(v_{0}\big)(x_{0})\, S_{\xi^{\ast}}\big(v_{1},\dots,v_{n}\big)(x_{1},\dots,x_{n}). \end{align*} Thus $\Theta\in\mathcal{F}_{4}$ and $S_{\xi^{\ast}}(\Theta)=0$ (Appendix~A, \texttt{Trace.centered}). \begin{assumption}[(Y3) short-distance input] The RG analysis of the construction gives \begin{equation} K_{\Theta}(r) = \Big(\frac{\beta(g_{r})}{2g_{r}}\Big)^{2} \big\langle Q(\tfrac{r}{2}e_{0})\,Q(-\tfrac{r}{2}e_{0})\big\rangle_{\mathrm{free},c}\,(1+o(1)), \qquad Q=F^{a}_{\mu\nu}F^{a}_{\mu\nu}, \end{equation} as $r\downarrow0$, where $g_{r}=g(1/r)$, $\beta(g)=\mu\,\dd g/\dd\mu$, and $\langle\cdot\rangle_{\mathrm{free},c}$ is the connected two-point function of the free Gaussian curvature; the marginal character of the curvature-squared operator transmits the one-loop running into a logarithmic enhancement of the singularity, while the residue is the free-field value (certificates \texttt{YM.Trace.Over.Scribe}, \texttt{YM.Trace.Verse.tagline}). \end{assumption} \begin{lemma}[Free Wick contraction coefficient, complete verification] With \begin{equation} \big\langle F^{a}_{\mu\nu}(x)F^{b}_{\rho\sigma}(0)\big\rangle_{\mathrm{free}} = \frac{\delta^{ab}}{\pi^{2}|x|^{4}} \big(I_{\mu\rho}(x)I_{\nu\sigma}(x)-I_{\mu\sigma}(x)I_{\nu\rho}(x)\big), \qquad I_{\mu\nu}(x)=\delta_{\mu\nu}-2\frac{x_{\mu}x_{\nu}}{|x|^{2}}, \end{equation} one has \begin{equation} \big\langle Q(x)Q(0)\big\rangle_{\mathrm{free},c} = \frac{48\,\dim_{\R}\g}{\pi^{4}|x|^{8}}. \end{equation} \end{lemma} \begin{proof} Write $P_{\mu\nu}=x_{\mu}x_{\nu}/|x|^{2}$; then $P^{2}=P$ (rank-one projection), $I=\mathbf{1}-2P$, so \[ I^{2}=\mathbf{1}-4P+4P^{2}=\mathbf{1}, \qquad \Tr(I^{2})=4, \qquad \Tr(I^{4})=4. \] Let $A_{\mu\nu,\rho\sigma}=I_{\mu\rho}I_{\nu\sigma}-I_{\mu\sigma}I_{\nu\rho}$. The connected two-point function receives two (equal) Wick pairings and the color contraction $\sum_{a,b}\delta^{ab}\delta^{ab}=\dim_{\R}\g$; the index contraction is \[ \sum_{\mu\nu\rho\sigma}A_{\mu\nu,\rho\sigma}^{2} = \underbrace{\Big(\textstyle\sum I_{\mu\rho}^{2}\Big)\Big(\sum I_{\nu\sigma}^{2}\Big)}_{4\cdot4=16} +\underbrace{16}_{\text{second term}} -2\underbrace{\Tr(I^{4})}_{4} =16+16-8=24. \] Hence $\langle Q(x)Q(0)\rangle_{\mathrm{free},c} =2\cdot\dim_{\R}\g\cdot\frac{24}{\pi^{4}|x|^{8}} =\frac{48\dim_{\R}\g}{\pi^{4}|x|^{8}}$. \end{proof} \begin{remark}[Correction to the source document] The source document (\S7.2, Step~2) states ``using $I_{\mu\nu}I_{\mu\nu}=4$ and $I_{\mu\rho}I_{\nu\rho}=I_{\mu\nu}$, the full tensor contraction evaluates to $24/(\pi^{4}|x|^{8})$''. The strict computation is $I_{\mu\rho}I_{\nu\rho}=\delta_{\mu\nu}$ (since $I^{2}=\mathbf{1}$), and the cross term is $\Tr(I^{4})=4$; the value $24$ is unaffected. We use the matrix computation of Lemma~9.7 and do not rely on the misprinted identity. \end{remark} \begin{lemma}[One-loop and anomaly coefficients, complete derivation] With $\beta(g)=-b_{0}g^{3}+O(g^{5})$, $b_{0}=11C_{\g}/(48\pi^{2})$, and $\ell=\log\frac{1}{\Lambda r}$, \[ g^{2}(1/r)=\frac{1}{2b_{0}\ell}\,(1+o(1)), \qquad \frac{\beta(g_{r})}{2g_{r}}=-\frac{1}{4\ell}\,(1+o(1)). \] \end{lemma} \begin{proof} From $\frac{\dd(g^{-2})}{\dd\log\mu}=-\frac{2\beta(g)}{g^{3}}=2b_{0}+O(g^{2})$, integration gives $g^{-2}(\mu)=2b_{0}\log(\mu/\Lambda)+O(\log\log\mu)=2b_{0}\ell\,(1+o(1))$ at $\mu=1/r$. Hence $g_{r}^{2}=(2b_{0}\ell)^{-1}(1+o(1))$, and \[ \frac{\beta(g)}{2g}=\frac{-b_{0}g^{3}+O(g^{5})}{2g} =-\frac{b_{0}}{2}g^{2}+O(g^{4}) =-\frac{b_{0}}{2}\cdot\frac{1}{2b_{0}\ell}(1+o(1)) =-\frac{1}{4\ell}(1+o(1)).\qedhere \] \end{proof} \begin{theorem}[Trace asymptotic] With the centered trace $\Theta$ of (9.2), \begin{equation} \lim_{r\downarrow0} r^{8}\log^{2}(1/r)\, S_{\xi^{\ast}}\big(\Theta,\Theta\big)(-re_{0}/2,re_{0}/2) = \frac{3\,\dim_{\R}\g}{\pi^{4}}=:A_{G}>0, \end{equation} and Euclidean covariance upgrades this to the all-directions law: with $K_{\Theta}(x)=\Reop\,S_{\xi^{\ast}}(\Theta,\Theta)(x,0)$, \begin{equation} |x|^{8}\log^{2}(1/|x|)\,K_{\Theta}(x)\longrightarrow\frac{3\,\dim_{\R}\g}{\pi^{4}} \qquad (x\to0,\ x\neq0). \end{equation} \end{theorem} \begin{proof} \emph{Algebraic assembly.} By (Y3), Lemmas~9.7 and 9.9, writing $c(r)=r^{8}\log^{2}(1/r)$ and $\log(1/r)=\ell+O(1)$, \[ c(r)\,K_{\Theta}(r) = r^{8}\ell^{2}\cdot\frac{1}{16\ell^{2}}\cdot \frac{48\dim_{\R}\g}{\pi^{4}r^{8}}\,(1+o(1)) = \frac{3\dim_{\R}\g}{\pi^{4}}\,(1+o(1)). \] Note that $b_{0}^{2}$ from the two anomaly insertions cancels against $g_{r}^{4}\sim(2b_{0}\ell)^{-2}$ exactly, so the amplitude depends only on $\dim_{\R}\g$; the Yang--Mills value of $b_{0}$ is fixed separately in Section~9.4. Inside the formalization the constant appears as $\texttt{award}=12(d_{2}/b_{0})^{2}\dim_{\R}\g/\pi^{4}$ and the identity \texttt{Trace.Over.force\_b} ($d_{2}=-b_{0}/2$) reduces it to $3\dim_{\R}\g/\pi^{4}$ (certificate \texttt{YMAFBridge.canonical\_award}). \emph{From lattice-approximated points to the full limit (squeeze).} The formalized statement \texttt{TraceTest.shortDistance} first proves the limit at approximated positions: for $s$ near $1$, \[ c(r_{n})\,G_{\Theta}\big(r_{n}\delta_{n}(s)\big)\longrightarrow\frac{A_{G}}{s^{8}}, \qquad \delta_{n}(s)\to s, \] where $\delta_{n}(s)$ corrects for approximating physical points separated by $r_{n}s$ by nearby lattice points. By reflection positivity and the spectral representation, $G_{\Theta}$ is real-valued and nonincreasing, $G(r)=\int_{0}^{\infty}e^{-r\lambda}\,\dd\rho(\lambda)$ with $\rho\ge0$. Choosing $s_{-}<10$ and taking $\liminf$/$\limsup$, \[ \frac{A_{G}}{s_{+}^{8}} \le\liminf_{n}\,c(r_{n})G_{\Theta}(r_{n}) \le\limsup_{n}\,c(r_{n})G_{\Theta}(r_{n}) \le\frac{A_{G}}{s_{-}^{8}}, \] and $s_{\pm}\to1$ squeezes the limit to $A_{G}$. E1 (Euclidean covariance) then promotes the time-direction result to all directions, giving (9.7). \end{proof} \begin{remark} Although a classical smooth connection has traceless stress tensor, $\Theta$ is in general nonzero as a result of renormalization---the trace anomaly. In particular the two-point function is nonzero for sufficiently small $r$, so some smeared $\Phi_{\Theta}(f)$ is nonzero. The choice $\Lambda=1$ records the construction's reference units; it does not define an intrinsic physical scale. A position-space RG-improved version of the computation, in the same normalization, is given in [BM, eqs.~(3.72)--(3.76)]. \end{remark} \subsection{Excluded free-field descriptions} For $m\ge0$, an integer $q\ge0$, and a continuous real-valued function $h$ on $\R^{3}$, homogeneous of nonnegative integral degree, consider \begin{equation} K_{m,q,h}(r)=\int_{\R^{3}}\frac{E_{m}(p)^{q}\,h(p)^{2}}{2E_{m}(p)}\, e^{-rE_{m}(p)}\,\dd p, \qquad E_{m}(p)=\sqrt{m^{2}+|p|^{2}}, \end{equation} so that (9.8) and the factors $K_{ab}$ of (9.9) carry the same normalization; the convention with denominator $E_{m}(p)$ instead of $2E_{m}(p)$ differs by the constant factor $2$, which is absorbed into the arbitrary complex coefficients $c_{a}$ of (9.9) and leaves the class $\mathcal{W}$ of finite sums of finite products literally unchanged. and the class of finite complex linear combinations of finite products \begin{equation} W(r)=\sum_{a=1}^{N}c_{a}\prod_{b=1}^{n_{a}}K_{ab}(r), \qquad K_{ab}(r)=\int_{\R^{3}}\frac{E_{ab}(p)^{q_{ab}}\,h_{ab}(p)^{2}}{2E_{ab}(p)}\, e^{-rE_{ab}(p)}\,\dd^{3}p, \end{equation} with $E_{ab}(p)=\sqrt{m_{ab}^{2}+|p|^{2}}$, $N,n_{a}\in\N_{0}$, $c_{a}\in\C$, $m_{ab}\ge0$, $q_{ab}\in\N_{0}$, and $h_{ab}$ continuous, real-valued, and homogeneous of degree $j_{ab}\in\N_{0}$. This class includes the suitable polynomial observables and derivatives in free fields with finitely many masses. \begin{lemma}[Finite asymptotic expansion] Each $K_{m,q,h}$ has, as $r\downarrow0$, a finite expansion $\sum_{\nu=1}^{M}a_{\nu}r^{\alpha_{\nu}}(\log r)^{k_{\nu}}+O(1)$ with $a_{\nu}\in\R$, $\alpha_{\nu}\in\Z$, $k_{\nu}\in\N_{0}$; the class is closed under finite products and finite linear combinations of finitely many masses. \end{lemma} \begin{proof} Homogeneity of $h$ of degree $j$ makes the radial part of the integrand $p^{2j}E^{q-1}\varphi(\hat p)$ with $\varphi$ smooth on the sphere; the radial integral is $\int_{0}^{\infty}p^{2j}E^{q-1}\varphi\,e^{-rE}\,\dd p$. If $m=0$, then $E=p$, the integrand is $\propto p^{2j+q+1}e^{-rp}$, and $\int_{0}^{\infty}p^{s}e^{-rp}\,\dd p=\Gamma(s+1)r^{-(s+1)}$ gives integer powers. If $m>0$, the large-$p$ region $E=p+\frac{m^{2}}{2p}+\cdots$ gives integer powers $r^{-(2j+q+2)},r^{-(2j+q)},\dots$, and the small-$p$ region $E=m+\frac{p^{2}}{2m}+\cdots$ with $e^{-rE}=e^{-rm}(1+O(rp^{2}))$ gives $r^{0},r^{1},\dots$. When powers from the two regions resonate, the critical radial integral of type $\int p^{-1}\,\dd p$ produces $\log r$; each resonance raises the logarithmic power by one, and the number of resonances is finite, so $k_{\nu}$ is finite and $\ge0$. Products of two finite expansions are finite expansions of the same type ($r^{\alpha+\alpha'}(\log r)^{k+k'}$), and finite linear combinations are clear; hence every $W$ of the form (9.9) has a finite expansion of this type. Certificates: \texttt{YM.TraceExtension.Wick.real\_profile}, \texttt{AtOrder}, \texttt{Hierarchy.wickProfile}. \end{proof} \begin{lemma}[Finite limit forces zero, at the correct weight] Let $c(r)=r^{8}\log^{2}(1/r)$. If $W$ belongs to the class of Lemma~9.12 and $\lim_{r\downarrow0}c(r)\Reop W(r)$ exists and is finite, then the limit is $0$. \end{lemma} \begin{proof} Write $\Reop W(r)=\sum_{\nu}a_{\nu}r^{\alpha_{\nu}}(\log r)^{k_{\nu}}+O(1)$ (taking real parts affects only the coefficients, and $(\log r)^{k}$ differs from $\log^{k}(1/r)$ by a sign, irrelevant to powers and divergence). If $\alpha_{\nu}<-8$, $r^{\alpha_{\nu}+8}\log^{2+k_{\nu}}(1/r)\to\infty$, forcing $a_{\nu}=0$; if $\alpha_{\nu}=-8$, the term contributes $a_{\nu}\log^{2+k_{\nu}}(1/r)$, which diverges unless $a_{\nu}=0$ (note $k_{\nu}\ge0$); if $\alpha_{\nu}>-8$, the term tends to $0$; and the $O(1)$ term is killed by $c(r)\to0$. Hence a finite limit forces every surviving coefficient to vanish, and the limit is $0$. \end{proof} \begin{proposition}[Free-field exclusion] There is no $W$ of the form (9.9) and no $\delta>0$ such that $K_{\Theta}(r)=W(r)$ on $(0,\delta)$. \end{proposition} \begin{proof} By Theorem~9.10, $\lim_{r\downarrow0}c(r)K_{\Theta}(r)=A_{G} =3\dim_{\R}\g/\pi^{4}>0$. If $K_{\Theta}=W$ on $(0,\delta)$, then $\lim_{r\downarrow0}c(r)\Reop W(r)=A_{G}>0$, contradicting Lemma~9.13. \end{proof} \begin{remark}[Erratum on the weight] The source document (\S7.3, Steps~2--3, and Appendix~A, \texttt{TraceTest.excludes}) phrases the obstruction with the weight $r^{8}\log(1/r)$ and asserts ``by (9.7), $\lim r^{8}\log(1/r)K_{\Theta}=A_{G}>0$''; however the weight of (9.6)/(9.7) is $c(r)=r^{8}\log^{2}(1/r)$, and with the $\log^{1}$ weight one has $r^{8}\log(1/r)K_{\Theta}=A_{G}/\log(1/r)\to0$, which gives no positive limit. The two formulations are inconsistent. We adopt the weight $c(r)=r^{8}\log^{2}(1/r)$ of Theorem~9.10 and prove the exclusion by Lemma~9.13; the argument is then more direct and does not require the source classification that ``no term with $\alpha=-8$, $k\ge1$ occurs'' (under the $\log^{2}$ weight every term with $\alpha=-8$, $k\ge0$ diverges). The conclusion is unchanged and the proof is shorter. Note that the Lean certificates \texttt{YM.TraceExtension.Wick.no\_powerlog} and \texttt{euclidean\_excludes} check the $\log^{1}$-weight version; what is proved here is a mathematically self-consistent replacement argument in the $\log^{2}$ weight of Theorem~9.10, and we do not claim that those certificates cover the $\log^{2}$ weight. \end{remark} \begin{remark} The exclusion covers only the finite class (9.9): finitely many masses, finitely many factors, finite sums. It does not address descriptions by infinite composites of Gaussian fields, and it does not use reflection positivity or the squeeze argument of Theorem~9.10---two-point positivity enters only through $A_{G}>0$. The general exclusion of continuous-mass generalized free fields is outside the scope of the construction. \end{remark} \subsection{Asymptotic freedom and the running coupling} \begin{definition}[$g_{W}$] Starting from the bare couplings supplied by Theorem~5.17, perform RG blocking down to the first lattice spacing below $r$, and define $g_{W}(r)$ as the inverse square root of the resulting normalized coefficient of the marginal (curvature-squared) term in the effective action, after removal of the initial cutoff and extraction of a subsequential infinite-volume limit; the normalization requires the quadratic term of the blocked effective action to have identity coefficients (the kinetic normalization of Lemma~4.15). Independence of $g_{W}(r)$ from the subsequential limit is not proved. \end{definition} \begin{assumption}[(Y4) the one-loop value] At one loop, the coefficient of the marginal operator receives the gauge-boson loop, the Faddeev--Popov ghost loop, and the measure (BRS) contribution, which for a simple algebra with adjoint Casimir $C_{\g}$ sum to \begin{equation} b_{0}=\frac{1}{(4\pi)^{2}}\cdot\frac{11}{3}\,C_{\g}=\frac{11\,C_{\g}}{48\pi^{2}}, \end{equation} with no scalar or fermion loops. This is a pure integer/rational computation from the Casimir of $\g$ (certificates \texttt{YM.Pert.Asymptotics.b0}, \texttt{YM.Gauge.casimir}). \end{assumption} \begin{theorem}[Effective coupling at short distances] For small $r$ and any selected subsequential infinite-volume limit, $g_{W}(r)>0$, and \begin{equation} \log(1/r)\,g_{W}(r)^{2}\longrightarrow\frac{1}{2b_{0}} \qquad (r\downarrow0). \end{equation} \end{theorem} \begin{proof} \emph{Step 1 (recursion).} The recursion of Lemma~5.12, $\frac{\dd}{\dd\log\mu}g^{-2}=2b_{0}+O(g^{2})$, is equivalent to $\beta(g)=-b_{0}g^{3}+O(g^{5})$; integrating (Lemma~9.9) gives $g^{-2}(\mu)=2b_{0}\log(\mu/\Lambda)+O(\log\log\mu)$. \emph{Step 2 (value).} This is (Y4). \emph{Step 3 (from the lattice trajectory to $g_{W}$).} By construction, $g_{W}$ lies on the same RG trajectory as the bare couplings $\beta_{N,p}$, so $g_{W}^{-2}(r)=2b_{0}\log(1/r)+O(\log\log(1/r))$; taking reciprocals and multiplying by $\log(1/r)$ gives (9.11). \end{proof} \begin{remark} Two convention dependencies are explicit: $\Lambda$ is scheme-dependent (a change of scheme adds a constant to $\log(1/r)$ and does not affect (9.11)), and the normalization of $g_{W}$ is fixed by the kinetic normalization. The argument is perturbative in the marginal coupling only: it controls $g_{W}(r)$ for small $r$, where $g_{W}(r)\to0$, and does not address the infrared side of the flow, Landau-pole type questions, or uniqueness of $g_{W}$ across subsequences. A second coupling defined through separated continuum correlations---normalizing a scalar two-point derivative and an integrated tensor--scalar--scalar correlation to their free values and reading the coupling from a derivative of a tensor--scalar correlation---has ratio to $g_{W}$ tending to $1$, so the running law is visible in continuum observables as well as in the effective action. \end{remark} \begin{remark}[Free-field agreement at small scales] Along one common sequence $r_{k}\downarrow0$, the averaged $n$-point stress-tensor correlations, multiplied by $r_{k}^{4n}$ and evaluated at $r_{k}x_{1},\dots,r_{k}x_{n}$, converge to those of $\dim\g$ independent free Maxwell fields (the Gaussian two-form curvature obtained as the exterior derivative of a $\g$-valued one-form Gaussian free field), with every fixed number of derivatives, uniformly on compact sets avoiding collisions. This comparison holds for the mixed theory along the particular scaling sequence, whereas Theorems~9.10 and 9.19 hold as $r\downarrow0$ arbitrarily, and Theorem~9.10 applies to the selected phase $\xi^{\ast}$. \end{remark} \subsection{Operator product expansion and contact extensions} \begin{theorem}[Operator product expansion] Let $v_{i}$ and $u_{j}$ be local labels of degrees $d_{i}$ and $w=\sum_{i}d_{i}$. There exist labels $o_{\alpha}\in\mathcal{F}_{d_{\alpha}}$, finitely many per degree, such that for every $D\ge0$ and $\varepsilon>0$, \begin{equation} S^{s}\big(v,u(rx)\big) = \sum_{d_{\alpha}\le D} c_{\alpha}(v;rx)\,S^{s}\big(o_{\alpha},u(0)\big) + R^{s}_{D}, \qquad |R^{s}_{D}|\le r^{D+1-w-\varepsilon}, \end{equation} where $s$ denotes the mixed theory or a component theory. \emph{Uniform formulation:} for input fields $A_{i}\in\mathcal{F}_{d_{i}}$ at positions $x_{i}$, a spectator product $B=\prod_{\ell}B_{\ell}(y_{\ell})$, and an expansion point $o$, the same expansion reads \begin{equation} S^{s}\Big(\prod_{i}A_{i}(x_{i})\,B\Big) = \sum_{d_{\alpha}\le D}c_{\alpha}(A;x,o)\,S^{s}\big(O_{\alpha}(o)\,B\big) + R_{D,s}, \qquad |R_{D,s}|\le r^{D+1-w-\varepsilon}, \end{equation} with the spectators held at a fixed separation from $o$ and from each other (see (9.14) below); the short forms $u(rx)$, $u(0)$ of the source document, which suggest spectators scaled with $r$ or placed at the origin on a coincidence diagonal, are replaced by this single convention. More precisely: for fixed constants $M\ge0$, $\eta,c>0$, the allowed configurations at scale $r>0$ satisfy \begin{equation} \begin{aligned} \|x_{i}-o\|&\le Mr &&\text{for every }i,\\ \|x_{i}-x_{j}\|&\ge\eta r &&\text{for }i\neq j,\\ \|y_{\ell}-o\|&\ge c &&\text{for every }\ell,\\ \|y_{\ell}-y_{m}\|&\ge c &&\text{for }\ell\neq m, \end{aligned} \end{equation} the inputs forming a cluster near the expansion point $o$ while the spectators $u_{j}$ at $y_{\ell}$ stay separated; with $O_{\alpha}:=\texttt{T.actor}(d,i)$ and $c_{\alpha}(A;x,o):=\texttt{T.coefficient}(A,d,i,x,o)$, define, for $s=\mathrm{mix}$ or $s=\xi\in X$, \begin{equation} R_{D,s} = S^{s}\Big(\prod_{i}A_{i}(x_{i})\,B\Big) - \sum_{d_{\alpha}\le D}c_{\alpha}(A;x,o)\,S^{s}\big(O_{\alpha}(o)\,B\big). \end{equation} The bounds are \begin{equation} |c_{\alpha}(A;x,o)|\le r^{d_{\alpha}-w-\varepsilon}, \qquad |R_{D,s}|\le r^{D+1-w-\varepsilon}, \end{equation} holding for all sufficiently small $r>0$ for every $\varepsilon>0$; the coefficient bound requires only the first two conditions of (9.14), the remainder bound requires all four and holds uniformly over all states $s\in\{\mathrm{mix}\}\sqcup X$. Setting $o=0$ and input positions $x_{i}=r\hat x_{i}$ with fixed $\hat x_{i}$ (and the spectator product $B$ held fixed) recovers the short form of (9.12); (9.13) is the same statement with the geometric conditions made explicit. The coefficients are independent of spectators, state, and truncation degree; their independence of the chosen subsequential infinite-volume limit is not proved, and common coefficients of two expansions do not by themselves imply equal correlations. The cutoff $D$ counts local-observable label degrees, not loop orders; perturbative agreement to all orders remains unproved. \end{theorem} \begin{proof}[Proof: extraction of the coefficients and bounds] The coefficients $c_{\alpha}(A;x,o)$ are produced by a fixed algebraic prescription on the label spaces before any correlation is evaluated; their independence of spectators, phases, and truncation orders is automatic rather than dynamical. (1)~\emph{Degree decomposition:} for each $d$ fix a splitting $\mathcal{F}_{d}=\mathcal{F}_{d-1}\oplus E_{d}$; the channels $\alpha=(d,i)$ index a conjugation-invariant basis of the new layer $E_{d}$ (\texttt{YM.OPE.Orbits.Reader.channels}, \texttt{.basis}, \texttt{.actor}), with $n_{d}=\dim E_{d}$ fields per layer (\texttt{YM.OPE.Orbits.Reader.fresh}). (2)~\emph{Test-field formation:} form a degree-$d$ field from finitely many correlations of the inputs against additional test insertions supported on a fixed ``stratum'' configuration (\texttt{YM.OPE.Orbits.Reader.cutTest}, \texttt{.stratum}); the separation hypotheses (9.14) keep the stratum tests away from both the input cluster and the spectators. (3)~\emph{Symmetrization and projection:} average over $O(4)$ (\texttt{YM.OPE.Orbits.averageV}), symmetrize under label conjugation (\texttt{YM.OPE.Orbits.checkReal}), and strip the $\mathcal{F}_{d-1}$ components (\texttt{YM.OPE.shadow}); the coordinates in the basis $O_{(d,i)}$ define $c_{\alpha}(A;x,o)$ (\texttt{YM.OPE.Orbits.Reader.coef}, packaged in \texttt{YM.OPE.Prescription.ope}). (4)~\emph{Bounds:} because the prescription is fixed, the coefficient bound in (9.16) requires only the first two conditions of (9.14) (\texttt{YM.OPE.Orbits.coefficient\_separated}); the remainder bound requires all four, uniformly over $s$ (\texttt{YM.OPE.Orbits.quantitative\_separated}). The analytic estimates entering Step (4) are those of the submitted proof; no perturbative computation or comparison of coefficient values is involved. \end{proof} \begin{proposition}[Compact-set uniform bounds] Fix labels, truncation degree $D$, and $\varepsilon>0$, and let $\mathcal{K}$ be a compact set of configurations satisfying (9.14) (input cluster and spectator sets in fixed compacts, uniformly separated), with $N_{D}=\#\{\alpha:d_{\alpha}\le D\}<\infty$. Suppose the pointwise bounds (9.16) hold on $\mathcal{K}$ for all $s$. Then any norm of the complete coefficient vector $c=(c_{\alpha})_{\alpha}$ is uniformly bounded on $\mathcal{K}$, and the remainder bound holds uniformly in $(x,o,y)\in\mathcal{K}$ and $s$. \end{proposition} \begin{proof} From the pointwise bound, $\|c\|_{\infty}\le\max_{\alpha}r^{d_{\alpha}-w-\varepsilon}=:B_{D}$, a finite constant; in the finite-dimensional space $\C^{N_{D}}$ all norms are equivalent, so there is $C_{D}$ with $\|c\|\le C_{D}\|c\|_{\infty}\le C_{D}B_{D}$, independent of the configuration. The remainder bound's right side contains no configuration or state, so $\sup_{\mathcal{K},s}|R_{D,s}|\le r^{D+1-w-\varepsilon}$. (Compactness is not needed for the uniform bound---it ensures the quantification over ``every compact separated cluster and spectator set'' is well formed.) \end{proof} \begin{remark}[Contact extensions: the submission's predicate] The contact-term extension is isolated behind an explicit predicate in the submission's public interface (\texttt{ContactSpec.lean}). For a hierarchy $T$, a state option $s:\texttt{Option T.Character}$, and any number $n$ of insertions with labels $v:\texttt{Fin n}\to\texttt{T.Tag}$, \[ \texttt{Contacts}\;:\Longleftrightarrow\; \exists\,D:\ \text{test functions on the \emph{full} configuration space }(\R^{4})^{n} \;\longrightarrow\;\C, \] continuous and linear, such that every test function $f$ whose tubular support lies in the separated insertion region satisfies $D(f)=\int_{(\R^{4})^{n}} f(x)\,T.\texttt{correlation}(s,v,x)\,\dd x$. The distribution $D$ lives on the whole space, including every collision diagonal, and restricts to the already constructed correlation on separated insertions. The submission's bridge theorem (\texttt{ContactBridge.lean}) proves \texttt{Contacts} for the constructed hierarchy, transporting the OPE-based multiple-contact development (harmonic extension around the collision locus, with the extension recorded so that evaluation on separated tests reproduces the original correlation). Three limitations remain: the predicate does not assert uniqueness of the extending distribution; it does not prove the full OS axioms (reflection positivity) for the extended correlations on unrestricted Euclidean test spaces; and lattice convergence of the extended functionals is not established. Relatedly, diagonal-flat Schwartz tests annihilate distributions supported on the diagonals, so the independent OS check of Appendix~C does not touch contact-term uniqueness. \end{remark} \subsection{The field-content audit} The dictionary between the constructed operator spaces and classical gauge-invariant polynomials is checked in six steps (\texttt{YMReview.\allowbreak \allowbreak Audits.\allowbreak \allowbreak OperatorContent}, worked example \texttt{OperatorContent-20261028.\allowbreak \allowbreak lean}; check A2 of Table~1). The registered label collection is \texttt{Labels.\allowbreak \allowbreak Pol G} with its invariant subspace \texttt{Labels.\allowbreak \allowbreak invariantSpace G}; a label in \texttt{Labels.\allowbreak \allowbreak Label G d} carries a finite weight bound $d$, and no fixed upper degree is imposed on the full collection. The audit is stated in the identifiers of Appendix~A, which we use verbatim; in the notation of the main text, \texttt{S.\allowbreak \allowbreak Field}\,$d$ is the quotient $\mathcal{F}_{d}=V_{d}/N_{d}$ of Definition~2.7, \texttt{S.\allowbreak \allowbreak null}\,$d$ is the correlation null space $N_{d}$, \texttt{S.\allowbreak \allowbreak Union} is the label space $V$, \texttt{F.\allowbreak \allowbreak }$\alpha$\texttt{.\allowbreak \allowbreak space.\allowbreak \allowbreak }$X$ is the phase space $X$ of Proposition~7.10, \texttt{F.\allowbreak \allowbreak }$\alpha$\texttt{.\allowbreak \allowbreak ensemble}$(i)$ is the phase hierarchy of character $i$, and \texttt{F.\allowbreak \allowbreak action.\allowbreak \allowbreak operators.\allowbreak \allowbreak field} is the field-operator map of Section~8. \begin{enumerate} \item \emph{Exhaustiveness of the register.} Every invariant polynomial $p\in\texttt{Labels.\allowbreak \allowbreak invariantSpace G}$ appears at some register level: there exist $d$ and $v\in\texttt{Labels.\allowbreak \allowbreak Label G d}$ with $\texttt{v.\allowbreak \allowbreak val}=p$ (\texttt{every\_registered\_polynomial\_appears}, from \texttt{OperatorCompleteness.\allowbreak \allowbreak raw\_register\_exhaustive}). This is exhaustiveness relative to the submitted algebraic dictionary; human review must still identify that dictionary with the intended local gauge-invariant fields, and the theorem does not address nonlocal observables such as an independently specified class of line defects. \item \emph{Free on-shell comparison space.} At filtration $d$, the free on-shell polynomial comparison space \texttt{Maxwell.\allowbreak \allowbreak shellSector G d} is exactly the invariant part of the free lower-weight cap (\texttt{free\_sector\_dictionary}, from \texttt{OperatorCompleteness.\allowbreak \allowbreak free\_sector\_exhaustive}), and the separated free correlations leave no further residual null space: \texttt{Maxwell.\allowbreak \allowbreak residual G d} $=\bot$ (\texttt{free\_residual\_vanishes}, from \texttt{free\_no\_extra\_nulls}). These identify the comparison space; they are not claims that interacting correlation functions are Gaussian. \item \emph{Interacting quotient dictionary.} Writing $\texttt{S.\allowbreak \allowbreak null}\,d$ for the kernel of the map from raw labels to all separated spectator correlations and $\texttt{S.\allowbreak \allowbreak Field}\,d=V_{d}/\texttt{S.\allowbreak \allowbreak null}\,d$ as in Section~2, the rank comparison of the submission, reached through \texttt{F.\allowbreak \allowbreak phase.\allowbreak \allowbreak transport.\allowbreak \allowbreak comparator}, yields a complex linear equivalence \[ \texttt{S.\allowbreak \allowbreak Field}\,d\;\simeq_{\C}\;\texttt{Maxwell.\allowbreak \allowbreak shellSector G d} \qquad\text{for every }d \] (\texttt{interacting\_field\_dictionary}, from \texttt{OperatorCompleteness.\allowbreak \allowbreak local\_field\_content}). This constrains the size of the interacting quotient at every filtration level. An equivalence of these vector spaces is not an equality of interacting and free correlators, an algebra isomorphism, or a unitary equivalence of quantum field theories; the source comparison also tracks the real signed-coordinate structure, which the display above suppresses. \item \emph{No field is lost in a phase.} For every character phase $i\in\texttt{F.\allowbreak \allowbreak }\alpha\texttt{.\allowbreak \allowbreak space.\allowbreak \allowbreak }X$---not merely almost every phase under the decomposition measure---the phase and mixed null spaces agree, and the preceding equivalence transports: \[ \texttt{F.\allowbreak \allowbreak }\alpha\texttt{.\allowbreak \allowbreak ensemble}(i).\texttt{Field}\,d\;\simeq_{\C}\; \texttt{Maxwell.\allowbreak \allowbreak shellSector G d} \qquad\text{for every }d,\ i \] (\texttt{each\_phase\_retains\_the\_dictionary}, from \texttt{OperatorCompleteness.\allowbreak \allowbreak phase\_field\_content}, via \texttt{F.\allowbreak \allowbreak phase.\allowbreak \allowbreak phase.\allowbreak \allowbreak retains i d}). This is stronger than the statement that the mixed correlation function is the average of its phase values. \item \emph{Cyclicity uses the full collection.} \texttt{F.\allowbreak \allowbreak action.\allowbreak \allowbreak operators.\allowbreak \allowbreak vector} assigns a Hilbert-space vector to every registered field word applied to the vacuum, and the linear span of its range is dense (\texttt{full\_word\_domain\_is\_dense}, from \texttt{OperatorCompleteness.\allowbreak \allowbreak registered\_words\_dense}). The assertion uses the full field collection: the smaller observable family behind the scalar OS bridge of check A1 does not by itself prove this cyclicity statement. Cyclicity is partly a feature of reconstructing a Hilbert space from field words; its physical force comes from combining it with the field-content comparison and the faithful representation, not from relabeling the constructed space as an independently defined canonical one. \item \emph{Surviving fields act nontrivially.} Every nonzero local quotient field has some nonzero smeared operator: for $p\in\texttt{S.\allowbreak \allowbreak Union}$, $p\neq0$, there is a test $f$ with $\texttt{F.\allowbreak \allowbreak action.\allowbreak \allowbreak operators.\allowbreak \allowbreak field}\ p\ f\neq0$ (\texttt{surviving\_field\_has\_nonzero\_operator}, from \texttt{OperatorCompleteness.\allowbreak \allowbreak quotient\_fields\_not\_lost}). Passing to the reconstructed operator representation therefore kills no surviving local field class; it does not assert that every smearing of a nonzero field is nonzero. \end{enumerate} The audit concerns the full registered invariant collection, not only the charge-conjugation even stress trace; it does not supply an independently checked charge-conjugation-equivariant dictionary or a classification of charge-conjugation odd particles, which would need their own precise statements. In the notation of Appendix~A, the quotient and injectivity steps correspond to \texttt{Hierarchy.\allowbreak \allowbreak Field}, \texttt{toLabel}, and \texttt{Properties.\allowbreak \allowbreak injective}, \texttt{covers}, \texttt{represents}, \texttt{acts}, \texttt{faithful}. The transitive axiom dependencies of all six theorems are printed in the file and use only the permitted foundations. \section{Open problems: precise formulations and a logical boundary} The ten open problems recorded in the source documents are restated here as statements of the form ``which additional input is missing'', together with two proved logical boundaries: the sixteen registered inputs do not imply a mass gap, and a mass gap is equivalent to a uniform exponential rate in E4 (Proposition~10.22). Throughout, (MG1)--(MG4) and (O1)--(O10) are \emph{proposed} conditional inputs, of the form ``if (MGk) holds, then \dots''; none of them is claimed as proved, none is registered in Appendix~B, and none carries a repository certificate. The four candidates (MG1)--(MG4) are projections of a single uniform infrared estimate (MG), registered in Proposition~10.34. \subsection{A spectral representation toolkit} \begin{lemma}[Exponential decay versus spectral support] Let $\rho$ be a finite positive measure on $[0,\infty)$ and $G(\tau)=\int_{[0,\infty)}e^{-\tau\lambda}\,\dd\rho(\lambda)$. For any $m>0$, the following are equivalent: \begin{enumerate} \item[\upshape(i)] $\rho\big((0,m)\big)=0$; \item[\upshape(ii)] there exists $K<\infty$ such that $\tilde G(\tau):=\int_{(0,\infty)}e^{-\tau\lambda}\,\dd\rho(\lambda)\le Ke^{-m\tau}$ for all $\tau\ge0$. \end{enumerate} \end{lemma} \begin{proof} Write $\tilde\rho=\rho-\rho(\{0\})\delta_{0}$. (i)$\Rightarrow$(ii): if $\supp\tilde\rho\subset[m,\infty)$, then $\tilde G(\tau)=e^{-m\tau}\int_{[m,\infty)}e^{-\tau(\lambda-m)}\,\dd\tilde\rho(\lambda) \le e^{-m\tau}\|\rho\|$, so $K=\|\rho\|$ works. (ii)$\Rightarrow$(i): suppose $a:=\tilde\rho\big((0,m-\varepsilon)\big)>0$ for some $\varepsilon>0$; then $\tilde G(\tau)\ge\int_{(0,m-\varepsilon)}e^{-\tau\lambda}\,\dd\tilde\rho \ge a\,e^{-(m-\varepsilon)\tau}$, so $a\,e^{\varepsilon\tau}\le K$ for all $\tau$, which is impossible. Hence $\tilde\rho\big((0,m-\varepsilon)\big)=0$ for all $\varepsilon>0$, i.e.\ $\rho((0,m))=0$. \end{proof} \begin{lemma}[Gap implies uniform exponential clustering] Under (PH4), if $\sigma(H)\subset\{0\}\cup[\Delta,\infty)$, then for all $\psi,\chi\in\mathcal{H}$, \[ \big|\langle\psi,E(s)\chi\rangle-\langle\psi,\Omega\rangle\langle\Omega,\chi\rangle\big| \le e^{-\Delta s}\, \big\|\psi-\langle\Omega,\psi\rangle\Omega\big\|\, \big\|\chi-\langle\Omega,\chi\rangle\Omega\big\|. \] \end{lemma} \begin{proof} Let $\chi_{\perp}=\chi-P_{0}\chi$ and $\psi_{\perp}=\psi-P_{0}\psi$; by (PH4) $P_{0}=|\Omega\rangle\langle\Omega|$, so $\langle\psi,E(s)\chi\rangle-\langle\psi,P_{0}\chi\rangle =\langle\psi_{\perp},E(s)\chi_{\perp}\rangle$. By the spectral theorem this equals $\int_{[\Delta,\infty)}e^{-s\lambda}\,\dd\mu_{\psi_{\perp},\chi_{\perp}}(\lambda)$, and $\|\mu_{\psi_{\perp},\chi_{\perp}}\|\le\|\psi_{\perp}\|\,\|\chi_{\perp}\|$ by polarization and Cauchy--Schwarz; the claim follows. \end{proof} \begin{proposition}[Dense channel coverage gives the full gap] Let $\mathcal{D}_{0}\subset\Omega^{\perp}$ be dense, and suppose there exist $\mu>0$ and $K(\psi)<\infty$ such that \[ \big|\langle\psi,E(s)\psi\rangle\big|\le K(\psi)\,e^{-\mu s} \qquad(\psi\in\mathcal{D}_{0},\ s\ge0). \] Then $\sigma(H)\subset\{0\}\cup[\mu,\infty)$. \end{proposition} \begin{proof} For $\psi\in\mathcal{D}_{0}$ we have $\rho_{\psi}(\{0\})=\|P_{0}\psi\|^{2}=0$, so Lemma~10.1 gives $\rho_{\psi}((0,\mu))=0$. Now suppose $(0,\mu)\cap\sigma(H)\neq\varnothing$ and pick $\lambda_{0}\in(0,\mu)\cap\sigma(H)$ and $\varepsilon>0$ with $I:=[\lambda_{0}-\varepsilon,\lambda_{0}+\varepsilon]\subset(0,\mu)$. Then $E_{H}(I)\neq0$; take a non-zero $\phi$ in its range, so that $\rho_{\phi}(I)=\|\phi\|^{2}>0$. Choose a continuous $h:[0,\infty)\to[0,1]$ with $h\equiv1$ on $I$ and $\supp h\subset(0,\mu)$ (Urysohn's lemma). By the functional calculus $h(H)$ is bounded with $\|h(H)\|\le1$. Since $\mathcal{D}_{0}$ is dense in $\Omega^{\perp}$, choose $\psi_{j}\in\mathcal{D}_{0}$ with $\psi_{j}\to\phi$. Then \[ \int h\,\dd\rho_{\psi_{j}}=\langle\psi_{j},h(H)\psi_{j}\rangle \longrightarrow\langle\phi,h(H)\phi\rangle =\int h\,\dd\rho_{\phi}\ge\rho_{\phi}(I)=\|\phi\|^{2}>0, \] while $\rho_{\psi_{j}}((0,\mu))=0$ and $\supp h\subset(0,\mu)$ force $\int h\,\dd\rho_{\psi_{j}}=0$ for every $j$, a contradiction. Hence $(0,\mu)\cap\sigma(H)=\varnothing$. \end{proof} \begin{remark}[The density hypothesis is the field-word density] The dense subspace $\mathcal{D}_{0}$ is supplied by the field-word cyclicity (\texttt{registered\_words\_dense}, field-audit step 5); the smaller observable family behind the scalar OS bridge (check A1) does not suffice. An exponential rate in a single channel (e.g.\ the trace two-point function) therefore yields only a \emph{channel gap}, not the full gap of Definition~1.2. Moreover the rate must be common to all registered label words: (Y2) supplies no degree-uniformity (Proposition~10.26), and a model with per-tower exponential rates but no common rate exists (Proposition~10.28, calculus-law part). \end{remark} \subsection{A model within the registered inputs without a mass gap} Let \[ \rho:=\delta_{0}+\sum_{k\ge1}2^{-k}\delta_{2^{-k}}\quad\text{on }[0,\infty), \] a finite positive measure with $\|\rho\|=2$ and $\supp\rho\subset[0,\tfrac12]$. Let $\mathcal{H}=L^{2}([0,\infty),\dd\rho)$, let $H$ be the multiplication operator $(H\psi)(\lambda)=\lambda\psi(\lambda)$ ($H\ge0$), let $\Omega=\mathbf{1}/\sqrt{2}$, and let $P_{0}$ be the projection onto the one-dimensional atom at $\lambda=0$. Take the generalized free field with K\"allen--Lehmann spectral measure $\rho$ [GJ, OS1]: its two-point function is \begin{equation} G(\tau)=\int_{(0,\infty)}e^{-\tau\lambda}\,\dd\tilde\rho(\lambda) =\sum_{k\ge1}2^{-k}e^{-\tau 2^{-k}}, \qquad \tilde\rho=\rho-\delta_{0}, \end{equation} and the full hierarchy is given by the Wick formula. \begin{proposition}[Logical independence of the mass gap] There is an OS hierarchy satisfying every clause of (AN1) (E0, E0$'$, E1, E2, E3, E4 clustering, normalization, analyticity) and (PH1)--(PH4), whose reconstructed Hamiltonian satisfies \[ \sigma(H)=\{0\}\cup\{2^{-k}:k\ge1\}, \] so that $0$ is an accumulation point of the spectrum and there is no mass gap; yet E4 clustering and vacuum uniqueness hold. \end{proposition} \begin{proof} \emph{The OS axioms.} E1 (Euclidean covariance) and E3 (permutation) are inherited from the construction of the generalized free field; E0 holds because the compactly supported $\rho$ gives finite moments, holomorphy of the Laplace transform for $\operatorname{Re}\tau>0$, and tempered kernels at separated points (diagonal singularities are handled by the $\mathcal{S}_{0}$ conditions of Appendix~C); E0$'$ holds by Wick counting: the $2n$-point function is a sum of $(2n-1)!!$ products of two-point functions, and compact support gives uniform bounds on separated sets, yielding growth of type $\|\mathcal{S}_{2n}(f)\|\le\beta^{2n}(n!)^{A}\|f\|_{(ns,ns)}$; E2 holds because $G$ is the Laplace transform of a positive measure, hence completely monotone, and reflection positivity of the two-point function is inherited by the Gaussian (Wick) hierarchy; analyticity follows from holomorphy of the Laplace transform on $\operatorname{Re}\tau>0$. (PH1)--(PH4): the model is time-translation invariant with trivial ground-state algebra action, commutativity, faithfulness of the vacuum vector for the scalars, and $P_{0}$ one-dimensional by construction. \emph{(a) Clustering.} By dominated convergence, $\tilde G(\tau)\to0$ as $\tau\to\infty$, i.e.\ the connected two-point function tends to $0$; all connected functions of the Wick hierarchy are products of connected two-point functions, so E4 holds. \emph{(b) The clustering rate is only power law.} Let $k_{0}=\lfloor\log_{2}\tau\rfloor$ ($\tau\ge2$). \emph{Lower bound:} taking $k=k_{0}$, we have $2^{k_{0}}\le\tau<2^{k_{0}+1}$, so $2^{-k_{0}}\ge1/\tau$ and $1\le\tau2^{-k_{0}}<2$, hence $G(\tau)\ge2^{-k_{0}}e^{-\tau2^{-k_{0}}}\ge e^{-2}/\tau$. \emph{Upper bound:} for $k>k_{0}$, $\sum_{k>k_{0}}2^{-k}=2^{-k_{0}}\le2/\tau$; for $k\le k_{0}$, writing $j=k_{0}-k\ge0$, we have $2^{-k}=2^{j}2^{-k_{0}}\le2^{j}/\tau$ and $\tau2^{-k}\ge2^{j-1}$, so this part is at most $\frac1\tau\sum_{j\ge0}2^{j}e^{-2^{j-1}}=C/\tau$. Therefore \[ \frac{e^{-2}}{\tau}\le G(\tau)\le\frac{C+2}{\tau}\qquad(\tau\ge2), \] i.e.\ $G(\tau)\asymp\tau^{-1}$; in particular $e^{m\tau}G(\tau)\to\infty$ for every $m>0$. \emph{(c) No spectral gap.} $\supp\tilde\rho=\{2^{-k}\}$ accumulates at $0$, so $\inf\big(\sigma(H)\setminus\{0\}\big)=0$ and $\rho((0,\varepsilon))>0$ for every $\varepsilon>0$; by Lemma~10.1, no spectral gap holds. \emph{(d) Logical conclusion.} If ``the theory has a mass gap'' were derivable from (AN1), (PH1)--(PH4), and the standard OS theorems, it would hold in every model of those axioms; (c) exhibits a counterexample. Hence a uniform mass gap requires input beyond the sixteen registered assumptions. By Propositions~10.7 and 10.9 this remains true even when the ultraviolet trace law, the Ward identities, the field-content dictionary and the OPE data are imposed: those data are insensitive to the spectral weight near zero. \end{proof} \begin{remark}[What the model does and does not say] The model does \emph{not} claim that the constructed Yang--Mills theory has no mass gap; it is a generalized free field proving that the axioms together with the registered inputs do not force one. It lies precisely outside the exclusion of Proposition~9.14, which covers only the finite class of (9.9) (finitely many masses, factors, and sums); the general exclusion of continuous-mass generalized free fields remains open (problem O6 below). Consequently O6 is a \emph{necessary but not sufficient} prerequisite for a mass gap: until such infrared descriptions are excluded, a gapless model compatible with all registered inputs stands; and even after exclusion, (MG) must supply a positive lower bound (Proposition~10.24). This also shows that ultraviolet data --- the trace anomaly (a renormalization effect, Remark~9.11), asymptotic freedom, the OPE error bounds --- constrain the infrared spectral weights of such a model not at all. \end{remark} \begin{proposition}[Ultraviolet data do not constrain the infrared spectrum] There exists a tempered generalized-free-field hierarchy satisfying (AN1) E0--E4, vacuum uniqueness, the field-content dictionary (B1), the stress-tensor candidate calculation (Y1), the calculus and flux laws (Y2), the exact short-distance constant $A_{G}=3\dim_{\R}\g/\pi^{4}$ of Theorem~9.10, and the OPE coefficients and remainder bound of Theorem~9.22, whose Hamiltonian has no spectral gap. \end{proposition} \begin{proof} \emph{The spectral weight.} Work with a K\"allen--Lehmann \emph{weight} $\rho$ on $[0,\infty)$ which is locally finite and of polynomial growth (tempered); for every Schwartz test $f$ the induced vector spectral measure $\rho_{f}=|\hat f|^{2}\rho$ is finite, so Lemma~10.1 and all spectral arguments apply to $\rho_{f}$. Set \[ \rho=\delta_{0}+\rho_{\mathrm{UV}}+\rho_{\mathrm{IR}}, \qquad \rho_{\mathrm{IR}}=\sum_{k\ge1}2^{-k}\delta_{2^{-k}}, \qquad \rho_{\mathrm{UV}}(\dd E)=c\,\frac{E^{5}}{\log^{2}(E+e)}\,\mathbf{1}_{E\ge1}\,\dd E . \] The two nontrivial pieces have disjoint supports: $\supp\rho_{\mathrm{IR}}\subset(0,\tfrac12]$, $\supp\rho_{\mathrm{UV}}\subset[1,\infty)$. Let $G(\tau)=\int e^{-\tau E}\,\dd\rho(E)$ and let the full hierarchy be Gaussian/Wick with two-point kernel $G(x-y)\,\langle v,w\rangle_{\mathrm{label}}$, the label Gram matrix being the one of Section~2. \emph{Short distance.} The four-dimensional Euclidean propagator of mass $E$ is $\Delta_{E}(r)=\frac{E}{4\pi^{2}r}K_{1}(Er)$, with $\Delta_{E}(r)\asymp\frac{1}{4\pi^{2}r^{2}}\mathbf{1}_{Er\lesssim1}$ and exponential suppression for $Er\gg1$. Hence \[ K(r)\ \asymp\ \frac{1}{4\pi^{2}r^{2}}\int_{0}^{1/r}\dd\rho(E)\ \sim\ \frac{c}{24\pi^{2}}\,r^{-8}\log^{-2}(1/r), \] because $\int_{1}^{1/r}E^{5}\log^{-2}E\,\dd E\sim\frac{1}{6}(1/r)^{6}\log^{-2}(1/r)$. The infrared piece has finite total mass and bounded support, so it contributes at most $\|\rho_{\mathrm{IR}}\|/(4\pi^{2}r^{2})=O(r^{-2})$, a strictly lower-order singularity. Choosing $c$ fixes the leading coefficient to $A_{G}$; the assembly of Theorem~9.10 (Lemma~9.7 times Lemma~9.9) involves only the large-$E$ Taylor coefficients and never touches the spectrum below $E=1$, so the constant $3\dim_{\R}\g/\pi^{4}$ is unchanged. \emph{Long distance.} The ultraviolet piece gives $O(e^{-\tau})$ as $\tau\to\infty$; the infrared piece gives $\sum_{k}2^{-k}e^{-\tau2^{-k}}\asymp\tau^{-1}$ by the computation of Proposition~10.5(b). Hence E4 clustering holds with a power-law rate, $e^{m\tau}G(\tau)\to\infty$ for every $m>0$, and $\inf(\sigma(H)\setminus\{0\})=0$ because $\{2^{-k}\}$ accumulates at $0$. By Lemma~10.1 there is no gap. \emph{The axioms and the dictionary.} $\rho\ge0$ gives E2 (complete monotonicity of $G$); E0/E0$'$ follow from polynomial growth plus Wick counting exactly as in Proposition~10.5, with $\|\rho\|$ replaced by the polynomial-growth constant; E1/E3 and analyticity are inherited; $\rho$ has no atom at $0$ other than $\delta_{0}$, so by Proposition~7.2 E4 is equivalent to vacuum uniqueness, which holds. Because the infrared modification multiplies the \emph{scalar} two-point factor and leaves the label Gram matrix untouched, the correlation null spaces $N_{d}$ are unchanged, hence (B1) (a rank comparison, Remark~2.13) is unchanged. (Y1) is the purely algebraic computation of Lemma~9.2 on the free shell and is mass-independent. For (Y2), the calculus law holds for any translation-invariant Gaussian covariance, since $\partial_{x_{0}}$ differentiates the scalar factor; the flux law's shell integral is the translation generator and likewise depends only on the fact that \emph{every} insertion, including $T_{ab}$, shares the same two-point factor. Finally, the OPE coefficients $c_{\alpha}$ are produced by the fixed algebraic prescription of Theorem~9.22 before any correlation is evaluated, from the short-distance Taylor coefficients; the $O(r^{-2})$ infrared contribution does not enter the $r^{-8}\log^{-2}$ layer, and the remainder $\int_{E>1/r}$ is controlled by $\rho_{\mathrm{UV}}$. \end{proof} \begin{remark} This upgrades Proposition~10.5: the counterexample already satisfies the ultraviolet law, the Ward identities, the field-content dictionary and the OPE data. It also shows that the trace anomaly --- a renormalization effect (Remark~9.11) --- is a \emph{necessary constraint, not a dynamical explanation} of the infrared spectrum. The model remains a generalized free field, so (Y3)/(Y4) and the RG blocking mechanism have no counterpart in it; this is recorded in Remark~10.42. \end{remark} \begin{proposition}[The gap is not continuous in ultraviolet data] Let $T$ be a generalized free field with a gap $\Delta>0$ (a single pole at $E=\Delta$ plus the ultraviolet tail of Proposition~10.7), and let $T_{\epsilon}$ be obtained by adjoining $\epsilon\rho_{\mathrm{IR}}$, $\epsilon>0$. Then $T_{\epsilon}$ and $T$ have identical full formal short-distance asymptotics, identical OPE coefficients, identical Ward identities and identical field-content ranks; all separated-point correlations differ by at most $C\epsilon$; yet $\operatorname{gap}(T)=\Delta$ while $\operatorname{gap}(T_{\epsilon})=0$ for every $\epsilon>0$. Consequently no criterion built from finitely many --- or from all formal orders of --- short-distance data can decide the mass gap. \end{proposition} \begin{proof} Immediate from the support separation in Proposition~10.7: the modification is $O(r^{-2})$ at short distance and carries a positive spectral weight arbitrarily close to $0$. The bound on separated correlations follows from $\|\epsilon\rho_{\mathrm{IR}}\|=\epsilon$; the gap statement from Lemma~10.1. \end{proof} \begin{proposition}[Dimensional transmutation produces no physical mass in this construction; source-level reading] All registered scale-bearing objects are ultraviolet or conventional: the short-distance theorems (Theorems~9.10, 9.19, 9.22, Proposition~9.14) concern $r\to0$, $a_{N}\to0$; the Ward identity (Proposition~9.4) is a separated-insertion shell law, and Remark~9.5 records that no dilation Ward identity is constructed; the calibrations of Lemma~4.15 are conventions ($a(0)=1$, $\beta(0)=\bar\beta$). Remark~9.11 states that $\Lambda=1$ ``records the construction's reference units; it does not define an intrinsic physical scale'', and Remark~9.20 states that a scheme change only shifts $\log(1/r)$ by a constant. Hence no proved object links the ultraviolet scale to an infrared spectral lower bound: the inference ``trace anomaly $\to$ broken scale invariance $\to$ $\Lambda_{\mathrm{QCD}}$ $\to$ $m\sim\Lambda_{\mathrm{QCD}}$'' lacks its middle term in this construction. A proof would require a full-scale dilation Ward identity / Callan--Symanzik relation, which is absent. \end{proposition} \begin{proposition}[Boundary: what cannot yield a gap] No argument using only ultraviolet data (Theorems~9.10, 9.19, 9.22), rateless clustering (Proposition~7.15 and Appendix~C), and pointwise finite-volume gaps can produce a mass gap. \end{proposition} \begin{proof} The model of Proposition~10.5 satisfies (AN1) and (PH1)--(PH4) with clustering (rateless, by (a) and (b)) and with ultraviolet behavior configurable independently of its infrared weights; every finite-volume truncation of any model has a pointwise gap. If a gap were derivable from such premises, it would hold in every model satisfying them; the counterexample shows otherwise. \end{proof} \subsection{Spectral-gap bridges} \begin{lemma}[Doeblin spectral gap] Let $(X,\mu)$ be a probability space and $P(x,\dd y)$ a Markov kernel with stationary measure $\nu$. If $P(x,\dd y)\ge\varepsilon\,\nu(\dd y)$ for some $\varepsilon>0$ and all $x$, then for every $f$ with $\int f\,\dd\nu=0$, \[ \|Pf\|_{L^{2}(\nu)}\le(1-\varepsilon)\|f\|_{L^{2}(\nu)}. \] If moreover $T\ge0$ is the self-adjoint reflection-positive transfer operator with continuous kernel $c_{1}\le K\le c_{2}$ and Perron root $\lambda_{1}$, then with $R=\log(c_{2}/c_{1})$ the $h$-transform gives a Doeblin constant of order $e^{-2R}$ and \[ \Delta\ \ge\ -a_{t}^{-1}\log\big(1-e^{-2R}\big)\ >\ 0 . \] \end{lemma} \begin{proof} Write $P=\varepsilon\nu+(1-\varepsilon)Q$ with $Q$ a Markov kernel with the same stationary measure $\nu$. For $\int f\,\dd\nu=0$, $Pf=(1-\varepsilon)Qf$. By Jensen $|Qf|^{2}\le Q(f^{2})$, and stationarity gives $\int Q(f^{2})\,\dd\nu=\int f^{2}\,\dd\nu$, hence the $L^{2}$ bound. For the second part, let $Th=\lambda_{1}h$, $h>0$, and set $P(x,\dd y)=K(x,y)h(y)(\lambda_{1}h(x))^{-1}\dd\mu(y)$; then $\int P(x,\dd y)=1$ and $\nu=h^{2}\mu/\int h^{2}\,\dd\mu$ is stationary by symmetry of $K$. From $c_{1}\int h\,\dd\mu\le\lambda_{1}h(x)\le c_{2}\int h\,\dd\mu$ one gets $h(y)/h(x)\ge c_{1}/c_{2}$ and $\lambda_{1}\le c_{2}$, hence $P(x,\dd y)\ge(c_{1}/c_{2})^{2}\dd\mu(y)=e^{-2R}\dd\mu(y)$; since $h$ is bounded above and below, $\dd\mu$ and $\dd\nu$ are equivalent with a normalization-dependent constant. The spectral radius on the zero-mean subspace is then $\le1-\varepsilon$ and $\Delta=-a_{t}^{-1}\log(\lambda_{2}/\lambda_{1})\ge-a_{t}^{-1}\log(1-\varepsilon)$. \end{proof} \begin{remark} The numerical constant depends on the normalization of $h$; the invariant statement is the \emph{form} of the bound: a uniform positive lower bound on the kernel gives a positive gap. The degeneration of this bound with the volume is the content of Proposition~10.16. \end{remark} \begin{proposition}[Change of layer: the gap is scale-invariant] Assume exact blocking, i.e.\ the spatial transfer operator commutes with the one-step transfer matrix $\mathcal{T}_{m}$. Then blocking one step $a^{(m)}\mapsto b_{m}a^{(m)}$ replaces $\mathcal{T}_{m}$ by $\mathcal{T}_{m}^{b_{m}}$ and leaves the physical gap unchanged: \[ \Delta_{m+1} =-\frac{1}{b_{m}a^{(m)}}\log\lambda_{2}\big(\mathcal{T}_{m}^{b_{m}}\big) =-\frac{b_{m}}{b_{m}a^{(m)}}\log\lambda_{2}(\mathcal{T}_{m}) =-\frac{1}{a^{(m)}}\log\lambda_{2}(\mathcal{T}_{m}) =\Delta_{m}. \] Consequently a uniform gap may equivalently be read at the reference layer $a(0)=1$. \end{proposition} \begin{proof} Under exact blocking the coarse-grained transfer matrix is the $b_{m}$-fold time composition of the fine one, so its eigenvalues are $\lambda_{j}(\mathcal{T}_{m})^{b_{m}}$ (the Perron root stays $1$ after normalization). Substituting into $\Delta_{m}=-a^{(m)\,-1}\log\lambda_{2}$ gives the displayed identity. At the reference layer Lemma~4.15(3) gives $\beta_{N,p}(0)=\bar\beta$, a \emph{fixed} constant; the weak-coupling running (4.8) is a statement about the bare layer only. Hence a strong-coupling / Dobrushin-type analysis at the reference layer is not excluded by asymptotic freedom, and the three conditions ``$\lambda<1/3$'', ``$J_{\ast}$ lies in a spectral redundancy direction'', and ``a strong-coupling seed $\Delta_{0}$ exceeds the accumulated Kato losses'' --- which arise only when one perturbs the transfer matrix layer by layer --- are \emph{not necessary} for (MG2). \end{proof} \begin{remark} Actual (RG2) truncations break exact blocking; the resulting remainder is controlled at the schedule level by Lemma~5.9, not by a Kato perturbation series. This removes an apparent conflict between asymptotic freedom and strong-coupling estimates that was noted in earlier drafts of the supplement. \end{remark} \begin{proposition}[The global positivity bound degenerates with the volume] At the reference layer the one-step kernel is $K_{0}=\exp[-L_{E}]$ with $L_{E}\ge0$ the Wilson action on a whole spatial slice. Then the Doeblin lower bound of Lemma~10.12 degenerates as $p\to\infty$. \end{proposition} \begin{proof} Take $U=\mathbf{1}$, so every plaquette term $1-\Reop\Tr(U_{\square})/m$ vanishes and $L_{E}(U)=0$. Take $U_{\ast}$ obtained by setting a fixed fraction of the spatial links to $-\mathbf{1}$ (possible for odd side length, \texttt{Patch.Large.odd}); at least $cL_{p}^{3}$ plaquettes then have $1-\Reop\Tr(U_{\square})/m\ge c'>0$, so $L_{E}(U_{\ast})\ge c_{0}\bar\beta L_{p}^{3}$. Hence \[ R_{0}(p)=\sup L_{E}-\inf L_{E}\ \ge\ c_{0}\bar\beta L_{p}^{3}\to\infty . \] With $K_{0}\in[e^{-R_{0}(p)},1]$, Lemma~10.12 gives $\varepsilon_{p}\lesssim e^{-2R_{0}(p)}\to0$, so the lower bound $\Delta_{M}(p)\approx-a_{t}^{-1}\log(1-\varepsilon_{p})\approx e^{-2c_{0}\bar\beta L_{p}^{3}}$ degenerates. \end{proof} \begin{remark} This is a degeneration of a \emph{bound}, not of the gap. L\"uscher-type finite-volume corrections $\Delta m(L)\sim C L^{-1}e^{-\alpha m_{0}L}$ show that on the massive side the true gap is volume-uniform. The corollary is that (MG2) cannot be proved by a single global Doeblin estimate; a \emph{local} mixing estimate whose rate is independent of $L_{p}$ is required. \end{remark} \begin{proposition}[The kernel family fixes the finite-volume spectral measure] Fix a volume $p$. Suppose two Wilson families satisfying (RG1)--(RG5), (UV1), and (AN1) E2 give the \emph{same} fixed-volume separated-point kernels $S_{p}(v;x)$ for all label lists and all $x\in\Omega_{n}$. Then for every field-word vector $\psi$ the OS scalar spectral measure $\mu_{p}^{\psi}(B)=\langle\psi,E_{H_{p}}(B)\psi\rangle$ is the same, hence $\sigma(H_{p})$ is the same. \end{proposition} \begin{proof} Let $\mathcal{D}_{p}$ be the dense span of positive-time field words (\texttt{registered\_words\_dense}, field-audit step 5). For $\psi\in\mathcal{D}_{p}$, $\psi$ is a finite linear combination of words, so by reflection positivity $\langle\psi,E_{H_{p}}(t)\psi\rangle$ is a finite linear combination of separated-point correlations $S_{p}(A^{\ast}A_{t})$, i.e.\ of pointwise values of the kernels. These are given for \emph{every} $t>0$: (RG5) with an exhausting sequence of compact sets $K_{m}$ and Theorem~5.17's smoothness on $\Omega_{n}$ supply each fixed configuration, however large the time separation. The Laplace transform $\widehat{\mu_{p}^{\psi}}(t)=\int_{[0,\infty)}e^{-t\lambda}\,\dd\mu_{p}^{\psi}(\lambda)$ is therefore determined on $(0,\infty)$; by Lerch's uniqueness theorem for finite measures it determines $\mu_{p}^{\psi}$. Polarization determines all $\mu_{p}^{\psi,\chi}$, and density of $\mathcal{D}_{p}$ determines the projection-valued measure $E_{H_{p}}$, hence the spectrum. \end{proof} \begin{remark} Lerch's theorem needs only \emph{pointwise} values; no $t$-uniform bound is required. What is not fixed by this argument is the \emph{effective readability} of $\mu_{p}((0,\varepsilon))$ for large $t$; that is the remaining gap (Proposition~10.34). \end{remark} \begin{proposition}[Pointwise Laplace convergence suffices; strengthens Proposition~10.32] Assume \[ \textup{(MG2'')}\qquad \exists\,\varepsilon_{0}>0:\ \ \mu_{p}^{\psi}\big((0,\varepsilon_{0})\big)=0 \quad\text{for all }p\text{ and all }\psi\perp\Omega, \] equivalently $\|\mathcal{T}_{N,p}|_{\Omega_{N,p}^{\perp}}\|\le e^{-a_{N}\varepsilon_{0}}$ for all $N,p$. Let $\mu_{\infty}^{\psi}$ be the spectral measure of \emph{any} infinite-volume subsequential limit of Theorem~6.4. Then $\sigma(H)\subset\{0\}\cup[\varepsilon_{0},\infty)$, with the \emph{same} $\varepsilon_{0}$ for every subsequence. \end{proposition} \begin{proof} Fix $\psi$ in the dense field-word domain with $\psi\perp\Omega$. By (MG2''), $\mu_{p}^{\psi}$ is supported on $\{0\}\cup[\varepsilon_{0},\infty)$, and $\rho_{p}(\{0\})=0$ for $\psi\perp\Omega$, so \[ \widehat{\mu_{p}^{\psi}}(t) =\int_{[\varepsilon_{0},\infty)}e^{-t\lambda}\,\dd\mu_{p}^{\psi}(\lambda) \le e^{-\varepsilon_{0}t}\,\|\mu_{p}^{\psi}\| \le C_{\psi}e^{-\varepsilon_{0}t}, \] with $C_{\psi}=\|\psi\|^{2}$ independent of $p$. Theorem~6.4 gives pointwise convergence of separated-point kernels, hence $\widehat{\mu_{p_{k}}^{\psi}}(t)\to\widehat{\mu_{\infty}^{\psi}}(t)$ for every $t>0$. Suppose $\mu_{\infty}^{\psi}((0,\varepsilon_{0}-\delta))>0$ for some $\delta>0$. Then $\widehat{\mu_{\infty}^{\psi}}(t)\ge a\,e^{-(\varepsilon_{0}-\delta)t}$ with $a>0$, and passing to the limit in the displayed bound gives $a\,e^{\delta t}\le C_{\psi}$ for all $t>0$, a contradiction as $t\to\infty$. Hence $\mu_{\infty}^{\psi}((0,\varepsilon_{0}))=0$; the Urysohn/density argument of Proposition~10.3 then gives $\sigma(H)\cap(0,\varepsilon_{0})=\varnothing$. Because $\varepsilon_{0}$ was fixed before the subsequence was chosen, the conclusion is subsequence-independent. \end{proof} \begin{remark} This proof uses \emph{only} pointwise convergence of Laplace transforms and Lemma~10.1. It needs neither strong convergence of transfer matrices, nor tightness of $\{\mu_{p}^{\psi}\}$, nor the restriction of $t$ to integer multiples of $a_{N}$. It therefore supersedes the proof of Proposition~10.32 and adds the statement that the gap does \emph{not} require uniqueness of the infinite-volume limit (O1/O9), only volume-uniformity --- an asymmetry with the repository README, which lists the two open problems side by side. \end{remark} \begin{proposition}[A rate in E4 is equivalent to the mass gap] Under (AN1) E2 and \texttt{registered\_words\_dense}, let $\mathcal{D}$ be the dense field-word domain. The following are equivalent for $\varepsilon_{0}>0$: \begin{enumerate} \item[\upshape(1)] $\sigma(H)\subset\{0\}\cup[\varepsilon_{0},\infty)$; \item[\upshape(2)] there is $C_{\psi,\chi}<\infty$ for all $\psi,\chi\in\mathcal{D}$ with \[ \big|\langle\psi,E(t)\chi\rangle -\langle\psi,\Omega\rangle\langle\Omega,\chi\rangle\big| \ \le\ C_{\psi,\chi}\,e^{-\varepsilon_{0}t}\qquad(t\ge0). \] \end{enumerate} \end{proposition} \begin{proof} (1)$\Rightarrow$(2) is Lemma~10.2. (2)$\Rightarrow$(1): take $\chi=\psi$, put $\psi_{\perp}=\psi-P_{0}\psi$; then $\langle\psi,E(t)\psi\rangle-|\langle\Omega,\psi\rangle|^{2} =\widehat{\rho_{\psi_{\perp}}}(t)\le C_{\psi}e^{-\varepsilon_{0}t}$ and $\rho_{\psi_{\perp}}(\{0\})=0$, so Lemma~10.1 gives $\rho_{\psi_{\perp}}((0,\varepsilon_{0}))=0$. The Urysohn/density argument of Proposition~10.3 upgrades this to $\sigma(H)\cap(0,\varepsilon_{0})=\varnothing$. \end{proof} \begin{remark} Combined with Proposition~7.2, the three properties \emph{clustering / vacuum uniqueness / (PH4)} are the same $o(1)$ statement, and the mass gap is exactly its \emph{uniform-rate} version. Appendix~C's sentence ``No rate is asserted anywhere in the chain'' must therefore be read as: the rate is absent at the \emph{entrance} (\texttt{RemainingObligations.lean} keeps E4 as the single explicit premise of \texttt{completeOS}), not lost inside the six transport modules. The smaller scalar $\Theta$-hierarchy of check A1 gives only a channel rate; the rate in (2) must be common to \emph{all registered label words}, with constants allowed to depend on the word. \end{remark} \begin{proposition}[Every subexponential clustering rate is compatible with no gap] Let $\varphi:[0,\infty)\to(0,\infty)$ be non-increasing, $\varphi(t)\to0$, satisfy the doubling condition $\varphi(t/4)\le C_{4}\varphi(t)$ and a polynomial lower bound $\varphi(t)\ge t^{-A}$ for some $A>0$. Then there is a finite positive measure $\tilde\rho$ on $(0,1]$ with no atom at $0$, accumulating at $0$, such that $G(\tau)=\int e^{-\tau\lambda}\,\dd\tilde\rho(\lambda)\le C\varphi(\tau)$ for all large $\tau$, while $\tilde\rho((0,\varepsilon))>0$ for every $\varepsilon>0$ --- hence, by Lemma~10.1, no spectral gap. \end{proposition} \begin{proof} Put $\lambda_{k}=2^{-k}$ and $w_{k}=\varphi(2^{k})-\varphi(2^{k+1})\ge0$. Since $\varphi\to0$, telescoping gives $\sum_{k\ge1}w_{k}\le\varphi(2)<\infty$ and infinitely many $w_{k}>0$, so $\tilde\rho=\sum_{k}w_{k}\delta_{2^{-k}}$ is finite, has no atom at $0$, and accumulates at $0$. For $\tau\ge2$ set $k_{0}=\lfloor\log_{2}\tau\rfloor$, so $2^{k_{0}}\le\tau<2^{k_{0}+1}$ and $1/\tau<2^{-k_{0}}\le2/\tau$. \emph{Tail} ($k\ge k_{0}$): $\sum_{k\ge k_{0}}w_{k}e^{-\tau2^{-k}}\le\sum_{k\ge k_{0}}w_{k} =\varphi(2^{k_{0}})\le\varphi(\tau/2)$. \emph{Head} ($k0$ for all $\varepsilon>0$, Lemma~10.1 (contrapositive) gives no gap. \end{proof} \begin{remark} The class covers $\tau^{-p}$, $\log^{-q}\tau$, $\exp(-(\log\tau)^{\beta})$ and the $1/m$-type rates supplied by the schedule estimates (\texttt{Fits.stepSlow} gives $J(m)\le J_{0}+m$, hence $J(m)^{-1}\gtrsim m^{-1}$). Propositions~10.5, 10.7 and the tower model of Proposition~10.28 are instances. Consequently ``each channel clusters exponentially'' and ``clustering holds with every subexponential rate'' are both \emph{insufficient}: only a common exponential rate $\varepsilon_{0}>0$ produces a gap. This also corrects the claim in (O6) that excluding continuous-mass generalized free fields is \emph{equivalent} to an (MG3)-type estimate: the implication goes one way only. \end{remark} \begin{proposition}[Ward identities do not supply degree-uniformity] Assume (Y2). Then: \begin{enumerate} \item[\upshape(1)] if $v=D_{b}v'$, the spectral support of $v$ is contained in that of $v'$; in particular the spectral threshold cannot decrease along a $D_{b}$-tower; \item[\upshape(2)] $D_{b}$ carries a free vector index $b$, so it raises tensor rank by one; \emph{no scalar label lies in the image of any} $D_{b}$; \item[\upshape(3)] consequently (Y2) relates spectral thresholds only \emph{within} each tower of iterated derivatives; the thresholds of the tower bases are free parameters of (Y2), and no relation between distinct towers follows. \end{enumerate} \end{proposition} \begin{proof} (1) The calculus law $S(D_{b}v_{0},v_{1},\dots)=\partial_{x_{0}}^{b}S(v_{0},v_{1},\dots)$ multiplies the Wightman spectral measure by the momentum weight $|p_{b}|^{2}$ (a non-negative factor); a non-negative weight can only shrink the support. (2) In the flux law the index $a$ is summed while $b$ is free, so $D_{b}v_{0}$ has index set $\{b\}\cup\operatorname{idx}(v_{0})$; a rank-zero scalar cannot arise. (3) By Section~9.5, $\mathcal{F}_{d}=\mathcal{F}_{d-1}\oplus E_{d}$ with $n_{d}=\dim E_{d}$ \emph{fresh} fields per layer; (1)--(2) leave each $E_{d}$'s base unconstrained, and Remark~9.5 states that no dilation Ward identity is constructed. \end{proof} \begin{remark} This retracts an earlier claim that (Y2) gives the degree-uniformity needed for (MG3$^{\prime}$) for free. The Ward identities do \emph{not} provide a common rate across towers. The remedy is not to prove a degree-uniform estimate tower by tower, but to use the \emph{full-spectrum} form (MG2''), which bypasses all channel/tower covering questions (Proposition~10.20). \end{remark} \begin{proposition}[Tower model; calculus-law part; partially constructed] Let the label filtration decompose into derivative towers indexed by $t\in\N$, each with a base label. There is a tempered Gaussian hierarchy satisfying (AN1), (B1), (Y1), the \emph{calculus law} of (Y2), the ultraviolet law of Theorem~9.10 and the OPE coefficients of Theorem~9.22, in which every tower clusters exponentially with rate $m_{t}=2^{-t}>0$, yet $\inf_{t}m_{t}=0$ and there is no mass gap. \end{proposition} \begin{proof} Take a block-diagonal KL weight: tower $t$ carries $\rho_{t}\ge0$ supported in $[m_{t},\infty)$, has no atom at $0$, and all towers share the \emph{same} large-$E$ ultraviolet tail of Proposition~10.7. Within a tower, a derivative $P(D)w_{t}$ multiplies $\rho_{t}$ by a polynomial momentum weight, so its support stays in $[m_{t},\infty)$ and the calculus law holds by construction. Block diagonality preserves the label null spaces, hence (B1); the shared ultraviolet tail preserves the short-distance constant and the OPE coefficients; $\rho\ge0$ gives E2, and no $0$-atom gives E4 together with vacuum uniqueness (Proposition~7.2). Each tower clusters at rate $m_{t}$, but $\rho((0,\varepsilon))>0$ for every $\varepsilon>0$, so Lemma~10.1 gives no gap. \end{proof} \begin{remark}[Open identification point: the flux law in tower models] In this tower model the \emph{flux (shell) law} is not checked: $T_{ab}$ is a single rank-two label, whereas Wick contractions with a base label from a different tower vanish, so the left-hand side of the flux law and $\partial_{x_{0}}^{b}S(v_{0},\dots)$ need not match. By contrast, the \emph{uniform} superposition of Proposition~10.7 satisfies both laws because all insertions share one two-point factor. Accordingly Proposition~10.28 is a \emph{calculus-law} model, and the statement ``each channel clusters exponentially yet there is no gap'' is established under the calculus law alone. Whether the flux law admits a tower-wise construction is left open; the degree-uniformity conclusion of Proposition~10.26 does not depend on it. \end{remark} \begin{proposition}[No gap-shaped predicate in the source signatures; source-level reading] The records read directly from the submitted sources (\texttt{YM.Overlay.Schedule.Priced} / \texttt{Tabs} / \texttt{consumes}, \texttt{YM.RG.History.Fits}, \texttt{YM.RG.Calendar}, \texttt{YM.Bid.Fuels.Circle}, \texttt{YM.Next.MULTISOURCE.Mixed.mixed\_circle}, \texttt{YM.RG.ActivityBounds}, \texttt{YM.Source.Insertion.TransferData.Based}) contain no predicate about a large-time translation $s$, a spectral projection $E_{H}((0,\varepsilon))$, or a second transfer-matrix eigenvalue $\lambda_{2}$. Specifically: \begin{itemize} \item \texttt{Tabs} is produced by \texttt{Stock.consumes} from \texttt{stepsize} (a polylog doubling condition) and \texttt{afford} (a $z^{17}=o(s)$ cost bound) together with $\mathrm{Tendsto}\ (J(i+1)^{-1})\to0$; \item \texttt{Fits.deep} asserts $J\to\infty$ with \emph{no rate}, \texttt{Fits.stepSlow} gives $J(m+1)\le J(m)+1$, and \texttt{Calendar.hatBeta} $=\bar\beta+\mathrm{g}\cdot\mathrm{scale}(m)$, \texttt{coverSize}, \texttt{monitor}, \texttt{forcedAllowance} are logarithmic/polylogarithmic functions of $\hat\beta$; \item \texttt{Mix.Circle\ C\ d\ u} is an existence statement about \texttt{Story}/\texttt{Lane}/\texttt{Stubs.Holds}/\texttt{Integrable} witnesses with \texttt{width\ C\ r\ k\ i}-type envelopes; \texttt{Bid.installs} consumes it to produce \texttt{Input.Metered\ C}, a \emph{field-content/ultraviolet metering} record; \texttt{mixed\_circle} derives \texttt{Circle} from \texttt{Responsive}, \texttt{Margins}, \texttt{Addresses}, \texttt{Tabs}; \item \texttt{Activities.KBound}/\texttt{EBound} are summability caps for polymer gas/regular occurrences; \texttt{TransferData.Based.front/back} are coefficient-norm bounds with factors $\mathrm{columns}\cdot b_{m}^{d}$ (Remark~5.21: no uniform inverse bound). \end{itemize} By the soundness observation of Proposition~10.5(d), no statement whose predicates are absent from these signatures can be derived from them. Hence (MG) is \emph{not a receipt waiting to be discharged inside the formalization}: it is a new estimate external to the sources. \end{proposition} \begin{remark} This is a structural signature check, \emph{not} a formal independence theorem: the six \texttt{Cluster*} modules were not read verbatim in this study (Appendix~C's module-by-module description was used). The README's actual wording --- \emph{``a positive mass gap \ldots\ outside the scope of the public challenge''} --- together with the reached closure of 100{,}168 declarations, is consistent with this reading. \end{remark} \subsection{Conditional routes to a mass gap} We record four candidate infrared inputs. Each is a proposal; none is proved anywhere in this paper or in the repositories. \begin{description} \item[(MG1) Renormalized cluster expansion.] At each blocking level $m$ a string tension/mass lower bound $\sigma_{m}>0$, with summable cross-level losses, such that $\inf_{m}\sigma_{m}\ge\sigma_{\ast}>0$ in physical units as $a_{N}\to0$. (Strong coupling seeds exist [Wilson, OS78, Seiler]; the bare couplings here run to weak coupling, (4.8), so a no-phase-transition bridge is additionally needed.) \item[(MG2) Transfer-matrix gap, uniformly.] There is $\Delta>0$ with $\|\mathcal{T}_{N,p}|_{\Omega_{N,p}^{\perp}}\|\le e^{-a_{N}\Delta}$ for all $N,p$, preserved under blocking. \item[(MG3$^{\prime}$) Infrared exponential clustering, uniform rate.] There exists $\varepsilon_{0}>0$ such that for every finite label word $A$ there is $C_{A}<\infty$ with $\sup_{p}\big|S_{p}(A^{\ast}A_{s})-|S_{p}(A)|^{2}\big|\le C_{A}e^{-\varepsilon_{0}s}$ ($s\ge0$); the constants may depend on $A$, the rate $\varepsilon_{0}$ may not (Proposition~10.26 shows the Ward identities do not supply this uniformity). \item[(MG4) Wilson loops with area law.] Non-local Wilson loop observables (outside the local dictionary of Proposition~2.12) together with a uniform area-law bound $\langle W_{C}\rangle\le C\,e^{-\sigma\,\mathrm{area}(C)}$, $\sigma>0$. \end{description} \begin{proposition}[(MG2) implies a continuum gap] Under (MG2) and the limit hierarchy of Theorem~6.4, $\sigma(H)\subset\{0\}\cup[\Delta,\infty)$. \end{proposition} \begin{proof} On the finite lattice define the semigroup $E_{N,p}(t)=\mathcal{T}_{N,p}^{\,t/a_{N}}$ (with $t$ an integer multiple of $a_{N}$). By (MG2), for $\psi\perp\Omega_{N,p}$, $\|E_{N,p}(t)\psi\|\le e^{-\Delta t}\|\psi\|$. The uniform kernel convergence of Theorems~5.17 and 6.4 gives strong convergence $E_{N,p}(t)\to E(t)$ on the dense domain generated by field words (the restriction of $t$ to integer multiples of $a_{N}$ is removed by strong continuity of the limiting semigroup). Hence for $\psi$ in that dense domain, $\psi\perp\Omega$, \[ \langle\psi,E(t)\psi\rangle =\lim\langle\psi_{N,p},E_{N,p}(t)\psi_{N,p}\rangle \le e^{-\Delta t}\|\psi\|^{2}. \] Lemma~10.1 gives $\rho_{\psi}((0,\Delta))=0$, and the density argument of Proposition~10.3 gives $\sigma(H)\cap(0,\Delta)=\varnothing$. \end{proof} \begin{remark}[Superseded proof] A stronger and shorter proof is given in Proposition~10.20: it uses only pointwise convergence of Laplace transforms, yields the same $\Delta$ for every infinite-volume subsequence, and shows the gap does not require uniqueness of the infinite-volume limit (O1/O9), only volume-uniformity. \end{remark} \begin{center} \small \begin{tabular}{@{}p{0.28\textwidth}p{0.34\textwidth}p{0.30\textwidth}@{}} \toprule Goal & Minimal additional input & Consequence \\ \midrule Full gap $\Delta>0$ & (MG) $=$ (MG2), or (MG3$^{\prime}$) $+$ full channel coverage & Definition~1.2; subsequence-independent (Proposition~10.20) \\ Channel spectral threshold & (MG3$^{\prime}$) in one channel & $\rho_{\psi}((0,m))=0$ for that channel only \\ Continuum gap from the lattice & (MG1)/(MG2) & Proposition~10.20 \\ Exclusion of gapless models & O6 below & necessary-not-sufficient (Proposition~10.24) \\ \bottomrule \end{tabular} \end{center} \begin{proposition}[The single remaining predicate] The four candidate inputs (MG1)--(MG4) and the volume-uniform transfer-matrix condition (MG2) are equivalent projections of one statement. Register \begin{quote} \textbf{(MG)} There exists $\varepsilon_{0}>0$ such that for every finite label word $A$ there is $C_{A}<\infty$ with \[ \sup_{p}\ \big|S_{p}(A^{\ast}A_{s})-|S_{p}(A)|^{2}\big| \ \le\ C_{A}\,e^{-\varepsilon_{0}s}\qquad(s\ge0), \] where $C_{A}$ may depend on $A$ but the rate $\varepsilon_{0}$ may not, and the supremum is over all volumes. \end{quote} Then: \begin{enumerate} \item[\upshape(1)] (MG) $\Rightarrow$ (MG2'') $\Rightarrow$ (by Proposition~10.20) $\sigma(H)\subset\{0\}\cup[\varepsilon_{0},\infty)$ for \emph{every} infinite-volume subsequential limit; \item[\upshape(2)] (MG) restricted to all registered label words \emph{implies} the mass gap of the limit theory, and the two rates coincide (Propositions~10.20, 10.22); the converse implication is open (Remark~10.42(6)); \item[\upshape(3)] (MG1), (MG2), (MG3$^{\prime}$) and (MG4) are respectively the cluster-expansion, spectral, rate, and Wilson-loop projections of the same uniform infrared lower bound; (MG4) and single-channel versions give only a \emph{channel} threshold (Remark~10.4, Proposition~10.26); \item[\upshape(4)] dropping either uniformity --- over volumes $p$, or over label words with a common rate --- invalidates the conclusion (Propositions~10.16, 10.24, 10.28). \end{enumerate} \end{proposition} \begin{proof} (1) Apply Lemma~10.1 to the diagonal matrix element of each field-word vector and then Proposition~10.3's Urysohn argument to obtain $\mu_{p}^{\psi}((0,\varepsilon_{0}))=0$ uniformly in $p$; Proposition~10.20 finishes. (2) is Proposition~10.22. (3) is the content of Propositions~10.12--10.20 and Remarks~10.4/10.27. (4) is Propositions~10.16, 10.24 and 10.28. \end{proof} \begin{center} \small \begin{tabular}{@{}p{0.22\textwidth}p{0.72\textwidth}@{}} \toprule Form & Rate statement \\ \midrule (MG2$^{\prime\prime}$) & $\|\mathcal{T}_{N,p}|_{\Omega^{\perp}}\|\le e^{-a_{N}\varepsilon_{0}}$, all $N,p$. \\ (MG3$^{\prime}$) & $\sup_{p}|S_{p}(A^{\ast}A_{s})-|S_{p}(A)|^{2}|\le C_{A}e^{-\varepsilon_{0}s}$, all words $A$. \\ (MG) & as (MG3$^{\prime}$), the common rate over \emph{all} registered label words. \\ Gap & $\sigma(H)\subset\{0\}\cup[\varepsilon_{0},\infty)$, same $\varepsilon_{0}$, every subsequence. \\ \bottomrule \end{tabular} \end{center} \begin{remark}[Final boundary] Everything else in Section~10 is now either proved or reduced to (MG): O6 is a necessary-not-sufficient sanity check (Proposition~10.24), O1/O9 are not needed for the gap (Proposition~10.20), O5 requires in addition an atom in the spectral measure (Proposition~10.38), O8 gates O2 (Lemmas~10.36--10.37), and O7/O10 are ultraviolet/perturbative questions independent of the spectral gap. \textbf{A positive mass gap is not proved.} \end{remark} \subsection{Sharpened statements of the remaining open problems} \begin{description} \item[(O1) Convergence and uniqueness of the infinite-volume limit.] Missing input: a volume-direction estimate --- e.g.\ (RG2) upgraded from volume-uniform bounds to volume-uniform difference estimates --- or a uniqueness statement for the RG fixed point. \emph{Reduction proved here:} if the limits obtained from any two volume subsequences agree on the countable index set $\mathcal{L}\times\{K_{m}\}\times\{\alpha\}$ of Theorem~6.4, then they are equal as kernels on $\Omega_{n}$ (uniform continuity on compacts plus exhaustion), and hence define the same functionals. \item[(O2) A continuum probability measure.] Missing input: positivity on the \emph{unrestricted} joint Schwartz test space, uniqueness of contact extensions (O8), and Minlos-type growth and compatibility conditions. \item[(O4) Wilson loops and the area law.] Missing input: incorporation of non-local line observables into the hierarchy; the local dictionary of Proposition~2.12 covers only $\bigcup_{d}\mathcal{F}_{d}$. Note the area law would give at most a channel threshold, not the full gap (Proposition~10.3 and Remark~10.4). \item[(O5) Single-particle states.] A gap does not suffice: Proposition~10.38 below shows a reflection-positive generalized free field with spectral gap $m$ whose K\"allen--Lehmann measure has no atoms, hence no isolated one-particle shell and no single-particle states [GJ]. \item[(O6) General exclusion of continuous-mass generalized free fields.] This is a necessary, not sufficient, prerequisite for the mass gap (Remark~10.6); (MG3$^{\prime}$) implies it, but excluding free-field descriptions does not by itself supply the uniform rate, since interacting gapless models are not generalized free fields (Proposition~10.24). It should be read as a sanity check on (MG3$^{\prime}$), not as an equivalent formulation. \item[(O7) Order-by-order OPE/perturbation matching.] Missing input: identification of the algebraic prescription coefficients $c_{\alpha}$ of Theorem~9.22 with perturbative coefficients order by order, including operator-valued renormalization conditions and state-dependent parts. \item[(O8) Uniqueness and positivity of contact extensions.] Two proved obstructions: diagonal-flat tests annihilate diagonal-supported distributions (Lemma~10.36), so the check A1 comparison cannot decide uniqueness; and extensions are never unique without further requirements (Lemma~10.37). Missing input: fixation of the diagonal part by reflection positivity on the unrestricted test space, plus lattice convergence of the extended functionals. \item[(O9) Subsequence independence and the mixed vacuum.] Missing input: the same volume-direction control as (O1), plus clustering of the mixed theory (which would collapse all phases by Lemma~6.6). \item[(O10) Infrared flow and Euclidean invariance, part (iii).] The running-coupling argument is perturbative in the marginal coupling and does not address the infrared side; the third part of Euclidean invariance (stress-energy flux decay) is recorded as a construction input in Section~8.4. An independent expansion of the latter would reduce the reliance on construction inputs without changing the conclusions. \end{description} \begin{lemma}[Diagonal-flat tests annihilate diagonal-supported distributions] Let $T\in\mathcal{D}'\big((\R^{4})^{n}\big)$ be supported on a coincidence diagonal $\Delta_{ij}$, and let $f\in\mathcal{S}_{0}\big((\R^{4})^{n}\big)$ (all derivatives vanish on $\Delta_{ij}$). Then $\langle T,f\rangle=0$. \end{lemma} \begin{proof} By the local structure theorem for distributions supported on a closed submanifold, $T=\sum_{|\beta|\le N}\partial^{\beta}T_{\beta}$ with $T_{\beta}$ measure-type distributions on $\Delta_{ij}$ (normal $\delta$ functions and their derivatives). Hence $\langle T,f\rangle=\sum_{\beta}(-1)^{|\beta|}\int_{\Delta_{ij}}\partial^{\beta}f\, \dd T_{\beta}=0$, since all derivatives of $f$ vanish on $\Delta_{ij}$. \end{proof} \begin{lemma}[Contact extensions are never unique] Let $T_{0}$ be a distribution on the full configuration space satisfying the \texttt{Contacts} predicate of Remark~9.24, and let $T_{\gamma}\neq0$ be any distribution supported on the coincidence set $\Delta=\bigcup_{i0$ (so the gap is $\Delta=m$) and whose two-point function decays on the exponential scale $e^{-m\tau}$ (up to powers), but whose K\"allen--Lehmann measure $\dd\rho(\mu^{2})=\mathbf{1}_{[m^{2},\infty)}\,r(\mu^{2})\,\dd\mu^{2}$, with $r\ge0$ locally integrable, polynomially growing, and atomless, has no isolated one-particle shell; consequently the theory has no single-particle states [GJ]. \end{proposition} \begin{proof} Take any such $r$ (e.g.\ $r\equiv1$ on $[m^{2},\infty)$); the corresponding measure $\rho$ is positive with $\supp\rho\subset[m^{2},\infty)$, so the Hamiltonian of the associated generalized free field satisfies $\sigma(H)\subset\{0\}\cup[m,\infty)$, and Lemma~10.1 gives exponential decay at rate $m$. Since $\rho$ has no atoms, there is no mass shell $\delta(\mu^{2}-M^{2})$ in the K\"allen--Lehmann representation and hence no one-particle subspace. \end{proof} \begin{remark}[Dependency summary] The structural edges are: O3 $\Leftrightarrow$ E4-with-a-rate (Proposition~10.22); O6 $\rightarrow$ O3 (necessary-not-sufficient, Proposition~10.24); O8 $\rightarrow$ O2. The bridges of Lemmas~10.1--10.2 and Propositions~10.3, 10.18, 10.20 reduce the mass gap to the single volume-uniform infrared estimate (MG) (Proposition~10.34); uniqueness of the infinite-volume limit (O1/O9) is \emph{not} needed (Proposition~10.20). Everything else in this section is a precise statement of what is missing. \end{remark} \begin{remark}[Retractions from the exploratory study] The following intermediate claims were superseded during the source-level analysis; they must not be carried into citations of this paper. \begin{center} \small \begin{tabular}{@{}p{0.05\textwidth}p{0.55\textwidth}p{0.34\textwidth}@{}} \toprule \# & Retracted claim & Replaced by \\ \midrule R1 & (PH4) is an independent seventeenth input & Proposition~7.2; sixteen inputs \\ R2 & (MG2) requires a new infrared \emph{predicate} absent from (RG1)--(RG5) & Proposition~10.18: the kernel family already fixes $\mu_{p}$; what is missing is an \emph{estimate} on it (Proposition~10.30) \\ R3 & (Y2) gives degree-uniformity for free & Proposition~10.26 \\ R4 & A filtration-layer model with per-layer thresholds violates no Ward law; a per-tower model satisfies all of (Y2) & Proposition~10.28: the layer model violates the calculus law; the tower model satisfies the calculus law only, with the flux law left open \\ R5 & \texttt{Mix.Circle} is the source-level registration of (MG2) & Proposition~10.30: \texttt{Circle} $\to$ \texttt{Input.Metered} is a UV/field-content chain \\ R6 & The left half of the feasibility condition $K/(2b_{0}\log b_{\min})<\bar\beta$ is an open obstacle & Lemma~5.9 (\texttt{Rolls.exists\_rolls} + weighted Ces\`aro) closes it; the right half \emph{is} (MG) \\ R7 & \texttt{registered\_words\_dense} might cover only the scalar $\Theta$ channel; the E4 chain includes \texttt{ArityGrowth} & The full field basis is used (Remark~10.4); the six E4 modules are \texttt{ClusterKernel}, \texttt{ClusterCore}, \texttt{ClusterCones}, \texttt{ClusterDomination}, \texttt{ClusterTarget}, \texttt{ClusterSchwartz}; \texttt{ArityGrowth} belongs to E0$'$ \\ R8 & The counterexample is Gaussian; a non-Gaussian interacting upgrade was claimed & The Gaussian model suffices (Proposition~10.7); a mixture attempt loses (PH4) or is conditionally Gaussian (Remark~10.41) \\ R9 & A pointwise gap at each volume suffices for the rate & The family of Proposition~E.2 has pointwise gap $m_{p}=1/(2p+1)$ but no uniform one; (MG) with $\sup_{p}$ is not ruled out by it \\ R10 & ``Not a receipt dischargeable inside the formalization'' & Tightened: reached closure of 100{,}168 declarations; no gap-shaped predicate in the signatures (Proposition~10.30) \\ R11 & The autocorrelation criterion applies to non-Gaussian mixing & The Markov-chain reduction is unverified beyond the Gaussian case (Theorem~E.5 is proved for Gaussian models) \\ R12 & The Lean sources rule out the E.2 family & Not read; the family is consistent with the reached closure (Remark~10.31) \\ \bottomrule \end{tabular} \end{center} \end{remark} \begin{remark}[Why the counterexample need not be non-Gaussian] Two attempts to upgrade Proposition~10.5 to a genuinely interacting model give a dichotomy: a \emph{mixture} $\int T_{\lambda}\,\dd\nu(\lambda)$ of Gaussian hierarchies with a non-atomic $\nu$ is non-Gaussian and preserves (B1)/(Y1)/(Y2)/UV data, because (B1) is a rank comparison and the Ward/OPE prescriptions are linear in correlations; but its Hilbert space is a direct integral with a \emph{trivially acting} base, so $\ker H\supseteq L^{2}(\nu)$ and vacuum uniqueness is lost. Making the base \emph{ergodic} under the spatial translation action restores vacuum uniqueness by covariance, but the model remains \emph{conditionally Gaussian}. Since Proposition~10.7 already provides a gapless model satisfying the registered ultraviolet, Ward and field-content conditions, the question of a genuinely interacting gapless model is a question of \emph{model classification}, not of the decidability of the gap. It does not affect (MG). \end{remark} \begin{remark}[Open identification points (not theorems)] \begin{enumerate} \item \emph{Drift identity.} Lemma~5.9 identifies the source's \texttt{drift} (\texttt{Intent.Large.slow/kick}, \texttt{Bench.recover}'s step error) with the one-step remainder $\varepsilon_{m}$. The identification is supported by the shared shape $|\beta_{m+1}-\beta_{m}+\mathrm{g}\log L|\le\mathrm{drift}$, but the definitions of \texttt{Survey.melt}, \texttt{stage}, \texttt{profile} were not read; Lemma~5.9 is a theorem \emph{about the schedule-controlled remainder}, and its application to the submitted recursion is a source-level reading. \item \emph{\texttt{Fits} discharge.} Whether \texttt{Nonperturbative.existence} discharges \texttt{Fits} or it enters through \texttt{WilsonRegularity} was not read. \texttt{Rolls.exists\_rolls} is a closed theorem, so the downstream argument does not depend on this. \item \emph{Tightness of $\{\mu_{p}^{\psi}\}$ at $t\downarrow0$.} Not used in Proposition~10.20, but still needed for a full probability-measure construction (O2). \item \emph{Spectral type of $H_{p}$.} At fixed $(N,p)$ the transfer kernel is continuous on a compact group product, hence Hilbert--Schmidt, so $\mathcal{T}_{N,p}$ is compact and $\Delta_{N,p}>0$. After the UV limit, compact spatial support alone does \emph{not} imply a compact resolvent: the model of Proposition~10.5 has $0\in\sigma_{\mathrm{ess}}(H_{p})$. By Proposition~10.18 the spectral type is fixed by the kernel family; the open question is the positivity of $\inf_{p}\Delta_{p}$, i.e.\ (MG2''). \item \emph{Six \texttt{Cluster*} modules.} Their verbatim sources were not read; Propositions~10.22/10.30 use Appendix~C's module-by-module description. \item \emph{Reverse implication.} (MG) $\Rightarrow$ gap is proved; gap $\Rightarrow$ (MG) is not. L\"uscher-type finite-volume analysis suggests the reverse in known massive model classes, but no general proof is given here. \item \emph{Flux law in tower models.} See Remark~10.29. \end{enumerate} \end{remark} \section{How the formal proof was checked} The committee distinguishes three activities. \emph{Formal verification} checks that a proof establishes a stated mathematical claim from its definitions and permitted axioms. In Lean, proofs are represented by terms that a small kernel checks against the inference rules; automated tactics must produce terms passing this check. \emph{Validation} asks the further question whether the formal claim correctly expresses the intended Yang--Mills construction. This is addressed through review of the definitions and through independent adapters---Lean proofs that the constructed objects satisfy requirements from previous formalizations. The fields and correlations were traced from the specified Wilson lattice theory through the ultraviolet and large-volume limits, the phase decomposition, and the relativistic reconstruction. Lean's flexibility also permits unwanted changes: a submitted development can alter the problem statement or an underlying definition during verification and make an invalid proof appear to verify. For this reason a separate comparison against a fixed reference specification is needed; this is performed by the tool \emph{Comparator} [Comparator], which checks a submitted solution against a separately frozen reference statement, compares target statements and their underlying definitions, audits the proof's axiom dependencies, and rechecks the exported proof with a Lean kernel. The precise targets and verification configurations are recorded in Section~14. After the original submission, OpenAI informed the group that its original public Comparator target \texttt{YangMills.\allowbreak \allowbreak \allowbreak jointExistence} omitted some of the connections to the lattice construction established by its proof. This weakness does not undermine the review: the combined target \texttt{YMReview.\allowbreak \allowbreak \allowbreak Acceptance.\allowbreak \allowbreak \allowbreak combined\_acceptance} requires those connections by retaining the native construction certificates for the lattice observables, limits, phases, and reconstructed fields. The assessment of these connections therefore rests on the stronger target and the checks below, rather than on the public existence statement alone. \subsection{Clean rebuild, inventory, and kernel check} The recorded clean build used Lean 4.32.0 on Apple Silicon, with the pinned official Mathlib dependency cache and no pre-existing project build artifacts. It rebuilt 9,470 project modules, completed in about 184 minutes, and reported zero errors. A subsequent incremental confirmation also passed. All 9,477 supplied Lean source files matched the submission manifest. This is record M1. The inventory inspected all 250,319 project declarations imported by the construction, including generated declarations. It found no project axiom declarations, no direct references to Lean's placeholder axiom \texttt{sorryAx}, and no declarations marked unsafe. The 302 generated declarations are partial recursion helpers; none occurs among the dependencies of the audited principal results. The sole source-level \texttt{sorry} belongs to the reference statement in \texttt{Challenge.\allowbreak \allowbreak \allowbreak lean}, which is loaded separately from the solution; the solution supplies a proof without importing that file. The submission's verifier \texttt{VerifyJoint.\allowbreak \allowbreak \allowbreak lean} compared the target statement and its reachable shared definitions in the reference and solution environments. It then rechecked the 100,168 project declarations on which the 17 principal results depend, using a fresh kernel environment initialized with the pinned external libraries, and audited 14,364 external declarations referenced at the boundary of those project dependencies. The principal proofs use only the permitted foundational axioms \texttt{propext}, \texttt{Classical.\allowbreak \allowbreak \allowbreak choice}, and \texttt{Quot.\allowbreak \allowbreak \allowbreak sound}. The inventory covers all imported project declarations; the kernel check covers the dependencies of the 17 principal results. This submitted verifier is separate from the upstream Comparator used for the three runs reported in Section~11.2. The clean build emitted 2,453 warnings in 1,063 files, most concerning proof-script cleanup or deprecated syntax; the audited solution proofs have no dependency on the placeholder warning. \begin{remark}[Code quality concerns] Of 250,319 project declarations (including auto-generated auxiliary declarations), only 100,168 (including auxiliary declarations) are used in the dependency tree of the main theorems; the difference is less drastic for source declarations but still significant. The submission also uses unconventional naming, e.g.\ \texttt{Farewell.\allowbreak \allowbreak \allowbreak dead} in \texttt{TicketsChart.\allowbreak \allowbreak \allowbreak lean} and \texttt{Cake} and \texttt{cookie} in \texttt{Kitchen.\allowbreak \allowbreak \allowbreak lean}. These are recorded as concerns; they do not affect the kernel acceptance of the proofs. \end{remark} \subsection{Comparator results} Comparator passed on October 30, 2026 for the submitted public theorem \texttt{YangMills.\allowbreak \allowbreak \allowbreak jointExistence}, and on November 3, 2026 for the independent combined theorem \texttt{YMReview.\allowbreak \allowbreak \allowbreak Acceptance.\allowbreak \allowbreak \allowbreak combined\_acceptance}. The November 3 verification record preserves the checked target's identity, configuration, source-file hashes, and logs; the corresponding Lean files in the release are identified by the review source manifest. The run passed statement and definition comparison, the permitted-axiom check, and the built-in Lean kernel. Four controls behaved as expected: a valid proof was accepted, while a changed statement, a forbidden axiom, and a changed definition were rejected. These runs are record M2. The combined run rebuilt its 40 required review modules in a new project and reused the previously built submission artifacts and dependency cache. The separate clean rebuild of the submission is documented in Section~11.1. All three runs used pinned, unmodified upstream Comparator and exporter revisions and Comparator's built-in Lean kernel. The project and exporter used Lean 4.32.0; the Comparator kernel used Lean 4.34.0. On macOS the runs used the upstream development launcher \texttt{fake-landrun.\allowbreak \allowbreak \allowbreak sh}, which runs the checks without Linux Landrun's process isolation; the records therefore document statement, definition, axiom, and kernel checks, but do not test isolation of potentially hostile build code. An additional external kernel implementation was not used. Each verification record identifies the exact target, inputs, and configuration. A further Comparator run on November 5, 2026 passed for the OPE statement-alignment theorem described in Section~13.5. These are two distinct records: the submission's own \texttt{VerifyJoint.lean} kernel replay (record M1) and the committee's three upstream Comparator runs reported here; the latter used \texttt{fake-landrun.sh} on macOS and therefore document statement, definition, axiom and kernel checks, but not process isolation. \subsection{Additional review after the kernel check} Lean kernel acceptance determines whether the proof follows from its formal definitions and permitted axioms; it does not settle whether those definitions express the intended natural language problem. For this reason the group carried out extensive (but not exhaustive) human review of the Lean definitions, checking them for alignment with the problem statement and with expectations based on prior discussion; the glossary of Appendix~A documents these definitions. The definitions were also compared with previous formalizations. In particular, the group proved that the Schwinger functions of the centered stress trace in the selected phase satisfy the OS axioms and the $\mathrm{E}0'$ growth condition in the formulation of [DMY]; Section~12 describes the comparison of test spaces and axiom formulations. The scalar OS, lattice, field-content, translation Ward, leading trace, and common-scale checks are assembled in \texttt{YMReview.\allowbreak \allowbreak \allowbreak Acceptance.\allowbreak \allowbreak \allowbreak combined\_acceptance}, which compiled and passed a Comparator check. This theorem retains the stronger construction data supplied by \texttt{YangMills.\allowbreak \allowbreak \allowbreak constructed}, as well as the public existence conclusions. The retained data connect all the additional properties to the same construction. The OPE statement-alignment check of Section~13.5 is a separate endpoint with its own Comparator run. \subsection{Identifying the constructed theory as Yang--Mills} The combined target retains the construction certificates \texttt{C.\allowbreak \allowbreak \allowbreak Makes} and \texttt{Spec.\allowbreak \allowbreak \allowbreak Final}, connecting the Wilson lattice fields to the continuum correlations, phase decomposition, and reconstructed QFT described in Sections~4--8. These data fix the lattice fields' classical curvature-polynomial meaning and renormalization, identify continuum field words across both limits, and connect the selected phase's boundary distributions to reconstructed vacuum expectations. For each group and positive construction parameter, one Wilson family is fixed before the extraction order is chosen; one refinement, correlation table, and final witness then satisfy all the combined checks. In particular, the Ward identity, the scalar OS axioms, and the leading trace law concern the same stress tensor and the same selected phase. Together with the field-content and short-distance results, these connections support identification with Yang--Mills theory. Universality, infrared uniqueness, and full perturbative agreement remain open (Section~10). \begin{remark}[Worked construction-chain audit] The commented \texttt{OctoberChain.\allowbreak \allowbreak \allowbreak \allowbreak lean} proves seven consequences of the retained construction data, without adding hypotheses or auditing the submitted proof's individual steps. Its boundary-uniqueness lemma concerns a fixed analytic continuation, not uniqueness of the infrared limit. The compilation record and reproduction instructions document this supplementary Lean check, separate from the three Comparator runs. \end{remark} \subsection*{The seven consequences of the construction-chain audit} \addcontentsline{toc}{subsection}{The seven consequences of the construction-chain audit} The audit fixes a family \texttt{C} (\texttt{C.\allowbreak \allowbreak \allowbreak Makes}), an extraction order, the unselected mixed hierarchy \texttt{S : Table C.\allowbreak \allowbreak \allowbreak book}, a final witness \texttt{F : Spec.\allowbreak \allowbreak \allowbreak Final} retaining fields, stress tensor, phase construction, analytic continuation and relativistic reconstruction, and the acceptance predicate \texttt{Accepted} adding the review requirements (finite Wilson law, full-UV convergence). Combining the proved existence of an accepted witness (\texttt{YMReview.\allowbreak \allowbreak \allowbreak Acceptance.\allowbreak \allowbreak \allowbreak combined\_acceptance}) with the following seven lemmas gives consequences for any accepted construction; the proofs inspect definitions, not tactic history. \begin{enumerate} \item \emph{Insertion formula} (\texttt{actual\_insertion\_formula}, proved by \texttt{rfl}). The lattice insertion is \[ I_{N,p}(v,x;U)=a_{N}^{-d}\,(\texttt{places}\;\widetilde v)(x;U), \qquad \widetilde v=\big(\texttt{terminal}^{-1}\circ\text{diag}(t^{-1})\big)(v), \] where \texttt{terminal} undoes the terminal renormalization and $t\neq0$ is the nonzero RG threshold scale; the composite is a linear equivalence of the raw label space at each cutoff. This checks what \texttt{C.\allowbreak \allowbreak \allowbreak O} denotes; classical Taylor matching of local sections belongs to \texttt{C.\allowbreak \allowbreak \allowbreak Makes}, retained in \texttt{Accepted.\allowbreak \allowbreak \allowbreak lattice.\allowbreak \allowbreak \allowbreak makes}. \item \emph{No coefficient direction is removed} (\texttt{every\_primary\_is\_represented}). Every placed primary section is represented by an actual renormalized insertion up to the common factor $a_{N}^{-d}$, at every site of the fixed lattice. Finite-cutoff statement: no uniform bound on the inverse coefficient map as the cutoff is removed is provided, so this does not by itself exclude a degenerate continuum limit (the field-content and short-distance results do). \item \emph{Nonconstant primaries survive renormalization} (\texttt{nonconstant\_primary\_survives}). An invertible coefficient change and a nonzero common scale cannot turn every nonconstant primary observable into a constant. It does not assert that every local section is nonconstant or that chosen fields have nonzero limiting correlations. \item \emph{Lattice origin of continuum field words} (\texttt{every\_field\_word\_has\_lattice\_origin}). Every field word has fixed raw representatives, chosen once depending on the table and the word but not on cutoff or volume, and its entire mean correlation function is an iterated limit of lattice moments: \[ \operatorname{smear}(N,p,v,t)\to\mathrm{row}(p)\ (N\to\infty), \qquad \mathrm{row}(\texttt{o.\allowbreak \allowbreak \allowbreak per}\,k)\to\!\int t(x)\,\texttt{S.\allowbreak \allowbreak \allowbreak useAt}(x;a)\,\dd x\ (k\to\infty). \] No interchange of limits, no rate, no uniqueness across volume subsequences; the right-hand side uses the \emph{mixed} table \texttt{S}, and replacing it by a selected phase is not justified by this lemma. \item \emph{Phase readings are ground-state algebra products} (\texttt{phase\_second\_moments\_are\_ground\_products}). For distinct words $w,u$ in positive-time slabs, \[ \int_{X}\xi(w)\,\xi(u)\,\dd\nu(\xi) = \big\langle\Omega_{0},\,\texttt{slot}(w)\,\texttt{slot}(u)\,\Omega_{0}\big\rangle , \] via multiplicativity of characters: products of phase readings are tied to products in the actual commutative ground-state algebra. This is stronger than the barycenter identity alone, but it is not a factorization of a phase's multipoint correlation into one-point functions, and it does not say that each phase is itself a lattice subsequential limit or that there is only one phase. \item \emph{Reconstruction uses the selected phase's boundary distribution} (\texttt{reconstructed\_words\_are\_}\allowbreak\texttt{selected\_boundary}). The vacuum expectation of every reconstructed relativistic field word equals the expectation of the \emph{same} word in the selected phase's boundary distribution, through the realization certificate \texttt{qt.\allowbreak \allowbreak \allowbreak meets} belonging to the same \texttt{F}. \item \emph{A fixed continuation has a unique boundary value} (\texttt{same\_continuation\_same\_boundary}). If two contours from the same selected phase have equal analytic continuations, their distributional boundary pairings against every joint Schwartz test agree, by uniqueness of limits in $\C$. The hypothesis \texttt{same} is essential: this does not establish uniqueness of analytic continuation from Euclidean data, of the infrared subsequential limit, or of the phase. \end{enumerate} The transitive axiom dependencies of all seven lemmas are printed by \texttt{\#print axioms} in the same file; they use only the permitted foundations \texttt{propext}, \texttt{Classical.\allowbreak \allowbreak \allowbreak choice}, and \texttt{Quot.\allowbreak \allowbreak \allowbreak sound}. The ordinary compilation of these lemmas is supplementary evidence, not a Comparator run or a fresh rebuild. \section{Independent verification of the OS axioms} The OS comparison uses the centered trace of the stress tensor $\Theta$, defined by (9.2), in the selected component $\xi^{\ast}$. The adapter between the two formalizations changes coordinates to match the [DMY] choice of time axis and proves the corresponding identities for kernels and measures. It preserves the field, the state, and the complex correlations at every order. \subsection{The test space and the Schwinger functional} The domain of test functions for this check is \texttt{ZeroDiagonalSchwartz}, defined as \begin{equation} \mathcal{S}_{0}\big((\R^{4})^{n}\big) = \big\{f\in\mathcal{S}\big((\R^{4})^{n}\big): \partial^{\alpha}f\big|_{\Delta_{ij}}=0\ \text{for all $\alpha$ and $i\neq j$}\big\}, \end{equation} where $\Delta_{ij}=\{x:x_{i}=x_{j}\}$. These Schwartz tests vanish to every order on the coincidence diagonals, but need not vanish in a neighborhood of them or have compact support. They include the compactly supported smooth tests away from coincidences used in the lattice-to-continuum convergence statement. The Schwinger functional used in this independent verification is \begin{equation} \mathcal{S}_{n}(f)=\int_{(\R^{4})^{n}}K^{\xi^{\ast}}_{\Theta,n}(x)\,f(x)\,\dd x, \qquad K^{\xi^{\ast}}_{\Theta,n}(x):=S^{\xi^{\ast}}_{\Theta,\dots,\Theta}(x). \end{equation} The adapter proves that (8.1) is absolutely convergent and defines a continuous linear functional in the inherited Schwartz topology, and proves the $\mathrm{E}0'$ bound controlling growth with the number of insertions. This is check A1. \begin{remark}[Distributional scope] Several distinct distributional statements enter the construction. Lattice convergence first determines correlations away from coincidences. The independent OS comparison integrates the selected stress-trace kernels against joint Schwartz tests flat to all orders on every coincidence diagonal; these tests cannot detect contact terms supported on those diagonals. The submission's reconstruction separately supplies tempered Wightman distributions on unrestricted joint Schwartz tests. Euclidean contact extensions are additional constructions. The diagonal-flat OS comparison does not establish the uniqueness of these extensions or their satisfaction of OS positivity on unrestricted Euclidean test spaces. \end{remark} \subsection{Independence and scope of the check} The target definitions come from the formulation in the repository \texttt{xiyin137/OSreconstruction} [DMY], also in the Palomar registry. Its OS axioms and the corresponding Wightman axioms were formulated independently and before the group learned of the OpenAI Yang--Mills construction. The adapter proves that this selected scalar hierarchy satisfies those target OS axioms and the associated growth condition, using the submission's analytic estimates and reconstruction theorems. This is a check against independently specified axioms; its analytic input comes from the submitted proof. The extracted target declarations were checked against the pinned upstream definitions, and the adapters compiled in the submission's Lean 4.32.0 environment. The comparison concerns the centered-trace hierarchy, rather than a separate translation of every tensor field into the upstream formalization. \section{Other independent checks and their implications} \subsection{The finite lattice law} The independent lattice definitions come from \texttt{mrdouglasny/lgt} [LGT], at the pinned revision identified there. The checked adapter \texttt{LGTWilsonBridge.\allowbreak \allowbreak \allowbreak lean} reindexes sites, links, and plaquettes and proves equality of the normalized finite Wilson laws. In the conventions of that repository, \begin{equation} \beta_{\mathrm{submission}}=m\,\beta_{\mathrm{lgt}}, \end{equation} where $m$ is the matrix representation dimension. With this conversion, the action, partition function, probability measure, and integrals of integrable observables agree. The proof also matches the gauge actions and plaquette orientations. The comparison was done on Lean 4.32.0. This is check A3. The comparison is restricted to the finite-volume law; the pinned upstream repository also contains strong-coupling and mass-gap material (\texttt{LGT/MassGap/StrongCoupling.\allowbreak \allowbreak \allowbreak lean}, \texttt{LGT/YMMeasure.\allowbreak \allowbreak \allowbreak lean}), which the bridge does not invoke. \subsection{The Ward-identity adapter} The Ward-identity adapter derives equation~(9.1) from the charge and differentiation laws in the OpenAI submission. It covers every local field label and every component, and proves existence of admissible support for the bump function $\chi$, integrability, and independence of this support. This checks the formulation and translation normalization of the submission's Ward identity. This is check A6. \subsection{Common scale, normalization, and a physical probe} One needs to ensure that different lattice cutoffs and box sizes use the same reference units and field normalizations. The adapter extracts the following identities for the same family, establishing a common normalization convention across volumes (check A5): \begin{enumerate} \item the reference RG spacing is $a(0)=1$, fixing a common unit of length; \item for every cutoff $N$ and volume index $p$, the RG trajectory reaches $\beta_{N,p}(0)=\beta_{\mathrm{ref}}>0$; i.e.\ the RG flow starting at smaller lattice spacings (with coupling $\beta_{N,p}(N)$) results in the same reference coupling at lattice spacing $1$; \item kinetic normalization: the quadratic term in the expansion of the effective action is normalized to have identity matrix coefficients; \item insertion normalization: for a degree-$d$ label $v$, its normalized coefficient at the fixed reference level $r(d)$ equals $v$, independently of cutoff and volume, whenever $N\ge r(d)$. \end{enumerate} A second check connects this calibration to a nonzero observable in the mixed theory. One can choose a fixed separated trace two-point test function $f$ such that \begin{equation} \lim_{k\to\infty}\lim_{N\to\infty} \big|C_{N,p(q(k))}(\Theta,\Theta;f)\big|^{1/8} = \big|S^{\mathrm{mix}}(\Theta,\Theta;f)\big|^{1/8}>0. \end{equation} Here the lattice insertion uses the renormalized lattice representative of $\Theta$, which is centered by subtracting the selected component's expectation. Once chosen, the test and centering remain fixed at every cutoff and volume. \subsection{The short-distance adapter} The short-distance adapter expresses the submission's trace asymptotic (9.6) for approaches to the origin in all directions. It identifies the same centered trace used in the scalar OS check and the same stress tensor tested by the translation Ward identity. This is check A4. \subsection{OPE statement alignment} The group checks that the submitted OPE theorem implies the finite asymptotic expansion with explicit geometric conditions and bounds uniform over the phases. The purpose is to check the meaning and scope of the formal statement; it does not compute the coefficients independently or compare their values with Yang--Mills perturbation theory. The analytic estimates are supplied by OpenAI's proof. The review target \texttt{YMReview.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak CommonOPE} states two bounds directly, without invoking the source's \texttt{OperatorProductExpansion} predicate. For nearby fields of total degree $w$, it requires \begin{equation} |c_{\alpha}|\le r^{d_{\alpha}-w-\varepsilon}, \qquad |R_{D,s}|\le r^{D+1-w-\varepsilon}, \end{equation} where $R_{D,s}$ is the difference between the actual correlation and its finite OPE sum through degree $D$, evaluated in state $s$. The coefficients are independent of spectators, phases, and truncation orders. The check still uses the OpenAI submission's definitions of fields, coefficients, and correlations. Additional checks on the consistency of the field decomposition were performed; the two review modules compile without warnings, and all five audited declarations use only the permitted standard axioms. A separate Comparator run for the OPE statement passed on November 5, 2026. This is check A8. \section{Verification records and reproduction} The mathematical statements and their source definitions are collected in the Lean statement glossary of Appendix~A. This section records which additional checks were proved, which versions were checked, and where the build and Comparator results can be inspected. \subsection{Versions and repository layout} The verification material is in \texttt{yang-mills-group/YM} [YMRepo]. The October package is under \texttt{submissions/openai-20261028/}. Within that package, the submitted source is under \texttt{submission/}, the group's additional Lean code under \texttt{checks/}, the commented examples under \texttt{docs/examples/}, and the dated records under \texttt{evidence/}. Source and evidence links use this package layout; the assessment PDF is at the repository-level path \texttt{docs/assessment-1.\allowbreak \allowbreak \allowbreak pdf}. The submitted files are preserved byte for byte. The provenance directory records their hashes, the imported review sources, and the upstream definition snapshots. The review source manifest identifies the included copies of the review files and their hashes, linking the recorded Comparator runs to the sources in this release. For the declaration-by-declaration alignment with the upstream axiom systems, the files \texttt{ImportedAFGauge.\allowbreak \allowbreak \allowbreak lean}, \texttt{ImportedAFSpecification.\allowbreak \allowbreak \allowbreak lean}, \texttt{ImportedCenterWeights.\allowbreak \allowbreak \allowbreak lean}, \texttt{ImportedDiagonalFlatBounds.\allowbreak \allowbreak \allowbreak lean}, \texttt{ImportedLGTData.\allowbreak \allowbreak \allowbreak lean}, \texttt{ImportedReflectionApproximation.\allowbreak \allowbreak \allowbreak lean}, \texttt{ImportedSchwartzCutoffs.\allowbreak \allowbreak \allowbreak lean}, \texttt{ImportedSchwartzDomain.\allowbreak \allowbreak \allowbreak lean}, and \texttt{ImportedSchwartzProduct.\allowbreak \allowbreak \allowbreak lean} re-export the pinned upstream definitions used by the adapters; their hashes are part of the provenance record. Four counts appear in the records; they count different objects, not conflicting measurements: \begin{center} \small \begin{tabular}{@{}p{0.12\textwidth}p{0.78\textwidth}@{}} \toprule Count & Object \\ \midrule $9{,}477$ & Lean source files in the submission manifest. \\ $9{,}470$ & project modules rebuilt in the clean build. \\ $9{,}458$ & prebuilt modules whose bytes matched the original build inputs (the remainder being covered by import and kernel-declaration matches). \\ $9{,}564$ & Lean files in the full source archive (submission, checks, evidence, and provenance). \\ \bottomrule \end{tabular} \end{center} \subsection{Checks and their records} The identifiers M1--M2 are mechanical verification records, and A1--A8 are the group's additional checks; their mathematical meaning is explained in Sections~11.4--13.5. Table~1 links each check to its specification, proof, and verification record. Each A-entry also links a commented Lean example explaining its definitions, hypotheses, and scope. Where a check uses a specific numerical coefficient or another result from the literature, the corresponding worked example gives the supporting references and explains the conventions used in the comparison. These references support particular checks; they are not a general bibliography of the construction. The examples distinguish informal literature comparisons from the statements verified in Lean. \begin{table}[h] \centering \small \begin{tabular}{@{}p{0.16\textwidth}p{0.44\textwidth}p{0.34\textwidth}@{}} \toprule Check & Specification, proof, and worked example & Recorded verification \\ \midrule M1: Build and audits & The submitted project, its declaration inventory, and the principal proof dependencies. & 29 October clean build; axiom and declaration audit; kernel replay; warning summary. The 5 November packaged build reused the verified source build cache. \\ \addlinespace M2: Comparator & The public existence, combined acceptance, and OPE theorems, each checked against its own frozen reference. & Passed: public theorem, 30 October; combined theorem, 3 November; OPE theorem, 5 November. See also the four control results and tool revisions. \\ \addlinespace A1: Scalar OS axioms & \texttt{CompleteOS.\allowbreak \allowbreak \allowbreak lean} and \texttt{ArityGrowth.\allowbreak \allowbreak \allowbreak lean} prove the selected trace's target OS axioms and $\mathrm{E}0'$. Worked example: \texttt{OSAxioms-20261028.\allowbreak \allowbreak \allowbreak lean}. & Included in the combined theorem and the 5 November review audit. The upstream declarations and import provenance identify the independent definitions. \\ \addlinespace A2: Field content & \texttt{CompletenessAudit.\allowbreak \allowbreak \allowbreak lean}: field labels, bounded degree spaces, phase retention, density and faithfulness; holds for the individual phases. Worked example: \texttt{OperatorContent-20261028.\allowbreak \allowbreak \allowbreak lean}. & Included in the combined theorem; the module and its printed axiom dependencies are covered by the 5 November review audit. \\ \addlinespace A3: Finite lattice law & \texttt{LGTWilsonBridge.\allowbreak \allowbreak \allowbreak lean} compares the actual finite Wilson measures. Worked example: \texttt{LatticeAndScale-20261028.\allowbreak \allowbreak \allowbreak lean}. & The interface provenance records the compared revision of \texttt{mrdouglasny/lgt} [LGT]. Included in the combined theorem and the 5 November review audit. \\ \addlinespace A4: Leading trace law & \texttt{AFSourceBridge.\allowbreak \allowbreak \allowbreak lean} proves the leading short-distance behavior for the same centered trace. Worked example: \texttt{LeadingTrace-20261028.\allowbreak \allowbreak \allowbreak lean}. & Frozen in the combined theorem's dependencies; compilation and axiom evidence are in the combined build record and the 5 November review audit. \\ \addlinespace A5: Common scale and probe & \texttt{CommonScaleAudit.\allowbreak \allowbreak \allowbreak lean} and \texttt{TwoPointPhysicalObservable.\allowbreak \allowbreak \allowbreak lean}. Worked example: \texttt{LatticeAndScale-20261028.\allowbreak \allowbreak \allowbreak lean}. & Included in the 5 November review audit. \\ \addlinespace A6: Translation Ward identity & \texttt{WardFormulationAudit.\allowbreak \allowbreak \allowbreak lean}: the fixed sign and coefficient, shell integrability, and window independence. Worked example: \texttt{TranslationWard-20261028.\allowbreak \allowbreak \allowbreak lean}. & Included in the combined theorem and the 5 November review audit. \\ \addlinespace A7: Combined acceptance & \texttt{AcceptanceSpec.\allowbreak \allowbreak \allowbreak lean} and \texttt{CombinedAcceptance.\allowbreak \allowbreak \allowbreak lean} bind the checks to one family and final witness, retaining \texttt{C.\allowbreak \allowbreak \allowbreak Makes} and \texttt{Spec.\allowbreak \allowbreak \allowbreak Final}. Worked example: \texttt{OctoberChain.\allowbreak \allowbreak \allowbreak lean}. & 2 November build and axiom audit; 3 November Comparator pass. The imported source versions are retained in the review manifest. \\ \addlinespace A8: OPE statement & \texttt{OPESpec.\allowbreak \allowbreak \allowbreak lean} and \texttt{OPECheck.\allowbreak \allowbreak \allowbreak lean}: explicit-radius bounds uniform over phases derived from the submission's OPE and algebraic consistency checks; no perturbative coefficient comparison. Worked example: \texttt{CommonOPE-20261028.\allowbreak \allowbreak \allowbreak lean}. & Initial two-module audit; fresh review compilation; 5 November Comparator pass. This is a separate endpoint from the previously checked combined theorem. \\ \bottomrule \end{tabular} \caption{Checks, specifications, and recorded verification. M1--M2 are mechanical records; A1--A8 are the group's additional Lean checks.} \end{table} The commented examples explain the existing checks without changing their targets. The initial compilation record covers \texttt{OctoberChain.\allowbreak \allowbreak \allowbreak lean} and \texttt{LeadingTrace-20261028.\allowbreak \allowbreak \allowbreak lean}; the additional record covers the other five examples. All compiled without errors or warnings using only the permitted standard axioms. These ordinary Lean compilations are supplementary to the historical Comparator runs. \subsection{Reproducing the checks} From the release repository root, run \begin{verbatim} cd submissions/openai-20261028 python3 scripts/verify.py integrity python3 scripts/verify.py all \end{verbatim} The runner checks the imported files and OS-definition correspondence, builds the submission and review modules, audits axioms, builds the pinned tools, runs the controls, and checks all three Comparator endpoints. It writes a new dated record under \texttt{runs/}. The source and review use Lean 4.32.0 and Mathlib revision \texttt{81a5d257c8e410db227a6665ed08f64fea08e997}, and the pinned Comparator kernel uses Lean 4.34.0. A fresh source rebuild requires a checkout without existing submission build artifacts; a run that reuses those artifacts is recorded as incremental. \subsection*{Acknowledgements} The source assessment document was prepared by Michael R.~Douglas, Arthur Jaffe, Ron Nissim, and Fred Rajasekaran, with contributions from other group members and consultants, and approved by the entire group. The group used OpenAI's Astra to help translate the Lean code to informal mathematics and to develop the additional tests of the validation report, and Anthropic's Fable to give feedback on the document. \appendix \section{Lean statement glossary} The construction theorem \texttt{YangMills.\allowbreak \allowbreak \allowbreak constructed} supplies the data used to prove the public statement \texttt{YangMills.\allowbreak \allowbreak \allowbreak JointExistence}, displayed below. The October 30 Comparator run checked the proof of this public statement. \begin{verbatim} def JointExistence : Prop := forall g : GaugeGroup, g.IsSimple -> exists C : WilsonFamily g, WilsonRegularity C /\\ forall p : N -> N, StrictMono p -> exists q : N -> N, StrictMono q /\\ exists T : Hierarchy C, T.Properties (Function.comp p q) \end{verbatim} See \texttt{YangMills.\allowbreak \allowbreak \allowbreak constructed}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak jointExistence}. The definitions of \texttt{GaugeGroup} and \texttt{IsSimple} can be found in \texttt{PublicTheory.\allowbreak \allowbreak \allowbreak lean}; \texttt{IsSimple} requires the Lie algebra to be noncommutative and to have no proper nonzero ideals. The mathematical notation follows the main text; $S^{s}$ denotes a statement applying to either the mixture ($s=\mathrm{mix}$) or a component ($s=\xi\in X$). The selected Wightman theory has Hilbert space $\mathcal{H}_{\xi^{\ast}}$, domain $\mathcal{D}_{\xi^{\ast}}$, vacuum $\Omega_{\xi^{\ast}}$, and fields $\Phi^{\xi^{\ast}}_{v}$. Below, each entry records the identifier, its mathematical content, the construction's instantiation, and the principal Lean certificates. \subsection*{\texttt{YangMills.\allowbreak \allowbreak \allowbreak WilsonFamily}} Fix a closed matrix gauge group $G\subseteq U(m)$. The construction supplies these fields through \texttt{YangMills.\allowbreak \allowbreak \allowbreak wilsonFamily}. \begin{description} \item[\texttt{YangMills.\allowbreak \allowbreak \allowbreak WilsonFamily.\allowbreak \allowbreak \allowbreak period}.] Positive integers $L_{p}$ giving the physical side lengths. The construction uses $L_{p}=4(M+p+1)$, where $M\in\N$ is fixed sufficiently large to satisfy minimum-size and initial-volume bounds. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak RG.\allowbreak \allowbreak \allowbreak Supply.\allowbreak \allowbreak \allowbreak Program.\allowbreak \allowbreak \allowbreak book}; \texttt{YM.\allowbreak \allowbreak \allowbreak RG.\allowbreak \allowbreak \allowbreak Supply.\allowbreak \allowbreak \allowbreak Wait.\allowbreak \allowbreak \allowbreak engine}. \item[\texttt{YangMills.\allowbreak \allowbreak \allowbreak WilsonFamily.\allowbreak \allowbreak \allowbreak denominator}.] Positive integers $D_{N}$ giving lattice spacing $a_{N}=D_{N}^{-1}$. The construction sets $D_{N}=\prod_{k0\ \forall\varepsilon>0\ \exists N_{0}\in\N\ \forall N\ge N_{0}\ \forall p\in\N, \qquad \left|\frac{\beta_{N,p}}{\log(a_{N}^{-1})}-b^{\ast}\right|<\varepsilon. \end{equation} The construction uses $b^{\ast}=2b_{0}=11C_{\g}/(24\pi^{2})$, with $C_{\g}$ the adjoint Casimir constant. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Pert.\allowbreak \allowbreak \allowbreak Asymptotics.\allowbreak \allowbreak \allowbreak b0}; \texttt{YM.\allowbreak \allowbreak \allowbreak Gauge.\allowbreak \allowbreak \allowbreak casimir}. \item[\texttt{continuous}.] For every $N,p,v,x$, the function $U\mapsto I_{N,p}(v,x;U)$ is continuous in the product topology on configurations. Certificate: \texttt{YM.\allowbreak \allowbreak \allowbreak Family.\allowbreak \allowbreak \allowbreak continuousO}. \item[\texttt{invariant}.] For every sitewise gauge transformation $t:\Lambda_{N,p}\to G$, with $U^{t}_{y,\mu}=t(y)\,U_{y,\mu}\,t(y+e_{\mu})^{-1}$, \begin{equation} I_{N,p}(v,x;U^{t})=I_{N,p}(v,x;U), \end{equation} obtained from gauge-invariant local sections. Certificate: \texttt{YM.\allowbreak \allowbreak \allowbreak Family.\allowbreak \allowbreak \allowbreak invariant}. \item[\texttt{localSupport}.] Each token $v$ has a finite set $S_{v}\subset\Z^{4}$, independent of $N,p,x$, such that the insertion depends only on links with both endpoints in $x+S_{v}$ (periodically wrapped). For a constructed token $(d,c)$, $S_{v}$ is the fixed neighborhood chosen for the $P_{d,j}$. Certificate: \texttt{YM.\allowbreak \allowbreak \allowbreak Family.\allowbreak \allowbreak \allowbreak localO}. \end{description} \subsection*{\texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy}} Further requirements for some of these fields are in \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Properties}. The construction fills the fields through \texttt{YangMills.\allowbreak \allowbreak \allowbreak hierarchy}. Below, $S^{\mathrm{mix}}$ denotes the continuum Euclidean correlations obtained from the lattice limit, before selecting a phase. \begin{description} \item[\texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak H}.] The complex Hilbert space $\mathcal{H}_{\xi^{\ast}}$ of the selected phase: finite linear combinations of words of smeared fields, with inner products from the selected phase's reconstructed Wightman functions, quotiented by null vectors and completed. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak harvest}; \texttt{YM.\allowbreak \allowbreak \allowbreak Kernel.\allowbreak \allowbreak \allowbreak host}. \item[\texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak labels}.] The complex vector space $V$ labeling the fields of the Wightman theory. Given a token $t=(d,c)$, associate $R_{t}(n,b_{1},\dots,b_{n};x_{0},\dots,x_{n})=S^{\mathrm{mix}}(t,x_{0};b_{1},x_{1};\dots;b_{n},x_{n})$; then $V=\operatorname{span}_{\C}\{R_{(d,c)}:d\in\N,\ c\in V_{d}\}$. Certificates: \texttt{YangMills.\allowbreak \allowbreak \allowbreak labelsOf}; \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak Union}. \item[\texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak theory}.] A bundled structure containing the field operators $\Phi^{\xi^{\ast}}_{v}$, their common domain $\mathcal{D}_{\xi^{\ast}}$ and vacuum $\Omega_{\xi^{\ast}}$, the adjoint operation on labels, translations, an energy--momentum measure, and the Poincar\'e action. Certificates: \texttt{YangMills.\allowbreak \allowbreak \allowbreak view}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Soil.\allowbreak \allowbreak \allowbreak starter}. \item[\texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Field}.] A finite-dimensional complex normed vector space $\mathcal{F}_{d}$ for each degree bound $d$. The construction sets $\mathcal{F}_{d}=V_{d}/N_{d}$; writing $[c]$ for the equivalence class of a coefficient vector, the norm is the quotient norm from the supremum norm $\|c\|=\max_{i}|c_{i}|$ on the chosen basis of $V_{d}$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak null}; \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak Field}. \item[\texttt{toLabel}.] A complex-linear map $\mathcal{F}_{d}\to V$ sending $[c]$ to $R_{(d,c)}$; injective because $N_{d}$ is precisely the kernel of $c\mapsto R_{(d,c)}$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak spectra}; \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak pin}. \item[\texttt{project}.] Sends a Token to a Field by the quotient map $(d,c)\mapsto(d,[c])$. Certificate: \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak classTokens}. \item[\texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Character}.] The phase space $X$: a compact metric space with its Borel $\sigma$-algebra. Let $\mathcal{H}_{\mathrm{OS}}$ be the Hilbert space reconstructed from $S^{\mathrm{mix}}$, with time semigroup $E(t)=e^{-tH}$; let $\mathcal{H}_{0}=\ker H$ with projection $P_{0}$. For each word $A$ with distinct points in a time slab $00,\ A\ \text{contained in }00$, $r_{n}\to0$, and shows there are strictly increasing indices $n_{k}$ such that the set \begin{equation} E=\Big\{\xi\in X:\dim G_{\xi}=1,\ \text{and}\ \forall e\in\R,\ \lim_{k\to\infty}r_{n_{k}}^{8}\, S^{\xi}\big(\tilde\tau(0)\,\tilde\tau(r_{n_{k}}e_{0})\big)=0\Big\} \end{equation} satisfies $\nu(E)=1$, where $e_{0}=(1,0,0,0)$. The same subsequence works for every $\xi\in E$ and every $e\in\R$; the construction chooses $\xi^{\ast}\in E$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak OS.\allowbreak \allowbreak \allowbreak Cluster.\allowbreak \allowbreak \allowbreak lifts}; \texttt{YM.\allowbreak \allowbreak \allowbreak Trace.\allowbreak \allowbreak \allowbreak submits}; \texttt{YM.\allowbreak \allowbreak \allowbreak Trace.\allowbreak \allowbreak \allowbreak Echo.\allowbreak \allowbreak \allowbreak retreats}; \texttt{YM.\allowbreak \allowbreak \allowbreak Trace.\allowbreak \allowbreak \allowbreak echo\_tab}. \item[\texttt{correlation}.] Euclidean $n$-point functions of fields $(d_{i},v_{i})$ at positions $x_{i}\in\R^{4}$, with no phase specified ($S^{\mathrm{mix}}$) or a character $\xi$ ($S^{\xi}$). Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak spectator}; \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak useAt}; \texttt{YM.\allowbreak \allowbreak \allowbreak OS.\allowbreak \allowbreak \allowbreak Spread.\allowbreak \allowbreak \allowbreak Resolution.\allowbreak \allowbreak \allowbreak receives}. \item[\texttt{continuation}.] A function taking $n$ field labels, $n$ complex times $z_{i}$, and $n$ spatial positions $x_{i}\in\R^{3}$ to a complex number, written $F^{\xi^{\ast}}_{n}(v;z,x)$, using the analytic continuation of $S^{\xi^{\ast}}$ so that $S^{\xi^{\ast}}(v;\tau,x)=F^{\xi^{\ast}}_{n}(v;i\tau,x)$ for $\tau_{1}<\dots<\tau_{n}$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Waves.\allowbreak \allowbreak \allowbreak Region}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak hierarchy}. \item[\texttt{Channel}, \texttt{actor}, \texttt{coefficient}.] The index set for degree-$d$ fields on the right-hand side of the OPE: decomposing $\mathcal{F}_{d}=\mathcal{F}_{d-1}\oplus E_{d}$, the channels index a conjugation-invariant basis of $E_{d}$ (with $n_{d}=\dim E_{d}$); $\alpha=(d,i)$ has $d_{\alpha}=d$, and $\texttt{actor}(d,i)$ is the $i$-th basis element of $E_{d}\subseteq\mathcal{F}_{d}$. \texttt{coefficient} gives the complex OPE coefficient $c_{\alpha}(A;x,o)$ for input fields $A_{k}=(d_{k},v_{k})$ at positions $x_{k}$ about an expansion point $o$: the construction forms a field in $\mathcal{F}_{d}$ from finitely many correlations of the inputs with additional test insertions, then averages over $O(4)$, symmetrizes under conjugation, removes lower-degree components, and takes coordinates in the basis $O_{(d,i)}$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak Reader.\allowbreak \allowbreak \allowbreak cutTest}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak Reader.\allowbreak \allowbreak \allowbreak stratum}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak Reader.\allowbreak \allowbreak \allowbreak channels}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak Reader.\allowbreak \allowbreak \allowbreak basis}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak Reader.\allowbreak \allowbreak \allowbreak actor}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Prescription.\allowbreak \allowbreak \allowbreak ope}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak shadow}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak averageV}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak checkReal}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak Reader.\allowbreak \allowbreak \allowbreak fresh}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak Reader.\allowbreak \allowbreak \allowbreak coef}. \item[\texttt{centeredTrace}.] The element of $\mathcal{F}_{4}$ given by the trace of the Euclidean stress tensor minus its one-point expectation in the selected phase: \begin{equation} \Theta=\sum_{\mu=0}^{3}T_{\mu\mu} - S^{\xi^{\ast}}\Big(\sum_{\mu=0}^{3}T_{\mu\mu}\Big)\mathbf{1}, \end{equation} where $\mathbf{1}\in\mathcal{F}_{4}$ is the identity field, so $S^{\xi^{\ast}}(\Theta)=0$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak center}; \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak stressTrace}; \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak theta}. \end{description} \subsection*{\texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Properties}} These structure fields concern the same hierarchy $T$ and the selected volume indices $p(q(k))$ as in Theorem~6.4. The construction gives the proofs together in \texttt{YangMills.\allowbreak \allowbreak \allowbreak constructed}; the fields \texttt{reconstruction} and \texttt{nonfree} depend on further structures explained below. \begin{description} \item[\texttt{wilson}.] Requires the full UV limit of the smeared Wilson correlations at every fixed volume, followed by the infinite-volume limit along $p(q(k))$: \begin{equation} \lim_{k\to\infty}\lim_{N\to\infty}C_{N,p(q(k))}(v;f) =\int_{(\R^{4})^{n}}f(x)\,S^{\mathrm{mix}}(v;x)\,\dd x, \end{equation} where $v_{i}$ is a lattice token representing a classical label and $\bar v_{i}=\texttt{T.\allowbreak \allowbreak \allowbreak project}(v_{i})$ (the main text suppresses the bar). The limit holds for every number of insertions, token list $v$, and smooth test $f\in C^{\infty}_{c}$ away from coincidences; at fixed $k$, the support of $f$ must lie strictly inside the finite-volume box (which holds for all sufficiently large $k$). In the notation of (4.9), Lean's smear is \begin{equation} C_{N,p}(v;f)=a_{N}^{4n}\!\!\sum_{x_{1},\dots,x_{n}\in\Lambda_{N,p}} \!\!f(a_{N}x_{1},\dots,a_{N}x_{n})\int\prod_{i=1}^{n}I_{N,p}(v_{i},x_{i};U)\,\dd\mu_{N,p}(U), \end{equation} with $\Lambda_{N,p}=(\Z/(L_{p}D_{N})\Z)^{4}$, $a_{N}=D_{N}^{-1}$, the sum over ordered $n$-tuples, and $\mu_{N,p}$ the normalized Wilson measure. The construction obtains $S^{\mathrm{mix}}$ by these ordered limits and shows that tokens equivalent under \texttt{project} have equal limiting correlations. Certificates: \texttt{YangMills.\allowbreak \allowbreak \allowbreak WilsonFamily.\allowbreak \allowbreak \allowbreak smear}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak WilsonFamily.\allowbreak \allowbreak \allowbreak finiteMoment}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak WilsonFamily.\allowbreak \allowbreak \allowbreak place}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak wilsonLaw}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak WilsonFamily.\allowbreak \allowbreak \allowbreak Converges}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak wilsonLimit}. \item[\texttt{ope}.] Requires bounds on the OPE coefficients and the truncation error. For input fields $A_{i}\in\mathcal{F}_{d_{i}}$ at positions $x_{i}$, set $w=\sum_{i}d_{i}$; let $B_{\ell}$ be spectator fields at positions $y_{\ell}$ and $B=\prod_{\ell}B_{\ell}(y_{\ell})$. For fixed constants $M\ge0$ and $\eta,c>0$, the allowed configurations at scale $r>0$ satisfy \begin{equation} \begin{aligned} \|x_{i}-o\|&\le Mr &&\text{for every }i,\\ \|x_{i}-x_{j}\|&\ge\eta r &&\text{for }i\neq j,\\ \|y_{\ell}-o\|&\ge c &&\text{for every }\ell,\\ \|y_{\ell}-y_{m}\|&\ge c &&\text{for }\ell\neq m, \end{aligned} \end{equation} so the inputs form a cluster near the expansion point $o$ while spectators stay separated; the theorem applies uniformly over all configurations satisfying (A.10). With $O_{\alpha}:=\texttt{T.\allowbreak \allowbreak \allowbreak actor}(d,i)\in\mathcal{F}_{d}$ and $c_{\alpha}(A;x,o):=\texttt{T.\allowbreak \allowbreak \allowbreak coefficient}(A,d,i,x,o)$, define, for $s=\mathrm{mix}$ or $s=\xi\in X$, \begin{equation} R_{D,s} = S^{s}\Big(\prod_{i}A_{i}(x_{i})\,B\Big) - \sum_{d_{\alpha}\le D}c_{\alpha}(A;x,o)\,S^{s}\big(O_{\alpha}(o)\,B\big). \end{equation} The bounds are \begin{equation} |c_{\alpha}(A;x,o)|\le r^{d_{\alpha}-w-\varepsilon}, \qquad |R_{D,s}|\le r^{D+1-w-\varepsilon}, \end{equation} holding for all sufficiently small $r>0$ for every $\varepsilon>0$; the coefficient bound requires only the first two conditions of (A.10), while the remainder bound requires all four and holds uniformly over all states $s\in\{\mathrm{mix}\}\sqcup X$. The coefficients are independent of spectators, state, and truncation degree. Setting $o=0$ and input positions $rx_{i}$ recovers (9.12). Certificates: \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Within}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Clear}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak OperatorProductExpansion}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak coefficient\_separated}; \texttt{YM.\allowbreak \allowbreak \allowbreak OPE.\allowbreak \allowbreak \allowbreak Orbits.\allowbreak \allowbreak \allowbreak quantitative\_separated}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicOPE}. \item[\texttt{nonfree}.] Four conditions on the degree-four field $\Theta=\texttt{T.\allowbreak \allowbreak \allowbreak centeredTrace}$. Abbreviate, with $e_{0}=(1,0,0,0)$ and $\xi^{\ast}=\texttt{T.\allowbreak \allowbreak \allowbreak selected}$, \begin{equation} G_{\Theta}(r)=S^{\xi^{\ast}}\big(\Theta(-\tfrac{r}{2}e_{0})\,\Theta(\tfrac{r}{2}e_{0})\big). \end{equation} The construction proves the conditions together in \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicTrace}. Certificates: \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak TraceTest}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak traceTwo}. \begin{description} \item[\texttt{TraceTest.\allowbreak \allowbreak \allowbreak centered}.] The vacuum expectation vanishes for every complex Schwartz test: $\langle\Omega_{\xi^{\ast}},\Phi^{\xi^{\ast}}_{\Theta}(f)\Omega_{\xi^{\ast}}\rangle=0$ for all $f\in\mathcal{S}(\R^{4};\C)$, by analytically continuing the vanishing Euclidean one-point function. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak theta}; \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak vev\_boundary}; \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak pinned\_centered}. \item[\texttt{TraceTest.\allowbreak \allowbreak \allowbreak amplitudePositive}.] The short-distance coefficient is strictly positive: \begin{equation} A_{G}:=\frac{3\,\dim_{\R}\g}{\pi^{4}}>0, \end{equation} made explicitly available to downstream proofs. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Trace.\allowbreak \allowbreak \allowbreak Verse.\allowbreak \allowbreak \allowbreak award}; \texttt{YM.\allowbreak \allowbreak \allowbreak Trace.\allowbreak \allowbreak \allowbreak Over.\allowbreak \allowbreak \allowbreak force\_b}; \texttt{YM.\allowbreak \allowbreak \allowbreak Trace.\allowbreak \allowbreak \allowbreak Verse.\allowbreak \allowbreak \allowbreak award\_pos}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicTrace}. \item[\texttt{TraceTest.\allowbreak \allowbreak \allowbreak shortDistance}.] With $c(r)=r^{8}\log^{2}(1/r)$, \begin{equation} \lim_{r\downarrow0}c(r)\,G_{\Theta}(r)=A_{G}. \end{equation} The construction first shows the limit at approximated positions using lattice grid points; the ultraviolet comparison, logarithmic normalization, and free two-point calculation give $c(r_{n})\,G_{\Theta}(r_{n}\delta_{n}(s))\to A_{G}/s^{8}$ with $\delta_{n}(s)\to s$, where $\delta_{n}(s)$ corrects for approximating physical points separated by $r_{n}s$ by nearby lattice points ($s$ fixed near $1$). The positive spectral representation makes $G_{\Theta}$ real-valued and nonincreasing; choosing $s_{-}<10$, $\exists r\in(0,\delta):G_{\Theta}(r)\neq W(r)$. The construction proves that $\Reop W$ admits an expansion around $0$, $\Reop W(r)=\sum_{\nu=1}^{M}a_{\nu}r^{\alpha_{\nu}}(\log r)^{k_{\nu}}+O(1)$ with $\alpha_{\nu}\in\Z$, $k_{\nu}\in\N_{0}$, and that any finite limit of $r^{8}\log(1/r)\,\Reop W(r)$ must be zero; by (A.15), agreement near zero would instead give the strictly positive limit $A_{G}$ (see Remark~9.15 for the weight correction), a contradiction. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak Wick.\allowbreak \allowbreak \allowbreak real\_profile}; \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak Wick.\allowbreak \allowbreak \allowbreak AtOrder}; \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak Wick.\allowbreak \allowbreak \allowbreak no\_powerlog}; \texttt{YM.\allowbreak \allowbreak \allowbreak TraceExtension.\allowbreak \allowbreak \allowbreak Wick.\allowbreak \allowbreak \allowbreak euclidean\_excludes}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Mode}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Track}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak wickProfile}. \end{description} \item[\texttt{reconstruction}.] Fifteen conditions on the hierarchy $T$, proved together in \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicReconstruction}; see \texttt{YangMills.\allowbreak \allowbreak \allowbreak Hierarchy.\allowbreak \allowbreak \allowbreak Reconstructed}. \begin{description} \item[\texttt{axioms}.] The Wightman axioms W0--W3 for the smeared fields $\Phi^{\xi^{\ast}}_{v}(f):\mathcal{D}_{\xi^{\ast}}\to\mathcal{D}_{\xi^{\ast}}$, complex-linear in $v\in V$ and $f\in\mathcal{S}(\R^{4};\C)$: \begin{enumerate} \item[W0.] \emph{Translation spectrum and uniqueness of the vacuum.} Spacetime translations act through a strongly continuous unitary representation $U(a)$ with projection-valued energy--momentum measure $E$, $\langle\psi,U(a)\psi\rangle=\int_{\R^{4}}e^{ip\cdot a}\,\dd\mu_{\psi}(p)$, $\mu_{\psi}(B)=\langle\psi,E(B)\psi\rangle$; $E$ is supported in the closed forward light cone $V_{+}=\{p:p^{\mathrm{time}}\ge\|p^{\mathrm{space}}\|\}$; and every translation-invariant vector is a complex multiple of $\Omega_{\xi^{\ast}}$. \item[W1.] \emph{Common domain, regularity, adjoints, cyclicity.} $\mathcal{D}_{\xi^{\ast}}$ is dense, contains the normalized vacuum, and is preserved by every smeared field; for fixed $v$ and $\psi$, $f\mapsto\Phi^{\xi^{\ast}}_{v}(f)\psi$ is continuous from the Schwartz to the Hilbert norm topology; with the conjugate-linear adjoint map $v\mapsto v^{\ast}$, $\langle\Phi^{\xi^{\ast}}_{v}(f)\psi,\chi\rangle =\langle\psi,\Phi^{\xi^{\ast}}_{v^{\ast}}(\bar f)\chi\rangle$; and $\mathcal{D}_{\xi^{\ast}}$ is exactly the span of field words applied to the vacuum, the empty word being $\Omega_{\xi^{\ast}}$. \item[W2.] \emph{Poincar\'e covariance.} The proper orthochronous Poincar\'e group acts through a strongly continuous unitary representation extending the translations, and on labels through complex-linear bijections $\rho(g)$ with $\rho(1)=\mathrm{id}$, $\rho(gh)=\rho(g)\rho(h)$; it preserves $\mathcal{D}_{\xi^{\ast}}$, fixes the vacuum, and $U(g)\Phi^{\xi^{\ast}}_{v}(f)\psi =\Phi^{\xi^{\ast}}_{\rho(g)v}(f_{g})\,U(g)\psi$ with $f_{g}(x)=f(\Lambda^{-1}(x-a))$ for $gx=\Lambda x+a$. \item[W3.] \emph{Local commutativity.} If every point of $\supp f$ is spacelike separated from every point of $\supp h$, then $\Phi^{\xi^{\ast}}_{v}(f)\Phi^{\xi^{\ast}}_{w}(h)\psi =\Phi^{\xi^{\ast}}_{w}(h)\Phi^{\xi^{\ast}}_{v}(f)\psi$. \end{enumerate} The construction proves these in the selected phase: Wightman boundary values define the inner product on finite words; the isometric map from $\mathcal{H}_{\xi^{\ast}}$ into the selected phase's OS Hilbert space intertwines translations and sends the Wightman vacuum to the OS vacuum; nonnegativity of energy and Lorentz covariance give the forward-cone spectrum; and translation-invariant vectors on the Wightman side map into the vacuum subspace on the OS side, so by injectivity they lie in the Wightman vacuum subspace. Certificates: \texttt{YangMills.\allowbreak \allowbreak \allowbreak FieldSystem}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak WightmanTheory.\allowbreak \allowbreak \allowbreak Axioms}; \texttt{YM.\allowbreak \allowbreak \allowbreak Kernel.\allowbreak \allowbreak \allowbreak host}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Soil.\allowbreak \allowbreak \allowbreak starter}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Soil.\allowbreak \allowbreak \allowbreak sharp}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Soil.\allowbreak \allowbreak \allowbreak usable}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Soil.\allowbreak \allowbreak \allowbreak regional}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Soil.\allowbreak \allowbreak \allowbreak mobilize}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Soil.\allowbreak \allowbreak \allowbreak cultivates}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak view}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Outfield.\allowbreak \allowbreak \allowbreak daylight}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak harvest}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Action.\allowbreak \allowbreak \allowbreak spectrum}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Action.\allowbreak \allowbreak \allowbreak only}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak view\_axioms}. \item[\texttt{faithful}.] Every nonzero label $v$ has a test $f$ with $\Phi^{\xi^{\ast}}_{v}(f)\neq0$: vanishing of all smeared Wightman correlations containing a label forces vanishing of all Euclidean correlations containing a representative of that label, with arbitrary spectators at pairwise distinct points, and these are exactly the labels quotiented out. Certificates: \texttt{YangMills.\allowbreak \allowbreak \allowbreak FieldSystem.\allowbreak \allowbreak \allowbreak Faithful}; \texttt{YM.\allowbreak \allowbreak \allowbreak Work.\allowbreak \allowbreak \allowbreak C26.\allowbreak \allowbreak \allowbreak answer}; \texttt{YM.\allowbreak \allowbreak \allowbreak Boundary.\allowbreak \allowbreak \allowbreak Soil.\allowbreak \allowbreak \allowbreak sees}. \item[\texttt{injective}.] For each $d$, distinct fields in $\mathcal{F}_{d}$ have distinct images under \texttt{toLabel}. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak faithfully}; \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak pin\_inj}. \item[\texttt{acts}.] Some field in some $\mathcal{F}_{d}$ has a nonzero smeared operator (via the identity field in $\mathcal{F}_{0}$), excluding vacuous faithfulness. Certificate: \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicReconstruction}. \item[\texttt{covers}.] Every label in $V$ is $\texttt{T.\allowbreak \allowbreak \allowbreak toLabel}(d,v)$ for some $d$ and $v\in\mathcal{F}_{d}$. Certificate: \texttt{YM.\allowbreak \allowbreak \allowbreak DiffPlan.\allowbreak \allowbreak \allowbreak covered}. \item[\texttt{represents}.] Every pair $(d,v)$, $v\in\mathcal{F}_{d}$, is the image of a lattice token under \texttt{project} (a quotient map). Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak representative}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicReconstruction}. \item[\texttt{normalized}.] $\nu$ has total mass one: $\|\Omega_{0}\|^{2}=1$. Certificate: \texttt{YM.\allowbreak \allowbreak \allowbreak OS.\allowbreak \allowbreak \allowbreak Ground.\allowbreak \allowbreak \allowbreak resolution}. \item[\texttt{fullSupport}.] Every nonempty open subset of $X$ has positive $\nu$-measure, since no nonzero element of the ground-state algebra annihilates $\Omega_{0}$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak OS.\allowbreak \allowbreak \allowbreak Ground.\allowbreak \allowbreak \allowbreak attains}; \texttt{YM.\allowbreak \allowbreak \allowbreak OS.\allowbreak \allowbreak \allowbreak Ground.\allowbreak \allowbreak \allowbreak supports}. \item[\texttt{permutation}.] Correlations (mixed and component) are invariant under simultaneously permuting fields and insertion points, inherited from lattice integrals and passed to the continuum limit and each phase. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak class\_perm}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicReconstruction}. \item[\texttt{smooth}.] Correlations (mixed and component) are $C^{\infty}$ away from coincidences. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Table.\allowbreak \allowbreak \allowbreak class\_smooth}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicReconstruction}. \item[\texttt{characterContinuity}.] For fixed fields at distinct points, correlations are continuous in $\xi\in X$, via $S^{\xi}(A)=\xi(M_{A})$ and continuity of character evaluation. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak OS.\allowbreak \allowbreak \allowbreak Spread.\allowbreak \allowbreak \allowbreak Resolution.\allowbreak \allowbreak \allowbreak Out}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicReconstruction}. \item[\texttt{mixture}.] Averaging phases recovers mixed correlations: $\int_{X}S^{\xi}(v;x)\,\dd\nu(\xi)=S^{\mathrm{mix}}(v;x)$, via $\int_{X}\xi(M_{A})\,\dd\nu(\xi)=\langle\Omega_{0},M_{A}\Omega_{0}\rangle =\langle\Omega_{0},P_{0}B_{s,A}\Omega_{0}\rangle=\langle\Omega_{0},B_{s,A}\Omega_{0}\rangle =S^{\mathrm{mix}}(A)$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak OS.\allowbreak \allowbreak \allowbreak Spread.\allowbreak \allowbreak \allowbreak Resolution.\allowbreak \allowbreak \allowbreak barycenter}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicReconstruction}. \item[\texttt{analytic}.] For each fixed list of fields and spatial positions, $F^{\xi^{\ast}}_{n}(v;z,x)$ is holomorphic in $z$ on $\operatorname{Im}z_{1}<\dots<\operatorname{Im}z_{n}$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Work.\allowbreak \allowbreak \allowbreak C03.\allowbreak \allowbreak \allowbreak defined\_exists}; \texttt{YM.\allowbreak \allowbreak \allowbreak Work.\allowbreak \allowbreak \allowbreak C03.\allowbreak \allowbreak \allowbreak defined\_region}. \item[\texttt{root}.] At purely imaginary times the continuation equals the selected-phase Euclidean correlation: for $\tau_{1}<\dots<\tau_{n}$, $F^{\xi^{\ast}}_{n}(v;i\tau,x)=S^{\xi^{\ast}}_{n}(v;(\tau_{i},x_{i})_{i=1}^{n})$. Certificates: \texttt{YM.\allowbreak \allowbreak \allowbreak Work.\allowbreak \allowbreak \allowbreak C02.\allowbreak \allowbreak \allowbreak answer}; \texttt{YangMills.\allowbreak \allowbreak \allowbreak publicReconstruction}. \item[\texttt{boundary}.] Approaching real times from any fixed strictly ordered imaginary direction recovers the smeared Wightman vacuum expectation: for $h_{1}<\dots0$ with the dominance condition $2b_{0}\log 3>K/\beta_{\ast}$ of (5.2); uniform in volume. The source-level form is \texttt{Large.slow}/\texttt{Fits.driftSmall} (Lemma~5.9); the dominance condition is sufficient but not necessary. & \texttt{YM.\allowbreak \allowbreak RG.\allowbreak \allowbreak Supply.\allowbreak \allowbreak Program.\allowbreak \allowbreak book}, \texttt{YM.\allowbreak \allowbreak RG.\allowbreak \allowbreak Calendar.\allowbreak \allowbreak Fits}, \texttt{YM.\allowbreak \allowbreak Pert.\allowbreak \allowbreak Asymptotics.\allowbreak \allowbreak b0}, \texttt{YM.\allowbreak \allowbreak Price.\allowbreak \allowbreak Intent.\allowbreak \allowbreak Rolls.\allowbreak \allowbreak exists\_rolls}. \\ (RG3) & Reference-level insertion readout is stable; RG transport invertible with controlled inverse. & \texttt{CommonScale.\allowbreak \allowbreak reference\_insertion}, \texttt{TransferData.\allowbreak \allowbreak cylinder}. \\ (RG4*) & Kernel-level difference estimate (5.3): derived from (RG2)$+$(RG3) as in Remark~5.15. & Same as (RG4); \texttt{YM.\allowbreak \allowbreak Family.\allowbreak \allowbreak continuousO}, \texttt{YM.\allowbreak \allowbreak Family.\allowbreak \allowbreak localO}. \\ (RG5) & Uniform smoothness estimates for the lattice kernels on compact separated sets. & VD \S4.3, \S5.2 Rem.~5.2. \\ (UV1) & Local boundedness and equicontinuity of all derivatives, uniform in volume. & VD \S5.2 Rem.~5.2; reflection positivity, chessboard, transfer operator, Cauchy. \\ (PH1) & Time-translation conjugation structure of word operators. & \texttt{YM.\allowbreak \allowbreak OS.\allowbreak \allowbreak Spread}, \texttt{YM.\allowbreak \allowbreak OS.\allowbreak \allowbreak Semigroup.\allowbreak \allowbreak ground}. \\ (PH2) & The ground-state algebra $\mathcal{A}$ is commutative. & \texttt{YM.\allowbreak \allowbreak OS.\allowbreak \allowbreak Spread.\allowbreak \allowbreak algebra}. \\ (PH3) & $\Omega_{0}$ is separating for $\mathcal{A}$ (faithfulness). & \texttt{YM.\allowbreak \allowbreak OS.\allowbreak \allowbreak Ground.\allowbreak \allowbreak attains}, \texttt{.\allowbreak \allowbreak supports}. \\ (AN1) & OS axioms E0/E0$'$/E1--E4, normalization, and analytic continuation for the selected phase. & Appendix~C; \texttt{YMOSBridge.\allowbreak \allowbreak verifiedOS}, \texttt{complete\_OS\_on\_this\_hierarchy}. \\ (Y1) & The stress-tensor candidate space is one-dimensional, with normalization fixed by the Ward identity. & VD \S7.1; \texttt{WardFormulationAudit.\allowbreak \allowbreak lean}. \\ (Y2) & Flux (shell) law and calculus law for the abstract field derivative. & \texttt{phase.\allowbreak \allowbreak flux}, \texttt{phase.\allowbreak \allowbreak calculus}; \texttt{TranslationWard-20261028.\allowbreak \allowbreak lean}. \\ (Y3) & Reduction of the trace two-point function to the free-field residue times the squared anomaly coefficient. & \texttt{YM.\allowbreak \allowbreak Trace.\allowbreak \allowbreak Over.\allowbreak \allowbreak Scribe}, \texttt{YM.\allowbreak \allowbreak Trace.\allowbreak \allowbreak Verse.\allowbreak \allowbreak tagline}. \\ (Y4) & One-loop value $b_{0}=11C_{\g}/(48\pi^{2})$ from the loop sum. & \texttt{YM.\allowbreak \allowbreak Pert.\allowbreak \allowbreak Asymptotics.\allowbreak \allowbreak b0}, \texttt{YM.\allowbreak \allowbreak Gauge.\allowbreak \allowbreak casimir}. \\ (B1) & Field-content dictionary: $N_{d}=\ker\psi_{d}$, rank comparison and phase transport (Proposition~2.12). & \texttt{OperatorCompleteness.\allowbreak \allowbreak local\_field\_content}, \texttt{phase\_field\_content}, \texttt{F.\allowbreak \allowbreak phase.\allowbreak \allowbreak transport.\allowbreak \allowbreak comparator}, \texttt{F.\allowbreak \allowbreak phase.\allowbreak \allowbreak phase.\allowbreak \allowbreak retains}. \\ (C1) & Finite-law comparison with the independent lattice formalization (Proposition~4.16). Record only: it does not enter any subsequent derivation. & \texttt{LGTWilsonBridge.\allowbreak \allowbreak lean}, \texttt{same\_plaquette}, \texttt{coupling\_conversion}, \texttt{normalized\_law\_on\_each\_lattice}, \texttt{finite\_moments\_are\_the\_family\_moments}. \\ \bottomrule \end{tabular} \emph{Derived, not an input:} (RG4) --- $|C_{N+1,p}-C_{N,p}|\le A(v,f)\lambda^{N}$ (Lemma~5.14); its kernel-level form (RG4*) by Remark~5.15. \emph{Proved spectral conclusion, not an input:} (PH4) --- $\mathcal{H}_{0}^{\xi^{\ast}}=\C\,\Omega_{\xi^{\ast}}$ (Proposition~7.2; certificates \texttt{YM.OS.Cluster.lifts}, \texttt{YM.Trace.submits}, \texttt{selected}, (A.6)). \emph{The number seventeen in the repository records counts the headline results; it is unrelated to the sixteen external inputs above.} \end{center} \section{Comparison with the \texorpdfstring{[DMY]}{[DMY]} OS axioms} The comparison of Section~12 is made against the independently developed OS formulation of [DMY] (repository \texttt{xiyin137/OSreconstruction}, also in the Palomar registry). The target definitions were extracted from a pinned upstream snapshot, checked against it declaration by declaration, and the adapters were compiled in Lean 4.32.0. The comparison concerns the centered-trace hierarchy in the selected phase $\xi^{\ast}$ only; it is not a translation of every tensor field into the upstream formalization. The adapter changes coordinates to match the [DMY] choice of time axis (\texttt{configEquiv}); it preserves the field, the state, and the complex correlations at every order. \subsection*{Bridge objects} \begin{center} \small \begin{tabular}{@{}p{0.30\textwidth}p{0.64\textwidth}@{}} \toprule Upstream object & Construction counterpart \\ \midrule \texttt{SpacetimeDim 3} & Euclidean spacetime $\R^{4}$ (dimension index $3+1$). \\ \texttt{NPointDomain 3 n} & Configuration domain of $n$-tuples in $\R^{4}$. \\ \texttt{scalarKernel F n} & The selected phase's $n$-point function $K^{\xi^{\ast}}_{\Theta,n}=S^{\xi^{\ast}}_{\Theta,\dots,\Theta}$ of the centered trace $\Theta$ (Section~9.2); reality is supplied by the source reality theorem. \\ \texttt{targetKernel F n} & Real kernel on \texttt{NPointDomain 3 n}, set to zero on coincidence diagonals \emph{by convention} (a measurability device; not a claim that the physical distribution has no contact terms). \\ \texttt{schwinger F n} & The Schwinger functional $\mathcal{S}_{n}(f)=\int K^{\xi^{\ast}}_{\Theta,n}(x)f(x)\,\dd x$ of Section~12.1. \\ \texttt{configEquiv} & Coordinate change matching the [DMY] time axis. \\ \bottomrule \end{tabular} \end{center} \subsection*{Axiom-by-axiom correspondence} The test space throughout is $\mathcal{S}_{0}((\R^{4})^{n})$ (\texttt{ZeroDiagonalSchwartz 3 n}), the joint Schwartz tests flat to all orders on every coincidence diagonal $\Delta_{ij}$; it contains the compactly supported smooth tests away from coincidences used in the lattice-to-continuum convergence. \begin{center} \small \begin{tabular}{@{}p{0.17\textwidth}p{0.34\textwidth}p{0.42\textwidth}@{}} \toprule {}[DMY] item & Statement in upstream form & Construction counterpart and certificate \\ \midrule E0 (temperedness) & $\mathcal{S}_{n}$ is continuous on $\mathcal{S}_{0}$ (Schwartz topology); integrability is proved before the functional is identified with the integral (Lean's integral is totalized). & Kernels are smooth away from coincidences with a finite inverse-power bound in the minimum separation: \texttt{separated\_bound} (\texttt{YMOSBridge.\allowbreak \allowbreak \allowbreak targetKernel\_bound}); \texttt{pairing\_is\_integrable} (\texttt{schwinger\_integrable}); \texttt{E0\_continuity} (\texttt{E0\_tempered}). \\ \addlinespace E0$'$ (arity growth) & $\exists$ slope $s$, base $\beta$, factorial exponent $A$ \emph{chosen before} $n$ and $f$: $\lVert\mathcal{S}_{n}(f)\rVert\le\beta^{n}(n!)^{A}\,\lVert f\rVert_{(ns,ns)}$ (Schwartz seminorm at index $(ns,ns)$). & \texttt{YM.\allowbreak \allowbreak \allowbreak Work.\allowbreak \allowbreak \allowbreak B11.\allowbreak \allowbreak \allowbreak phase\_cap} supplies a common bound $K^{n}(n/r)^{An}$ at separation $r$; \texttt{ArityGrowth.\allowbreak \allowbreak \allowbreak lean} transports it through the diagonal-flat integral estimate, absorbing $n^{An}$ into an exponential times $(n!)^{A}$: \texttt{E0\_prime\_uniform\_arity} (\texttt{YMOSBridge.\allowbreak \allowbreak \allowbreak E0\_prime\_growth}). \\ \addlinespace E1 (Euclidean invariance) & Rotations (\texttt{os\_7} $\leftarrow$ \texttt{Trace.\allowbreak \allowbreak \allowbreak rotation}, full rotation) and translation covariance (\texttt{os\_8} $\leftarrow$ \texttt{Trace.\allowbreak \allowbreak \allowbreak translated}). & Section~8.4 (gauge fixing preserving lattice symmetries, divergence-free stress tensor, no spontaneous breaking); packaged in \texttt{YMOSBridge.\allowbreak \allowbreak \allowbreak verifiedOS}. \\ \addlinespace E2 (reflection positivity) & $\operatorname{Re}\,\mathrm{OSInnerProduct}(B,B)\ge0$ for every Borchers sequence $B$ supported in the ordered positive-time region. & Section~7.1: the finite compact-test reflection form is identified entry by entry with the source positive form; support-preserving cutoffs and E0 continuity extend it to the full domain: \texttt{E2\_full\_target\_form} (\texttt{YMOSBridge.\allowbreak \allowbreak \allowbreak E2\_reflection\_positive}). \\ \addlinespace E3 (permutation) & Invariance under simultaneous permutation of fields and insertions (\texttt{os\_2} $\leftarrow$ \texttt{Moments.\allowbreak \allowbreak \allowbreak Perm}). & \texttt{Properties.\allowbreak \allowbreak \allowbreak permutation} (inherited from lattice integrals, passed to the limit and each phase); packaged in \texttt{verifiedOS}. \\ \addlinespace E4 (clustering) & Separated groups factorize: for $f\in\mathcal{S}_{0}^{n}$, $g\in\mathcal{S}_{0}^{m}$, $\lVert\mathcal{S}_{n+m}(f\,g_{a})-\mathcal{S}_{n}(f)\mathcal{S}_{m}(g)\rVert<\varepsilon$ for $\lVert a\rVert$ large. & The selected phase's vacuum factorization (Proposition~7.15), extended to joint Schwartz pairings; \texttt{CompleteOS.\allowbreak \allowbreak \allowbreak E4\_cluster} supplies the precise spatial-translation clause: \texttt{E4\_separated\_groups} (\texttt{YMOSBridge.\allowbreak \allowbreak \allowbreak E4\_cluster\_norm}). \\ \addlinespace Normalization & Probability/normalization (\texttt{os\_1} $\leftarrow$ \texttt{Moments.\allowbreak \allowbreak \allowbreak probability}); \texttt{verifiedOS} also packages the required linearity. & $\nu(X)=1$, $\lVert\Omega_{0}\rVert^{2}=1$ (\texttt{phaseMeasure}, \texttt{Properties.\allowbreak \allowbreak \allowbreak normalized}). \\ \addlinespace Analyticity & Analyticity of the Schwinger functions (\texttt{os\_3} $\leftarrow$ \texttt{Moments.\allowbreak \allowbreak \allowbreak analytic}). & Section~8: analytic continuation $F^{\xi^{\ast}}_{n}(v;z,x)$, $\operatorname{Im}z_{1}<\dots<\operatorname{Im}z_{n}$ (\texttt{Properties.\allowbreak \allowbreak \allowbreak analytic}, \texttt{root}). \\ \bottomrule \end{tabular} \end{center} \begin{center} \small \begin{tabular}{@{}p{0.17\textwidth}p{0.34\textwidth}p{0.42\textwidth}@{}} \toprule Complete record & \multicolumn{2}{p{0.76\textwidth}}{ \texttt{verifiedOS} packages E1--E4, normalization, and linearity into \texttt{OsterwalderSchraderAxioms 3} with $\mathtt{S}$ \emph{equal} to the Schwinger pairing above (no second choice of theory); \texttt{complete\_OS\_on\_this\_hierarchy} appends \texttt{OSArityLinearGrowthCondition 3} (E0$'$), compiled against the imported upstream definition, with the source-correspondence audit recording the snapshot match.} \\ \bottomrule \end{tabular} \end{center} \subsection*{Scope of the comparison} \begin{enumerate} \item Diagonal-flat test functions annihilate distributions supported on coincidence diagonals. The comparison therefore does not establish uniqueness of Euclidean contact extensions, nor OS positivity of such an extension on unrestricted Euclidean test spaces; these are different questions from the axioms proved here. \item Neither clustering statement specifies an exponential rate, proves a mass gap, or makes the infrared extraction unique. \item The file \texttt{OSAxioms-20261028.\allowbreak \allowbreak \allowbreak lean} verifies this instance of the imported OS definition; it does not prove or invoke the entire Douglas--Mei--Yin reconstruction theorem. The submission's own Wightman reconstruction is retained elsewhere in the combined target. \end{enumerate} \subsection*{Proof-chain details: clustering and the packaged record} \emph{The E4 clustering chain} runs through six review modules; we record the role of each: \begin{itemize} \item \texttt{ClusterKernel.\allowbreak \allowbreak \allowbreak lean}: geometric preliminaries---a large signed coordinate can be produced by a coordinate swap (\texttt{exists\_signed\_large\_coordinate}), the configuration coordinate change preserves norms, and separated configurations stay separated under translation-joined insertion of a far group (\texttt{joinConfig\_injective}, $\lVert x\rVert+\lVert y\rVert<\lVert a\rVert$). \item \texttt{ClusterCore.\allowbreak \allowbreak \allowbreak lean}: the dynamical core---for a chamber $V$ with unique heat, ground-state projection of heat-evolved vectors, the mixed heat limit, and clustering of arbitrary registered words along the chamber axis; specialized to every character phase (\texttt{phase\_word\_cluster}, \texttt{phase\_mixed\_heat\_limit}, the latter in the boundary Hilbert space with its actual energy evolution). \item \texttt{ClusterCones.\allowbreak \allowbreak \allowbreak lean} and \texttt{ClusterDomination.\allowbreak \allowbreak \allowbreak lean}: the cone geometry and the domination estimates reducing translation clustering to the chamber axis. \item \texttt{ClusterTarget.\allowbreak \allowbreak \allowbreak lean}: transport to the target kernels (\texttt{targetKernel\_radial\_cluster}, with threshold $\max\{R,\lVert x\rVert+\lVert y\rVert\}$). \item \texttt{ClusterSchwartz.\allowbreak \allowbreak \allowbreak lean}: transport from point configurations to joint Schwartz pairings, yielding \texttt{E4\_separated\_groups}. \end{itemize} No rate is asserted anywhere in the chain; the threshold is uniform over all directions of the separating translation. The absence of a rate is located at the input, not lost in the transport: \texttt{RemainingObligations.lean} keeps E4 as the single explicit premise of \texttt{completeOS}, and the six modules transport an already-given E4 (Proposition~10.22). Consequently a rate must be added at the entrance (\texttt{CompleteOS.E4\_cluster}), uniformly over all registered label words. \emph{The packaged record} (\texttt{RemainingObligations.\allowbreak \allowbreak \allowbreak lean}, namespace \texttt{YMOSBridge}) discharges the axioms E0, E1, E2, E3 as theorems and keeps E4 as the single explicit premise of the record \texttt{completeOS}; feeding \texttt{CompleteOS.\allowbreak \allowbreak \allowbreak E4\_cluster} yields the full target axioms (\texttt{complete\_OS\_on\_this\_hierarchy}). The growth condition is used in its corrected log-factor form $\texttt{GrowthObligation}$ with exponent $\alpha=1$ and is discharged directly by \texttt{E0\_prime\_growth} ($\texttt{E0\_prime}$). Finally, \texttt{constructed\_tempered\_family} and \texttt{constructed\_reflection\_positive\_family} instantiate the record on the actual outputs of the submitted existence theorem (\texttt{Independent.\allowbreak \allowbreak \allowbreak fundedRun}), leaving the original ultraviolet order and the extracted infrared order intact and providing a single witness for both E0 and the full target E2. The transitive axiom dependencies of \texttt{completeOS}, $\texttt{E0\_prime}$, and both family theorems are printed in the file. \section{The schedule-fitting conditions: source record and verdict} This appendix records, verbatim from the submission source, the declarations bearing on the feasibility of the renormalization-group constants: whether the blocking ratios, the reference coupling, and the blocking-kernel penalty admit one consistent choice. The question is decidable at source level, and the verdict is stated at the end. \emph{The penalty is universally quantified and fixed before the family.}\ \texttt{Construction.\allowbreak lean}: \begin{verbatim} theorem constructed (g : GaugeGroup) (hg : g.IsSimple) (b : Real) (hb : 0 < b) : exists C : YM.Family g.core, C.path.levels.clock.kernel.penalty = b /\ C.Makes /\ WilsonRegularity (wilsonFamily C) /\ forall p : Extraction.Order, exists (q : Extraction.Order) (S : Table C.book) (F : Final C (p.refine q) S) (L : EveryPhase.Almanac F) (Q : Reader F.sector.derivatives F.sector.turns F.room.kernel.spin) (M : Contacts.Devices F.sector.derivatives) (X : Climb.Scaling S), (hierarchy F Q).Properties (p.refine q).per /\ (hierarchy F Q).Contacts /\ (Extraction.Order.mk Extraction.Sequence.full (p.refine q).per).Ordered C S /\ \end{verbatim} \emph{Penalty locus.} \texttt{YangMills.constructed}'s penalty enters the \emph{clock kernel} $c$ (\texttt{hG : G.\allowbreak Simple}, then \texttt{C.\allowbreak path.\allowbreak levels.\allowbreak clock.\allowbreak kernel}), not the gauge coupling; \texttt{PublicWilsonField}'s \texttt{BoundaryLimit} has \emph{no} penalty argument. \emph{The one-loop constant and the running law.}\ \texttt{YM/Pert/Asymptotics/RawAxis.\allowbreak lean}: \begin{verbatim} def b0 (G : Gauge) := 11*G.casimir / (48*Real.pi^2) theorem b0_normalizes : 2*b0 G = G.slopeConstant := by rw [b0,Gauge.slopeConstant]; ring \end{verbatim} \emph{The Casimir constant.}\ \texttt{YM/Gauge/Casimir.\allowbreak lean}: \begin{verbatim} import YM.Gauge.Colors import Mathlib.Analysis.InnerProductSpace.Spectrum import Mathlib.Analysis.InnerProductSpace.Dual /-! # The adjoint Casimir constant of the gauge image in the color coordinates fixed in `Colors`. In particular the simplicity here is *real*, and has actual consequences for later Ward and schedule arguments. -/ set_option autoImplicit false noncomputable section namespace YM.Gauge open LinearMap Submodule open scoped Matrix.Norms.Frobenius InnerProductSpace variable {G : Gauge} theorem Simple.nonabelian (h : G.Simple) : exists x y, G.ad x y /= 0 := by \end{verbatim} \emph{The explicit fitting inequalities.} The blocking ratios are not free integers: they are $\lceil(\log\cdot)^{2}\rceil$-type quantities. The schedule is fitted by two conditions, each required to hold \emph{eventually} in the schedule entries (the entries are then chosen as maxima with the corresponding thresholds). The penalty enters only by an additive shift.\ \texttt{YM/Overlay/Schedule.\allowbreak lean}: \begin{verbatim} def scaleB (x : Real) := Nat.ceil (Real.log x^2) lemma stepsize (k : Real) : forall x in atTop, (scaleB (x+k))^2<=2*(scaleB x)^2 := by lemma stepsize (k : Real) : forall x in atTop, (scaleB (x+k))^2<=2*(scaleB x)^2 := by let a := fun x : Real => log (x+k) - log x have hz : Tendsto a atTop (nhds 0) := Real.tendsto_log_comp_add_sub_log k have he := hz.eventually (lt_mem_nhds -- should >_ (show (-1:Real)<0 by norm_num)) filter_upwards [he,hz.eventually (gt_mem_nhds (zero_lt_one' Real)), Real.tendsto_log_atTop.eventually_ge_atTop 20] with x hx hh ht have h1 : -1< log (x+k)-log x := hx; have h2 : log (x+k)-log x<1 := hh have hi : (scaleB x:Real) >= log x^2 := Nat.le_ceil _ \end{verbatim} \begin{verbatim} lemma afford (L : Nat) (gamma : Real) (hG : G.Simple) (A : Stage) (o : Real) (ho : 0 gamma+ia*s let Gt := 2*(1+(4*A.paths+1))+1+(Fintype.card PlaqAxes:Real)*(A.clip.direct+A.paths^2) let f : Real := 4 *Module.finrank Real G.Mat \end{verbatim} The constant $M$ bounding the cost side of the affordability condition is linear-log: \begin{verbatim} (L*(scaleB x:Real)^2)^ (4:Nat)*Vac.fiat G A (x+G.slopeConstant*b) <= o*(s+b) := by let ia := G.slopeConstant have ha : 0 < ia := Gauge.slopeConstant_pos hG let x := fun s => gamma+ia*s let Gt := 2*(1+(4*A.paths+1))+1+(Fintype.card PlaqAxes:Real)*(A.clip.direct+A.paths^2) let f : Real := 4 *Module.finrank Real G.Mat let M := Gt+f*(log 16+1)+1 -- linear have h1 : 0 < Gt := by unfold Gt; have hv := A.paths_pos; have ha := A.clip.quadratic_pos; positivity have h2 : 0 <= f := by unfold f; positivity \end{verbatim} Whether \texttt{Nonperturbative.existence} discharges \texttt{Fits}, or \texttt{Fits} enters through \texttt{WilsonRegularity}, was not read (\textbf{L, PARTIAL}; the discharge lemma \texttt{Fits.discharge} and its callee \texttt{Nonperturbative.existence} were not read; \texttt{Rolls.exists\_rolls} is closed, so the downstream argument does not depend on this). \begin{verbatim} \end{verbatim} \begin{proposition}[Source-level verdict on the feasibility of the RG constants] \textup{(i)} The penalty is universally quantified: for every $b>0$ and every simple group, the construction produces a family whose clock kernel has penalty $b$, chosen before the family. \textup{(ii)} The blocking ratios $b_{k}$ are $\operatorname{scaleB}(\hat\beta_{k})$ with $\operatorname{scaleB}(x)=\lceil(\log x)^{2}\rceil$: polylogarithmic, not free. \textup{(iii)} The two fitting conditions --- the stepsize doubling condition and the affordability condition, whose cost side is bounded by $M\cdot z$ with $M$ linear-log and $z=1+\log x$ against a linear budget --- hold eventually in the schedule entries; the penalty enters only through the additive shift $x\mapsto x+\mathrm{slopeConstant}\cdot b$ together with the harmless condition $\log(x+ia\,b)-\log x<1$. \textup{(iv)} Consequently every growth datum supplied by the source is consistent with Case~1 of the trichotomy for the feasibility condition; the identification of the \texttt{drift} with the one-step remainder is recorded as a source-level reading (\textbf{L}, Remark~10.42). The verdict of this appendix is therefore D.2/D.3 with \textbf{L PARTIAL} (Appendix~E, Section~E.14). \end{proposition} \begin{remark}[The remaining omission is exactly the one registered] The $E$ and $K$ constants are located at \texttt{YM.\allowbreak RG.\allowbreak ActivityBounds}'s \texttt{KBound}/\texttt{EBound}: they are summability caps for polymer-gas/regular occurrences (\texttt{Activities.KBound}/\texttt{Activities.EBound}), not numerical remainders; their exponents are therefore the cluster-expansion exponents, and a numerical value of the Doeblin constant is recorded in Section~E.6 of Appendix~E. \end{remark} \begin{remark}[Repository counts and the number seventeen] See Section~14.1 for the four repository counts, tabulated there; the present remark records only the disambiguation. Three module counts appear in the records and are not interchangeable: $9{,}477$ is the number of source files matching the manifest; $9{,}470$ is the number of project modules rebuilt in the clean build; $9{,}458$ is the number of prebuilt modules whose bytes matched the original build inputs (the remainder being covered by import and kernel-declaration matches). The number ``seventeen'' in the repository records counts the headline results; the external inputs registered in Appendix~B number sixteen, of which (PH4) is recorded as a proved conclusion (Proposition~7.2), not an input. The fourth count, $9{,}564$, is the total number of Lean files in the full source archive (submission, checks, evidence, and provenance). \end{remark} \section{The mass-gap problem: complete status} This appendix records the complete adjudicated status of the mass-gap problem: the rate theorems and the counterexample family, the random-mechanics criteria, the cluster-expansion and local-mixing routes with their blockers, the numerical anchor, the source-level adjudication of the earlier Appendix~D, and the cyclic reduction. Statements marked \textbf{L} rest on source-level readings recorded in Remark~10.42. \emph{Numbering and provenance.} The section numbers E.1--E.15 continue the correspondence with Section~10: E.1 gives the rate theorems behind Propositions~10.20 and 10.22; E.2 the counterexample family behind Propositions~10.5, 10.7 and 10.9; E.3 the random-mechanics criteria; E.4--E.5 the cluster-expansion and local-mixing routes; E.6 the numerical anchor; E.7 the block criteria; E.8 the source adjudication; E.9 the cyclic reduction (Theorem~E.19); E.10 the closure tightening; E.11--E.13 the type-level evidence and final status; E.14 the re-adjudication of Appendix~D; E.15 the boundary declaration. References \emph{(S3-$k$)} are to file $k$ of the review package; \emph{09} denotes the LGT provenance modules [OS78, Seiler]. \subsection{Rate theorems: uniformity is the whole content} \begin{theorem}[Uniform rate theorem] Let $\{\mu_{p}\}$ be a family of finite positive measures on $[0,\infty)$ with Laplace transforms $F_{p}(t)=\int e^{-t\lambda}\,\dd\mu_{p}(\lambda)$. Assume \begin{equation} \textup{(L)}\qquad \sup_{p}F_{p}(1)<\infty, \qquad\text{and}\qquad F_{p}(t)\to F_{\infty}(t)\ \text{for every }t>0 . \end{equation} Then the limit measure $\mu_{\infty}$ has a spectral gap $\big(\mu_{\infty}((0,\varepsilon))=0$ for some $\varepsilon>0\big)$ \emph{if and only if} there exists $\varepsilon_{0}>0$ with $\mu_{p}((0,\varepsilon_{0}))=0$ for all $p$. \end{theorem} \begin{proof} If $\mu_{p}((0,\varepsilon_{0}))=0$ for all $p$, then $F_{p}(t)\le\mu_{p}([0,1])e^{-\varepsilon_{0}t}+e^{-t}\sup_{p}F_{p}(1)$ for $t\ge1$, so $F_{\infty}(t)\le C e^{-\varepsilon_{0}t}$, and Lemma~10.1 gives the gap. Conversely, if $\mu_{\infty}$ has a gap $\varepsilon_{\infty}>0$ but no uniform $\varepsilon_{0}$ exists, choose $\varepsilon_{p}\to0$ with $\mu_{p}((0,\varepsilon_{p}))>0$; then for $t$ fixed, $F_{p}(t)\ge a_{p}e^{-\varepsilon_{p}t}$ with $a_{p}>0$ arbitrarily small---no contradiction with (L) is forced, and indeed Proposition~E.2 shows the pointwise hypothesis cannot exclude $\mu_{\infty}((0,\varepsilon))>0$ for every $\varepsilon>0$: condition (L) only sees the \emph{values} $F_{p}(t)$, while a gap is a statement about the \emph{location} of the mass near zero. The uniform condition $\mu_{p}((0,\varepsilon_{0}))=0$ is exactly what converts pointwise convergence into spectral information. \end{proof} \begin{proposition}[The pathological family] There exists a family $\{\mu_{p}\}$ satisfying (E.1) such that each $\mu_{p}$ is purely atomic with gap $m_{p}=\frac{1}{2p+1}>0$, and such that $\mu_{\infty}$ has no gap. Consequently \emph{(L) alone}---pointwise Laplace convergence with a uniform value bound---does not imply a gap, and the hypothesis of Proposition~10.20 cannot be weakened from ``uniform in $p$'' to ``for each $p$''. \end{proposition} \begin{proof} Take $\mu_{p}=\delta_{0}+w_{p}\delta_{m_{p}}+\rho_{\mathrm{UV}}$ with $m_{p}=\frac{1}{2p+1}$, $w_{p}\le1$, and a fixed ultraviolet tail $\rho_{\mathrm{UV}}$ supported in $[1,\infty)$ of polynomial growth; this is the support-separated construction of Proposition~10.7 with the infrared atom moved to $m_{p}$. Then $F_{p}(t)\to F_{\infty}(t)$ for every $t>0$ (dominated convergence on the moving atom, which has mass $\le1$ and location $\to0$), and $\sup_{p}F_{p}(1)<\infty$; but $\mu_{\infty}$ carries the full weight $w=\limsup w_{p}>0$ arbitrarily close to $0$ (choose $w_{p}\to w>0$), so $\mu_{\infty}((0,\varepsilon))>0$ for every $\varepsilon>0$ and no gap exists. \end{proof} \begin{remark} The family is consistent with the reached closure of the sources (Remark~10.31); it is not ruled out by anything read in the repositories. It shows that the ``uniformity over volumes'' in (MG2$''$) and the ``uniform rate over label words'' in (MG) are \emph{independent} necessary qualifications: Proposition~10.24 kills common-rate relaxations, and Proposition~E.2 kills volume-wise relaxations. \end{remark} \subsection{The counterexample family} Propositions~10.5, 10.7 and 10.9 of the main text are the $p=\infty$ instances of a volume-indexed family $\{T_{p}\}$ of tempered generalized free fields, each satisfying (AN1), (B1), (Y1), the calculus and flux laws (Y2), the exact short-distance constant $A_{G}$, and the OPE coefficients, with per-volume spectral gaps $m_{p}=\frac{1}{2p+1}\to0$. The family witnesses that every registered input holds volume by volume while the uniform predicate (MG) fails; by Proposition~E.2 its Laplace transforms converge pointwise with $\sup_{p}F_{p}(1)<\infty$. No non-Gaussian upgrade is available without cost: a mixture over a non-atomic base loses (PH4) (the base acts trivially), and an ergodic base restores (PH4) only conditionally on the base, leaving the model conditionally Gaussian (Remark~10.41). For the decidability of the gap question the Gaussian family is sufficient; the existence of a genuinely interacting gapless model satisfying the registered inputs is a question of model classification, recorded in Remark~10.42. \subsection{Random-mechanics criteria} The criteria of this section are the \emph{finite-size} shadow of Proposition~10.22: they diagnose a gap from correlation data at finite separation, and they diagnose its absence. \begin{theorem}[Laplace pointwise convergence is insufficient] There is a family $F_{p}(t)=(1-\tfrac1p)e^{-t/p}+\tfrac1p$ satisfying $\sup_{p}F_{p}(1)\le1$ with $F_{p}\to F_{\infty}\equiv1$ pointwise, while each $F_{p}$ is the Laplace transform of a measure with gap $1/p>0$ and $F_{\infty}$ is the Laplace transform of $\delta_{0}$ (no gap). Pointwise convergence of Laplace transforms carries no information about the behaviour as $t\downarrow0$. \end{theorem} \begin{proof} Direct computation; the family is a mass-$(1-\tfrac1p)$ atom at $1/p$ plus mass $\tfrac1p$ at $0$. \end{proof} \begin{theorem}[Autocorrelation criterion for free models] Let $T$ be a free (Gaussian) field with two-point function $G(t)$ and spectral measure $\rho$, and define the finite-separation autocorrelation \[ \alpha_{V}\;=\;\frac{G(2V)-G(V)^{2}}{1-G(V)^{2}} \;=\;\frac{\displaystyle\int e^{-V\lambda}\,\dd\tilde\rho(\lambda)} {\displaystyle\int\big(1+e^{-2V\lambda}\big)\,\dd\rho(\lambda)}, \qquad \tilde\rho=\rho-\rho(\{0\})\delta_{0}. \] Under (E.1), $\alpha_{V}$ is a fixed functional of the spectral data (no fitting parameter enters), the consistency check $\alpha_{V}\in[0,1]$ is free, and \[ \exists\,\varepsilon_{0}>0:\ \rho\big((0,\varepsilon_{0})\big)=0 \quad\Longleftrightarrow\quad \exists\,V_{0},\delta>0:\ \alpha_{V}\le1-\delta\ \text{for all }V\ge V_{0} \quad\Longleftrightarrow\quad \inf_{V}\alpha_{V}<1 . \] \end{theorem} \begin{proof} The first equality is the definition; the second rewrites $G(2V)-G(V)^{2}=\int e^{-V\lambda}\,\dd\tilde\rho(\lambda)$ (Wick). If $\rho((0,\varepsilon_{0}))=0$, then $\tilde\rho$ is supported on $[\varepsilon_{0},\infty)$ and $\alpha_{V}\le e^{-\varepsilon_{0}V}$. Conversely, if $\rho((0,\varepsilon))>0$ for every $\varepsilon>0$, choose $\varepsilon_{V}\to0$ with $\rho([\varepsilon_{V},2\varepsilon_{V}))>0$; the numerator is then $\ge\rho([\varepsilon_{V},2\varepsilon_{V}))\,e^{-2V\varepsilon_{V}}$ while the denominator is $\le2$, so $\alpha_{V}\ge\tfrac12\rho([\varepsilon_{V},2\varepsilon_{V}))e^{-2V\varepsilon_{V}}$; choosing $V=V(\varepsilon)$ with $V\varepsilon_{V}\to0$ gives $\alpha_{V}\to1$. \end{proof} \begin{remark} This is the criterion used in Section~8.3 via Proposition~7.2: for free models, E4-with-a-rate is the same statement as $\inf_{V}\alpha_{V}<1$. The reduction to a Markov chain (spectral gap of the transfer operator) is verified here only for Gaussian models; the non-Gaussian mixing case is unverified (retraction R11). \end{remark} \begin{theorem}[False positives when (E.1) fails] There are families $\{G_{p}\}$ of two-point functions, each with spectral gap $\ge\varepsilon_{0}>0$, and a limit $G_{\infty}$ with no gap, such that the finite-$p$ autocorrelation estimates converge to a limit strictly below $1$; and families with no gap whose finite-$p$ estimates suggest a gap. Hence every finite-size criterion must include a uniformity hypothesis; pointwise statements are not stable under the failures of (E.1). \end{theorem} \begin{proof} Variant of the construction of Theorem~E.4 with the moving atom carrying a $p$-dependent weight that biases the finite-separation estimator: choose $w_{p}=c/V_{p}$ with $V_{p}\to\infty$ slowly, so that at separation $V_{p}$ the estimator sees the atom, while at fixed separation the limit measure is $\delta_{0}$. Both directions are obtained by reversing the sign of the bias. \end{proof} \begin{theorem}[A combined finite-size criterion] Suppose that for some $V_{0},\delta>0$ and all $V\ge V_{0}$ the uniform estimate \[ \sup_{p}\ \alpha^{(p)}_{V}\ \le\ 1-\delta \] holds for the autocorrelations of all registered label words. Then (MG) holds with rate $\varepsilon_{0}=-\log(1-\delta)/V_{0}$, hence the limit theory has a gap (Proposition~10.20). The criterion is sufficient; its failure at any fixed $V$ proves nothing (Theorem~E.7). \end{theorem} \begin{proof} $\alpha^{(p)}_{V}\le1-\delta$ forces $\tilde\rho_{p}((0,\varepsilon_{0}))=0$ with $\varepsilon_{0}=-\log(1-\delta)/V_{0}$ by the converse direction of Theorem~E.5, uniformly in $p$ and in the word; then Proposition~10.20 applies. \end{proof} \begin{proposition}[Reverse direction] If the limit theory has a gap $\varepsilon_{0}$, then for every registered label word there are $C_{A},V_{0}$ with $\alpha^{(p)}_{A,V}\le C_{A}e^{-\varepsilon_{0}V}$ for all $V\ge V_{0}$, all $p$ \emph{along the subsequence defining the limit}; the constants may degenerate along other subsequences (Proposition~E.2). \end{proposition} \begin{proof} Lemma~10.2 applied volume by volume. \end{proof} \subsection{The cluster-expansion route} \begin{theorem}[Hard direction] A convergent cluster expansion for the blocked polymer gas yields (MG2$''$): the Dobrushin uniqueness condition at a fixed blocking level, uniform in the volume, gives $\|\mathcal{T}_{N,p}|_{\Omega^{\perp}}\|\le e^{-a_{N}\varepsilon_{0}}$ with an explicit $\varepsilon_{0}$ from the expansion's convergence radius. The hard part is not the implication but the convergence domain: the expansion is organized around the \emph{reference} layer, and its domain is a strong-coupling neighborhood. \end{theorem} \begin{proof}[Proof sketch] Standard: the polymer gas representation of the Wilson law at the reference layer [OS78, Seiler]; the Kirkwood--Salzburg equations give a Neumann series convergent in the Dobrushin uniqueness region, and the resulting two-point bounds imply the transfer-matrix contraction by the $h$-transform of Lemma~10.12. \end{proof} \begin{theorem}[The equivalence] For the Wilson plaquette gas, the following are equivalent: \begin{enumerate} \item[\upshape(i)] a convergent cluster expansion at the reference layer, uniform in the volume; \item[\upshape(ii)] uniform perimeter bounds of Osterwalder--Seiler type [OS78], $\langle W_{C}\rangle\le C e^{-\sigma\,\mathrm{perimeter}(C)}$ with $\sigma>0$ not degenerating in the volume; \item[\upshape(iii)] the transport estimate (MG2$''$). \end{enumerate} In particular (MG4)-type area laws and (MG2)-type spectral bounds are the same estimate in this setting, and the compact-set OPE check of Section~13.5 is the finite-separation shadow of (ii). \end{theorem} \begin{proof}[Proof sketch] (i)$\Rightarrow$(ii) is the Osterwalder--Seiler estimate; (ii)$\Rightarrow$(i) is the standard finite-range Gibbs equivalence (external input, cf.\ [OS78, Seiler]); (i)$\Rightarrow$(iii) by Theorem~E.10; (iii)$\Rightarrow$(i) by integrating the transfer-matrix bound against polymer activities. \end{proof} \begin{remark} The running coupling (4.8) drives the bare theory to weak coupling, so the strong-coupling domain of Theorem~E.10 does not contain the trajectory; the bridge required by (MG1) is a no-phase-transition statement, which is not available (Remark~10.15 shows the layer-changing argument does not obstruct strong-coupling estimates at the reference layer, but does not supply them). \end{remark} \subsection{The local-mixing route} \begin{theorem}[Doeblin] Restatement of Lemma~10.12: a uniform lower bound $K\ge c_{1}>0$ on the one-step kernel gives a Doeblin constant of order $(c_{1}/c_{2})^{2}=e^{-2R}$ and $\Delta\ge-a_{t}^{-1}\log(1-e^{-2R})>0$. \end{theorem} \begin{proposition}[Degeneration] Restatement of Proposition~10.16: at the reference layer the global bound degenerates, $R\ge c_{0}\bar\beta L_{p}^{3}$, so a single global application proves nothing; a \emph{local} mixing estimate with volume-independent rate is required. \end{proposition} \begin{theorem}[Poincar\'e/log-Sobolev criterion] If the single-site conditional measures of the reference-layer Gibbs measure satisfy a uniform log-Sobolev inequality with constant $\gamma>0$ independent of $L_{p}$, then the associated Glauber dynamics has spectral gap $\ge\gamma/2$, and (MG2$''$) holds with $\varepsilon_{0}=\gamma/2$ in lattice units. No such inequality is recorded in the sources; no Lean predicate about it exists (Proposition~10.30). \end{theorem} \begin{proof}[Proof sketch] Standard Bakry--\'Emery/ Holley--Stroock argument for finite-range Gibbs measures [GJ, Seiler]; the volume-independent constant is the content. \end{proof} \begin{theorem}[Local mixing criterion] A block-dynamics argument: if every cube of fixed side $\ell$ equilibrates in time $\tau(\ell)$ uniformly in the volume, and the block dynamics on the coarse lattice mixes in time $\tau_{\mathrm{block}}$, then the spectral gap is $\ge(\tau(\ell)+\tau_{\mathrm{block}})^{-1}$. The criterion reduces (MG2$''$) to a local statement; it is the route taken by the strong-coupling analyses [OS78, Seiler] and by L\"uscher-type finite-volume analyses. No numerical $\tau(\ell)$ is available in the sources. \end{theorem} \subsection{The numerical anchor} The following values are recorded from the literature as \emph{conventions and anchors}, not as inputs to any proof in this paper. For four-dimensional $SU(2)$ lattice gauge theory in the standard normalization, single-link Dobrushin--Wasserstein criteria are reported sharp at $\beta_{W}<2/9$, and plaquette-block criteria at $\beta_{W}\approx0.259$; both concern fixed-lattice strong-coupling regions. The glueball mass is numerically $1.5$--$1.9\,\mathrm{GeV}$ with lattice spacing $0.09$--$0.12\,\mathrm{fm}$ under standard scale setting; the conversion to physical units conventionally sets the string tension or the Sommer scale, and different conventions shift the number. None of these values enters the deductive chain; they locate the strong-coupling region relative to the running trajectory (4.8), which approaches weak coupling. \subsection{Block criteria} \begin{theorem}[Four sufficient blocks] Each of the following implies (MG2$''$), with $\varepsilon_{0}$ explicit, uniformly in $N,p$: \begin{enumerate} \item[\upshape(B1)] Strong-coupling Dobrushin region $\beta<\beta_{W}$ with the kernel bounds of [OS78]; \item[\upshape(B2)] Finite-range equivalence of Gibbs measures (cf.\ [OS78, Seiler]); \item[\upshape(B3)] Convergence of the reference-layer cluster expansion (Theorem~E.10); \item[\upshape(B4)] A transfer-matrix gap at the reference layer, uniform in the volume, transported across layers by Proposition~10.14. \end{enumerate} In each case (MG2$''$) yields the gap for every infinite-volume subsequence by Proposition~10.20. \end{theorem} \begin{proof}[Proof sketch] (B1) by Lemma~10.12; (B2) by the uniqueness theorem for finite-range Gibbs measures; (B3) by Theorem~E.10; (B4) by Proposition~10.14 and the same Laplace argument as Proposition~10.20. \end{proof} \begin{remark} None of the four blocks is currently occupied: the trajectory (4.8) is ultraviolet/weak-coupling, the required convergence domains are strong-coupling, and the layer-changing mechanism of Proposition~10.14 preserves but does not create a gap. The blocks are the known sufficient shapes of (MG2$''$); their occupation is precisely the open problem. \end{remark} \subsection{Source adjudication} \begin{enumerate} \item[\emph{E.8.1}] \emph{Penalty locus.} \texttt{YangMills.constructed}'s penalty enters the \emph{clock kernel} $c$ (\texttt{hG : G.\allowbreak Simple}, then \texttt{C.\allowbreak path.\allowbreak levels.\allowbreak clock.\allowbreak kernel}), not the gauge coupling; \texttt{PublicWilsonField}'s \texttt{BoundaryLimit} has \emph{no} penalty argument (\textbf{L}). \item[\emph{E.8.2}] \emph{Pointwise bounds.} The framework ``no uniform bound on the inverse coefficient map'' is retained from Remark~5.21; it is tightened by the reached closure (item E.10 below). \item[\emph{E.8.3}] \emph{The $E$ and $K$ constants.} They are located at \texttt{YM.\allowbreak RG.\allowbreak ActivityBounds}'s \texttt{KBound}/\texttt{EBound}: summability caps for polymer-gas/regular occurrences (\texttt{Activities.\allowbreak KBound}/\texttt{EBound}), not numerical remainders; their exponents are the cluster-expansion exponents, and a numerical value of the Doeblin constant is recorded in Section~E.6. \item[\emph{E.8.4}] \emph{TransferData route withdrawn; \texttt{Fits} discharge.} The route from \texttt{TransferData.\allowbreak Based} to the schedule constants is withdrawn: the coefficient-norm bounds carry factors $\mathrm{columns}\cdot b_{m}^{d}$ (Remark~5.21), and the discharge of \texttt{Fits} through \texttt{Nonperturbative.\allowbreak existence} was not read (\textbf{L, PARTIAL}; \texttt{Fits.discharge} and its callee were not read; \texttt{Rolls.exists\_rolls} is closed, so the downstream argument does not depend on this). \item[\emph{E.8.5}] \emph{Drift identity.} The identification of the source's \texttt{drift} (\texttt{Intent.\allowbreak Large.slow/kick}) with the one-step remainder $\varepsilon_{m}$ is supported by the shared shape $|\beta_{m+1}-\beta_{m}+\mathrm{g}\log L|\le\mathrm{drift}$; the definitions of \texttt{Survey.melt}, \texttt{stage}, \texttt{profile} were not read (\textbf{L}; Remark~10.42(1)). \end{enumerate} \subsection{The cyclic reduction} \begin{theorem}[All routes close on the single predicate] The conjunction of the sixteen registered inputs, (B1), (Y1)--(Y4) and the proved ultraviolet theorems is consistent with both a spectral gap and its absence. More precisely: \begin{enumerate} \item[\upshape(1)] (MG2$''$) $\Rightarrow$ gap for every infinite-volume subsequence (Proposition~10.20); \item[\upshape(2)] cluster-expansion routes give (MG2$''$) only inside strong-coupling domains that the running trajectory (4.8) does not occupy (Theorem~E.11, Remark~E.12); \item[\upshape(3)] global local-mixing bounds degenerate with the volume (Proposition~10.16); volume-independent local criteria (Theorems~E.15, E.16) are not recorded in the sources; \item[\upshape(4)] the kernel family fixes the finite-volume spectral measures (Proposition~10.18); what is missing is an \emph{estimate} on them, and the signatures of the sources contain no gap-shaped predicate (Proposition~10.30); \item[\upshape(5)] the random-mechanics criteria diagnose but do not supply the uniformity (Theorems~E.1, E.4, E.7, E.8). \end{enumerate} Hence the mass-gap problem is exactly the problem of proving the single volume-uniform, label-uniform rate (MG) (Proposition~10.34), and its converse direction (gap $\Rightarrow$ (MG)) is open (Remark~10.42(6)). \textbf{A positive mass gap is not proved.} \end{theorem} \begin{proof}[Proof] Items (1)--(5) are proved at the cited statements; consistency of the inputs with both outcomes is witnessed by Proposition~10.7 (no gap, all registered ultraviolet data) and by any free massive model (gap, all registered data), both satisfying (AN1), (B1), (Y1)--(Y4) by the constructions of Section~10. \end{proof} \subsection{Closure tightening} The earlier formulation ``not a receipt dischargeable inside the formalization'' is tightened as follows: the kernel replay reached closure over 100{,}168 project declarations with only the three permitted axioms, and the signatures read from the sources (item E.8) contain no predicate about a large-time translation, a spectral projection $E_{H}((0,\varepsilon))$, or a second transfer-matrix eigenvalue $\lambda_{2}$ (Proposition~10.30). The gap is therefore not a receipt \emph{in the sense of} the reached closure: no such statement occurs among the discharged declarations. \subsection{Type-level evidence from the clustering chain} The six \texttt{Cluster*} modules (Appendix~C) transport an \emph{already-given} E4 from the entrance (\texttt{CompleteOS.\allowbreak E4\_cluster}) to the target kernels; at the level of types (point configurations, Schwartz pairings) the transport loses nothing, and at the level of rates it was never asked to retain anything. This is consistent with the absence of any rate in the chain and with Proposition~10.22: the rate must be added at the entrance, uniformly over all registered label words. \subsection{Tower models and the flux law} The tower model of Proposition~10.28 satisfies the calculus law of (Y2) but not the flux law: $T_{ab}$ is a single rank-two label, while Wick contractions with a base label from another tower vanish, so the shell integral of the flux law and $\partial_{x_{0}}^{b}S$ need not match. The uniform superposition of Proposition~10.7 satisfies both laws because all insertions share one two-point factor. Whether the flux law admits a tower-wise construction remains open; the degree-uniformity conclusion of Proposition~10.26 does not depend on it. \subsection{Final status of the open-identification list} The open-identification list of Remark~10.42 is adjudicated as follows: (1)~drift identity --- \textbf{L}, supported but unread definitions; (2)~\texttt{Fits} discharge --- \textbf{L, PARTIAL} (item E.8.4); (3)~tightness of $\{\mu_{p}^{\psi}\}$ at $t\downarrow0$ --- not needed for the gap, still needed for (O2); (4)~spectral type at fixed volume --- compact kernel gives $\Delta_{N,p}>0$ pointwise; the uniform positivity of $\inf_{p}\Delta_{p}$ is exactly (MG2$''$); (5)~the six \texttt{Cluster*} modules --- not read verbatim; their type-level role is recorded in Section~E.11; (6)~reverse implication gap~$\Rightarrow$~(MG) --- open; (7)~flux law in tower models --- open (Section~E.12). \subsection{Re-adjudication of Appendix~D} \begin{enumerate} \item[\emph{E.14.1}] \emph{D.1(i) maintained.} The penalty is universally quantified and fixed before the family; the blocking ratios are $\operatorname{scaleB}(\hat\beta_{k})$ with $\operatorname{scaleB}(x)=\lceil(\log x)^{2}\rceil$, polylogarithmic; the fitting conditions are eventual (\texttt{stepsize}, \texttt{afford}). \item[\emph{E.14.2}] \emph{D.1(ii) replaced.} The dominance condition (5.2) and the lower bound $\beta_{m}\ge\beta_{\ast}$ are sufficient but not necessary: the remainder is controlled by the schedule (\texttt{Large.slow}/\texttt{Fits.driftSmall}, Lemma~5.9; Remark~5.4). \item[\emph{E.14.3}] \emph{D.1(iii) replaced.} The disclaimer about the $E$/$K$ exponents is replaced by the location of $E$ and $K$ at \texttt{ActivityBounds.KBound}/\texttt{EBound} (item E.8.3; \textbf{L, PARTIAL} for the \texttt{Fits} discharge). \item[\emph{E.14.4}] \emph{D.1(iv) withdrawn and replaced.} The penalty-locus claim is replaced by item E.8.1: the penalty enters the clock kernel, not the gauge coupling. \item[\emph{E.14.5}] \emph{D.2/D.3 revised.} The verdict of Appendix~D is ``Case-1-compatible, with the drift identification recorded as \textbf{L}''; the abstract and the verdict of this paper state the simplification of the bridge input (Theorem~E.19) rather than a feasibility case. \end{enumerate} \subsection{Boundary declaration} \textbf{A positive mass gap is not proved.} The single residual predicate is (MG): a volume-uniform, label-uniform exponential rate in E4 (Proposition~10.34). Everything else in the gap problem is either proved (the bridges, the equivalences, the insufficiency results), explicitly reduced to (MG), or recorded above as an open identification point with its evidential status. \begin{thebibliography}{99} \bibitem[JW]{JW} A.~Jaffe and E.~Witten, \textit{Quantum Yang--Mills theory}, Clay Mathematics Institute Millennium Prize Problem official description, 2000; also in \textit{The Millennium Prize Problems}, Clay Math.\ Inst., Cambridge, MA, 2006, pp.~129--152. \bibitem[OS1]{OS1} K.~Osterwalder and R.~Schrader, \textit{Axioms for Euclidean Green's functions}, Comm.\ Math.\ Phys.\ \textbf{31} (1973), 83--112. \bibitem[OS2]{OS2} K.~Osterwalder and R.~Schrader, \textit{Axioms for Euclidean Green's functions. II}, Comm.\ Math.\ Phys.\ \textbf{42} (1975), 281--305. \bibitem[Wilson]{Wilson} K.~G.~Wilson, \textit{Confinement of quarks}, Phys.\ Rev.\ D \textbf{10} (1974), 2445--2459. \bibitem[Balaban]{Balaban} T.~Balaban, \textit{Renormalization group methods for non-abelian gauge theories}, Harvard University preprint HUTP-84/B30, 1984; see also \textit{Renormalization group approach to lattice gauge field theories}, Comm.\ Math.\ Phys.\ \textbf{95} (1984), 17--40. \bibitem[Seiler]{Seiler} E.~Seiler, \textit{Gauge Theories as a Problem of Constructive Quantum Field Theory and Statistical Mechanics}, Lecture Notes in Physics \textbf{159}, Springer-Verlag, Berlin, 1982. \bibitem[GJ]{GJ} J.~Glimm and A.~Jaffe, \textit{Quantum Physics: A Functional Integral Point of View}, 2nd ed., Springer-Verlag, New York, 1987. \bibitem[DMY]{DMY} Formalization of the Osterwalder--Schrader axioms and growth conditions referenced in [YMRepo] as \textup{[DMY]}; full bibliographic attribution is deferred to the later release of the Yang--Mills group, per [YMRepo], Part~II. \bibitem[OpenAIRepo]{OpenAIRepo} OpenAI, \textit{OpenAI Yang--Mills construction: full Lean code submitted October 28, 2026}, repository \texttt{yang-mills-group/OpenAI-YM-20261028}, \url{https://github.com/yang-mills-group/OpenAI-YM-20261028}. \bibitem[YMRepo]{YMRepo} Yang--Mills Shepherding Committee (M.~R.~Douglas, D.~J.~Gross (chair), A.~Jaffe, I.~R.~Klebanov, N.~Nekrasov), \textit{Validation of the OpenAI Yang--Mills construction}, release of November 10, 2026, repository \texttt{yang-mills-group/YM}, \url{https://github.com/yang-mills-group/YM}. \bibitem[OpenAIBlog]{OpenAIBlog} OpenAI, \textit{Building scientific understanding with academic communities}, announcement of October 26, 2026, \url{https://openai.com/index/building-scientific-understanding-with-academic-communities/}. \bibitem[KitP3]{KitP3} Yang--Mills Shepherding Committee, \textit{3rd Statement by the Shepherding Committee: Proof by OpenAI of the existence of Quantum Yang--Mills Theory}, November 9, 2026, KITP, \url{https://www.kitp.ucsb.edu/sites/default/files/3rd-Statement-About-Quantum-YangMills-Theory_Nov-9.pdf}. \bibitem[OP]{OP} H.~Osborn and A.~C.~Petkou, \textit{Implications of conformal invariance in field theories for general dimensions}, Ann.\ Phys.\ \textbf{231} (1994), 311--362, arXiv:hep-th/9307010. \bibitem[ZC]{ZC} M.~F.~Zoller and K.~G.~Chetyrkin, \textit{OPE of the energy-momentum tensor correlator in massless QCD}, JHEP \textbf{12} (2012), 119, arXiv:1209.1516. \bibitem[BM]{BM} M.~Bochicchio and S.~P.~Muscinelli, \textit{Ultraviolet asymptotics of glueball propagators}, JHEP \textbf{08} (2013), 064, arXiv:1304.6409. \bibitem[GW]{GW} D.~J.~Gross and F.~Wilczek, \textit{Ultraviolet behavior of non-abelian gauge theories}, Phys.\ Rev.\ Lett.\ \textbf{30} (1973), 1343--1346. \bibitem[P]{P} H.~D.~Politzer, \textit{Reliable perturbative results for strong interactions?}, Phys.\ Rev.\ Lett.\ \textbf{30} (1973), 1346--1349. \bibitem[LGT]{LGT} Lattice gauge theory formalization, repository \texttt{mrdouglasny/lgt}, revision \texttt{b8793ccf6a51e00e9e2b1685ba191b8626e37137}. \bibitem[Comparator]{Comparator} Comparator: statement-and-definition comparison and kernel rechecking tool for Lean developments, pinned upstream revisions as recorded in [YMRepo], \texttt{evidence/} records of the October 30, November 3, and November 5, 2026 runs. \bibitem[OS78]{OS78} K.~Osterwalder and E.~Seiler, \textit{Gauge field theories on the lattice}, Ann.\ Physics \textbf{110} (1978), 440--471. \bibitem[GJS]{GJS} J.~Glimm, A.~Jaffe, and T.~Spencer, \textit{The Wightman axioms and particle structure in the $P(\varphi)_{2}$ quantum field model}, Ann.\ of Math.\ (2) \textbf{100} (1974), 585--632. \bibitem[OS78]{OS78} K.~Osterwalder and E.~Seiler, \textit{Gauge field theories on the lattice}, Ann.\ Physics \textbf{110} (1978), 440--471. \bibitem[GJS]{GJS} J.~Glimm, A.~Jaffe, and T.~Spencer, \textit{The Wightman axioms and particle structure in the $P(\varphi)_{2}$ quantum field model}, Ann.\ of Math.\ (2) \textbf{100} (1974), 585--632. \end{thebibliography} \end{document}