Pure Yang-Mills Theory on Lattice The Mass Gap Problem (Clay) , On the Proof of the Yang Mills Mass Gap: Gradient Flow, GEVP Variational Construction and Spectral Localization

Description of the Manuscript Title: Proof of the Yang-Mills Mass Gap: Gradient Flow, GEVP Variational Construction and Spectral Localization Author: Rodrigue Pekar (independent researcher) This manuscript addresses one of the seven Clay Mathematics Institute Millennium Prize Problems: proving that pure Yang-Mills gauge theory in four dimensions possesses a strictly positive mass gap. The work is set on a four-dimensional Euclidean hypercubic lattice with gauge group SU(N) and Wilson action. The central technical innovation is the systematic use of the Wilson gradient flow at fixed positive flow time t > 0 as a non-perturbative UV regulator, combined with a Generalized Eigenvalue Problem (GEVP) variational construction to extract spectral information. The proof is organized along a primary architecture (Architecture A) proceeding in seven steps. First, the transfer matrix on any finite lattice is shown to be a strictly positive trace-class operator, so the Perron-Frobenius theorem (Kreĭn-Rutman) guarantees a spectral gap for each fixed lattice. Second, a GEVP built from gradient-flow-smoothed Wilson loop operators of distinct sizes provides a constructive lower bound on this gap, uniform across the UV window, provided the overlap coefficients between the smoothed operators and the low-lying eigenstates remain bounded away from zero. Third and this is the heart of the paper a six-step argument (Theorem 13.6, called Theorem U) establishes this uniform positivity of the GEVP overlap: (U1) uniform operator bounds from the unitarity of Wilson loops, (U2) finite-dimensional Hörmander hypoellipticity on the compact lattice configuration space SU(N)^{4|Λ|}, (U3) Poisson-Jacobi theta-function bounds for volume uniformity, (U4) entry-wise Symanzik convergence of individual correlation matrix entries which breaks a circularity with the spectral gap (U5) tree-level non-degeneracy from the linear independence of Wilson loop form factors at distinct momentum scales, promoted to a rigorous bound via the analyticity of lattice correlators established by Balaban's renormalization group program, and (U6) spectral assembly using Weyl's perturbation theorem and Lüscher-Wolff error control from the GEVP formalism. The remaining steps transport the lattice gap to the continuum infinite-volume limit: Poisson-Jacobi exponential bounds control finite-volume corrections, Symanzik O(a²) improvement at positive flow time controls lattice artifacts, a step-scaling procedure propagates the bound through the renormalization group trajectory, and Osterwalder-Schrader reconstruction yields a Wightman quantum field theory with mass gap m₁ > 0. A second, independent proof architecture the Localization Route is developed in detail as a conditional alternative. It uses a Fractal Localization Principle to construct self-similar partitions with arbitrarily small IMS cost, local coercivity from Poincaré-Korn inequalities on each patch, a multi-type Galton-Watson branching criterion ensuring sub-criticality of inter-patch interactions, and a Spectral Extinction Theorem that globalizes the gap. This architecture is conditional on the uniformity in the lattice spacing of the lattice Poincaré constant, which remains an open technical point. The manuscript maintains an explicit rigor table classifying every intermediate result as proved, proved (variational), or sketch. Some auxiliary results in the proof chain notably the Gaussian decay of the discrete heat kernel, the relative operator bound for the Kato-Rellich perturbation argument, and the local gap lower bound for the Localization Route are classified as sketches. The key uniformity condition (Theorem 13.6) is proved analytically using the Balaban cluster expansion for non-perturbative control, and the paper carefully distinguishes what Balaban's program establishes (UV stability, analyticity of correlators in the coupling) from what it does not (exponential decay, mass gap). The mathematical tools deployed across the 15 chapters include transfer matrix spectral theory, Perron-Frobenius and Kreĭn-Rutman positivity, the GEVP formalism of Lüscher and Wolff, Symanzik effective theory, Poisson summation and Jacobi theta functions, Kato-Rellich analytic perturbation theory, IMS localization, Hörmander hypoellipticity, Peter-Weyl decomposition on compact groups, Balaban's constructive renormalization group, and Osterwalder-Schrader reconstruction. Pistes