The Depth of Almost Nothing: Near-Null Geometry in a Prime–Archimedean System

The Depth of Almost Nothing: Near-Null Geometry in a Prime-Archimedean System This document studies a mathematical structure built from two operator channels on the same finite-dimensional space: an arithmetic "prime channel" p, assembled from a finite sum over prime powers, and an analytic "archimedean channel" a, assembled from an integral over a fixed density. Compared through a common energy H0 = p^2 + a^2, the residual R0 = p + a develops an extreme near-null geometry on a distinguished deep spectral subspace D_r: eigenvalue depths of the associated operator T0 reach 10^-100 and beyond, produced by two layers of cancellation sitting on top of an exact common integral architecture shared by both channels -- a residual layer that is fully established, and a scalar layer that an unfitted, pre-registered audit found to be asymmetric between the two channels rather than symmetric. The document separates what is proved from what is numerically observed from what remains open, using a consistent five-level marker system throughout (exact identity, derived asymptotic result, computational regularity, open problem, eliminated), and records at comparable length eight distinct plausible mechanisms for the near-null geometry that were formulated precisely, tested computationally, and eliminated -- including a rank-5 "universal core" construction retracted once its defining measure is shown to diverge. Two further exact structural facts are established directly from the channels' own integral architecture, independent of that audit: D_r can be located a priori as the bottom eigenspace of T0 alone, agreeing with its original definition to a fraction of a degree at every reference point; and R0 itself is shown to be exactly a scaled Loewner matrix, R0 = 2D L_f D, of a single function f built from the two channels' defining sums, with an exact displacement identity of rank at most two. A pre-registered kill test on that identity eliminates low polynomial complexity of f as the mechanism behind the near-null block -- the residual stays ten to sixty in magnitude against target eigenvalues of 10^-38 to 10^-90, a gap of forty to ninety orders, with no cancellation between channels in this specific metric. Why R0 develops this near-null geometry at all remains the central open problem. A second, independent result concerns the near-null projector once it exists: fixing a rank and a trajectory of configurations, the response of the projector to smooth trajectory deformations is shown to admit an exact perturbative expansion, Q(eps) = 1 + A(G) eps + B(G) eps^2 + O(eps^3), with both coefficients derived -- not fitted -- from the exact projector family and confirmed against independent numerical measurement across multiple geometries, including a sign reversal of the second-order curvature B that the same formula predicts correctly in both directions. A further addendum shows the trajectory's own base point is an exact algebraic coherence point -- an affine blend of the residual at two reference configurations -- and proves, from the general fact that T0 is a Gram form A^T A, that any near-null amplitude there is structurally forbidden a linear correction: the quadratic response observed away from that point is a provable consequence of this structure, not merely a fitted regularity.