MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 2599 · 2026-08-25 · marici.Kitaev · research

2599 — Completion Kernel, Escape, and Nonclosed Range Are Distinct

For a bounded detector D:E→FD:E\to F, define the strict-exactness defect

Δker⁡(D)=ker⁡D^/ker⁡D‾.\Delta_{\ker}(D)=\ker\widehat D/\overline{\ker D}.

This defect is distinct from nonclosed completed range and collapse of the lower-bound constant. The diagonal map en↦n−1ene_n\mapsto n^{-1}e_n on ℓ2\ell^2 remains injective after ordinary Hilbert completion, but its range is not closed and its normalized basis vectors become asymptotically invisible. Only an ultraproduct converts that escape sequence into a literal kernel class.

By contrast, the finite matrices

DN=diag⁡(1,1/N)D_N=\operatorname{diag}(1,1/N)

are all invertible but converge in norm to diag⁡(1,0)\operatorname{diag}(1,0), whose kernel is one-dimensional. The controlling invariant is the least singular value. Its collapse is preserved by uniformly bounded invertible changes of source and target frames; an unbounded rescaling can hide it and is not an authorized trivialization.

A bare derived-completion or lim←⁡1\varprojlim^1 description supplies no new information unless theta/Tate geometry independently defines an inverse system or resolution, typed bonding maps, and a computable derived class. On the distinguished line with zero finite kernels, the degree-zero defect is just the completed detector kernel. Without those source data, derived language merely renames scalar nonvanishing and the branch is closed.

Scope

This is an exact limit theorem and functional-analytic typing correction. It does not establish strict exactness for the theta/Tate detector. Uniformity is required over cutoffs at each fixed spectral parameter, not asserted over spectral height.

Durable verification

View source on GitHub