Retired external-programme boundaries

This record replaces five retired programme notes and preserves why their implementations did not establish TNFR emergence. It is an archive of scope boundaries, not a research queue. The removed sources remain recoverable from Git history. Retirement does not prove that every future correspondence is impossible, or that the remaining engine has complete dynamical closure.

Retired programme	Boundary established by the implementation review	Useful work retained elsewhere
Yang-Mills	The selected finite matrix and imposed graph sweeps did not derive an independent gauge dynamics. The bundled vertex-phase connection was pure gauge, with floating cycle-closure residuals. Report classifications encoded missing structure; they did not discover a physical mass gap.	Gauge diagnostics retain their auxiliary coordinate and numerical-residual scope.
Navier-Stokes	A separately supplied fluid solver did not implement the nodal equation's declared engine path. Vector vorticity was not identified with wrapped scalar curvature, and continuum Fourier shell ordering was not interchangeable with normalized graph spectral ordering. Combining these objects did not establish a common dynamics or justify identifying their different limiting questions.	Structural diffusion retains the graph operator, conditional energy balance and separately declared damped-wave comparison.
P versus NP	A supplied antialignment heuristic and exhaustive finite search did not derive autonomous TNFR selection or establish a complexity result. A gradient identity for one admitted diffusion channel did not extend to that heuristic or to every operator.	The variational boundary and operator contracts retain the distinction between conditional descent, named events and selection.
Birch-Swinnerton-Dyer	Known rank labels and external arithmetic point counts were inputs. Renaming them pressure and accumulated coherence did not supply a nodal evolution, predict the labels independently or create a physical measurement bridge.	The arithmetic implementation remains a separately specified mathematical model; shared diagnostic names do not identify its dynamics with graph evolution.
Hodge	The example supplied its own simplicial complex. Nodewise unsigned phase gradients and wrapped node curvature did not furnish its oriented edge and face cochains. The finite construction therefore supplied no derivation of that complex from the tetrad.	Phase-cycle geometry retains explicitly defined graph observations; minimal structural degrees records the limits of reconstruction from lossy fields.

The retained research evidence infrastructure records assumptions, provenance, numerical scope and circularity. These are reusable methods for assessing a stated model, not evidence that any external programme has been recovered. Any renewed correspondence would require a new, explicit bridge and independent evidence; the retired milestone lists do not create work obligations.