Author notes — full detail, auditor-facing

Cipher v11 was the version that closed the long-running self-derivation problem: predict every element's crystal structure parameters from the atomic number Z *alone*, with no external lookup, no NIST cross-references, no fitted parameters.

The path to v11:

v7 introduced the dimensional architecture (2D/3D/4D regimes, 96.9% property matches) but still relied on external cross-references.

v8 added the lattice-resonance mechanism (geometry rings at coord/{2,3}), bringing the cross-scale unification — same mechanism explains crystal structure, magic numbers, and (eventually) galactic framerate.

v9 added the eigenvalue-susceptibility layer (CN-derived) and the snap/tunneling mechanism (pre-snap f-state at 99° vs post-snap |t-state at 109.5° = tunneling). Reached 71.9% exact / 94.8% match-plus-partial, zero regressions vs v8.

v10 experimented with corrections — eigenvalue blends, spherical limit handling, percolation thresholds. Most corrections HURT accuracy (see the no-corrections finding).

v11 consolidated: bicone simultaneous-derivation (f-down for depth + |t-up for spread, simultaneous not sequential), snap at percolation topology, Thomson/Fibonacci hybrid for shell electron seating.

The v11 derivation chain

For an element of atomic number Z:

1. Z → electron shell topology. Z electrons fill seats on the spiral coordinate (the C_potential's internal geometry), seated at Thomson positions on early shells, Fibonacci positions on outer shells. No NIST configuration data used — the seating rule derives from minimizing repulsion under the constraint that the spiral has discrete seat density.

2. Shell topology → coordination number. The outer-shell seating determines the local coordination geometry (CN). v11 derives CN from the spiral seat count, not from external sources.

3. CN → eigenvalue spectrum. Each coordination geometry has its own eigenvalue spectrum (number of distinct values in its symmetry group's character table). BCC=9, HCP=7, FCC=5, Diamond=1. Eigenvalue count = susceptibility = how readily the geometry resonates.

4. Eigenvalue spectrum + snap state → unfolding angle. The bicone simultaneous read produces a unfolding angle. Pre-snap (percolation topology < 1.29) gives the f-state at ~99°; post-snap gives the |t-state at ~109.5°. Snap = tunneling.

5. Unfolding angle + spiral coordinate → predicted crystal structure parameter (lattice ratio c/a, melting temperature trend, conductor class, etc.).

Results

• Scale 2 sweep (107 elements): median angular error 1.0°,

98.1% A+P (acceptable or partial match), 2 outright misses.

• Self-contained: every prediction derives from Z. No NIST lookups,

no external configuration data.

• Non-Euclidean discovery: the 2–3° offset between cipher and

crystallographic-measurement values is *real curvature*, not error. Crystallography uses Euclidean projection; the cipher reads true non-Euclidean geometry. This was an unexpected finding — first X-ray-less structural measurement.

Why v11 is now marked superseded

After v11, the next experimental cycle ran v11-with-various-corrections vs pure v12 (no corrections, just clean cycle-aware terrain). The result was unambiguous:

• v11 corrected: 71/107
• v12 clean: 82/107

Every correction layer added to v11 HURT accuracy. v12 returned to pure terrain (no snap layer, no eigenvalue correction, no shell logic) and beat v11 on the same bench. See the corrections-hurt-accuracy log for the full comparison.

The lesson v11 taught: the framework is *capable* of complete self-derivation. v12 took the v11 result and stripped the corrections, proving the terrain alone carries the predictive power.

v11 stays preserved, not deleted. The 98.1% A+P result at median 1.0° is the historical high-water mark for the "complete derivation + corrections" approach. v12 beat it without corrections, but v11 demonstrated the self-derivation chain was sound.