Historical derivation — preserved in full
Source: engines/quan_loncar_v1/lineage/phases_01_28/quan_loncar_complete_lineage_phases_01_28/fryett_q_series_phase13/DERIVATION_PHASE13.md
Snapshot: Quan–Lončar complete lineage, phases 1–28. The body below is unabridged.
Phase 13 derivation: separated even-x in-plane degree four¶
Controlled sequence¶
On the corrected Fryett sector (s_x,s_y,s_z)=(+1,-1,+1), hold every cutoff,
contour, quadrature, and thickness setting fixed and compute
- hole Zernike degree 3, waveguide y degree 3;
- hole Zernike degree 3, waveguide y degree 4;
- hole Zernike degree 4, waveguide y degree 4.
This separation is required because the Phase-12 restart changed the physical hole basis and the bare-waveguide Green basis simultaneously.
At thickness degree zero, every real disk mode retains exactly one vector
component in the target (s_y,s_z)=(-1,+1) block. The degree-four radial
family contains five real modes: three cosine modes enter as E_y, while two
sine modes enter as E_x. The reduced dimension grows from 450 to 675.
Result¶
The separated relative changes are
with a net degree-three to degree-four change of -2.453504%.
The waveguide-only overlap is 0.993938 and the lifted hole-basis overlap is 0.999107. The final null-vector coefficient fractions are
Eigenvalue and smallest-singular-value residuals are below 1.7e-11 for the
waveguide row and 2.0e-13 for the joint row. Reciprocity remains below
3.6e-16.
Interpretation¶
The degree-four physical hole correction is already far below the 2% provisional threshold. The bare-waveguide y correction remains slightly above that threshold, so the complete in-plane axis cannot yet be called converged. A separated degree-five row is mandatory.
This behavior does not trigger the abandonment gate. The pole is stable, its overlap is high, the new physical-basis increment is small, and the remaining waveguide correction is in the 2-10% range for which the protocol requires one more order.