XYZ-symmetric SiN/SiO2 y-atom air-strip Q/V optimizer

Worker: not active · stage: projected_q_over_v_continuation · status: stopped_by_request_final_validation · checkpoint age: 5657.3 min

This paused campaign starts from density 0.5 in a 100 nm SiN design layer embedded between 300 nm SiO2 layers in a 16 µm × 2 µm design. A centered y-oriented atom lies in an immutable one-micron vacuum strip through every material layer. Full x/y/z symmetry leaves one octant and gives the unfolded bright source pair unit total power. Raw Adam first minimizes the fixed absolute temporal error. At step 50, a trustworthy Q≥100 activates the pole handoff. Stage B uses adaptive-trust L-BFGS-B to maximize only log(Q_harmonic)-log(V_atom/(lambda_pole/n_air)^3) over every raw topology pixel, the full SiN thickness, and the SiO2 thickness on each side. The thickness map enforces t_SiO2 >= t_SiN + 0.050 µm for every proposal. After the raw-pixel and tangent-binarity directions plateaued, Stage C switched to conventional unfiltered tanh-projection continuation. It optimizes only Q/V for six trials at fixed beta, raises beta by 0.5, and repeats from beta=1 to 32. Near-binary bulk hands off by exact replay to a topology-preserving cubic-spline level-set boundary, which then resumes direct Q/V optimization. There is no spatial filter or binarity term in the scalar objective.

quantity	current
temporal Adam updates	50 / at least 50
pole accepted updates / trials	528 / 1611
Q	23337.9
best recorded Q (trust gate shown above)	23408.9
field / energy Q	23337.9 / 23480.3
Q-estimator ratio	1.0061
pole wavelength	837.526 nm
source-center wavelength	836.564 nm
field-monitor wavelength	836.564 nm
design-region slab thickness	0.293695 µm
Stage-B thickness coordinate / range	True / 0.050–0.500 µm
SiO2 thickness per side	0.343712 µm
Stage-B SiO2 coordinate / maximum	True / 1.000 µm
stack-thickness constraint	t_SiO2 − t_SiN ≥ 0.050 µm
SiO2 full width	2 µm
joint-stage SiO2 width coordinate / range	False / 2.000–2.000 µm
physical mode volume	3.89716 µm³
normalized mode volume	6.63366 (λ_pole/n_air)³
Q / normalized atom volume	3528.81
Purcell from Q/V	268.157
fitted atom-pole amplitude at source clear	0.0641632
absolute atom-trace peak / handoff	0.122394
source spectral power at pole / actual carrier	0.844164
reciprocal guided-Purcell proxy	0.030625
guided-Purcell proxy / handoff	3902.89
Stage-B volume treatment	none — log Q / normalized atom volume is optimized
three-window V span	0.000118202
active Q/V run time / window	— / — ps
gradient backend / storage	unknown / unresolved
structural backend decision	not recorded
exact reversible PML tape size	— GB
frozen measured value+gradient speedup	—x
Q/V fit start / nominal lifetime per window	— ps / — τ_amplitude
precision-triggered tape growths	0
final ringdown field-power / energy fraction	— / —
finite-step adjoint direction verified	False
ideal temporal fidelity	0.483419
active objective / value	log Q / normalized atom volume / 8.16871
stage plateau / reason	False / collecting_optimizer_memory
plateau samples / evidence window	0 / 0
objective slope / detectable slope	— / —
trust-floor restarts since progress	0 / 2
practical fixed-beta Q/V gain over one memory	—% / 1% limit
practical plateau samples / required	0 / 11
practical trust radius still expanding	False
healthy / cumulatively pruned curvature pairs	0 / 0
L-BFGS / current-gradient predicted-ascent ratio	—
replay-triggered curvature recoveries	0
maximum last density motion	0.0886365
trust radius	0.00961084
binarity score (1 is binary)	0.951442
trainable gray fraction (0.05<rho<0.95)	6.9299%
archived tangent-stage Q/V floor	694.152
projection beta / eta	20 / 0.5
fixed-beta accepted Q/V updates / transition	1 / objective plateau
material parameterization	unfiltered_centered_tanh_projected_50nm_pixels
hard atom access	1.000 µm full-y air strip, immutable
pixel / grid	50 nm / [181, 35, 45]
non-PML z span / material-stack-to-PML air gap	3.50 / 1.26 µm
z PML	0.50 µm per side
z-padding convergence audit	not_run
1.50→3.40 µm replay: residual / field-Q / energy-Q change	nan% / nan% / nan%
binarization / material projection / filter	True / True / off
physical waveguide escape eta	not part of the active objective

Q, V, objective, and optimizer history

The Q axis is linear. A pole is not trusted until the centered Ey and total energy decay estimates agree. Mode volume uses the geometric mean of their three same-pole windows and the same fitted pole wavelength; the local permittivity is exactly air by the hard geometry constraint. The predecessor vacuum-clad branch is preserved as a qualified success at Q=1,978, normalized atom volume 4.210, Q/V=469.9, and Purcell=35.71. Its frozen atom-volume controller reached a trust-floor plateau; this campaign is a clean three-material restart and does not inherit its geometry.

Ideal and measured atom response

The target has fixed amplitude and phase for Q=10,000, beta one, V=0.1 µm³, and perfect Ey-polarized overlap. It is not rescaled to fit the simulation.

Geometry and fields

Downloadable layout

Layout status: provisional density-0.5-contour GDS preview. The ZIP includes the GDS, JSON stack/metric manifest, and README. Before the spline handoff the GDS is intentionally marked as a thresholded preview; after handoff it is the fabrication-oriented smooth boundary representation. GDS is planar, so use the manifest for SiN and SiO2 thicknesses.

GDS bundle

Current GDS

Layout manifest

Exact campaign methodology