One-sided SiN atom-hole temporal-to-pole optimizer

Worker: not active · stage: pole_q_volume_constrained · status: paused_resumable_q1767_atom_volume_constrained · checkpoint age: 4307.9 min

This paused campaign launches the fundamental TM-like mode of one positive-x waveguide into an 8 µm × 8 µm, 300 nm-thick SiN design. A centered z-oriented atom occupies an immutable one-micron through-slab air hole. The first 150 or more raw Adam updates minimize the absolute vector temporal error; a trustworthy Q≥100 then activates frequency-free, adaptive-trust L-BFGS-B on the fitted pole lifetime. Its Q directions are projected against the atom-mode volume gradient, and every accepted geometry must remain below the frozen pre-Q volume ceiling. No filter, material projection, or binarization is active.

quantity	current
temporal Adam updates	150 / at least 150
pole accepted updates / trials	211 / 348
Q	1767.37
best trusted Q	1767.37
field / energy Q	1768.15 / 1767.37
Q-estimator ratio	1.00044
pole wavelength	773.456 nm
source-center wavelength	773.502 nm
field-monitor wavelength	773.502 nm
design-region slab thickness	0.3 µm
Stage-B thickness coordinate / range	False / 0.300–0.300 µm
physical mode volume	1.94042 µm³
normalized mode volume	4.19363 (λ_pole/n_air)³
Q / normalized atom volume	—
Purcell from Q/V	—
fitted atom-pole amplitude at source clear	—
absolute atom-trace peak / handoff	—
source spectral power at pole / actual carrier	—
reciprocal guided-Purcell proxy	—
guided-Purcell proxy / handoff	—
Stage-B volume treatment	2.0675 µm³
three-window V span	0.000125812
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	0.553248
active objective / value	log Q / normalized atom volume / 7.47747
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	0.00029999
trust radius	0.00045
binarity score (1 is binary)	—
trainable gray fraction (0.05<rho<0.95)	—%
archived tangent-stage Q/V floor	—
projection beta / eta	— / 0.5
fixed-beta accepted Q/V updates / transition	0 / objective plateau
material parameterization	unfiltered_direct_50nm_pixels
hard atom access	1.000 µm minimum air hole, immutable
pixel / grid	50 nm / [202, 95, 25]
non-PML z span / material-stack-to-PML air gap	1.50 / 0.60 µm
z PML	0.50 µm per side
z-padding convergence audit	passed_keep_1p5um_interior
1.50→3.40 µm replay: residual / field-Q / energy-Q change	7.5144e-05% / -0.004968% / -0.7036%
binarization / material projection / filter	False / False / 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 Ez 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 rejected Q-only branch reached Q=3,619 while normalized V deteriorated to 17.85; it is archived and is not part of this curve.

Ideal and measured atom response

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

Geometry and fields

Exact campaign methodology