---
title: Optimization campaign atlas
---

# Optimization campaign atlas

Every FDTDX cavity and atom-interface optimization in one visual directory.
Cards pair the actual optimized geometry with its simulated field, so related
experiments remain distinguishable even when their internal campaign names are
similar. **Live** pages continue to refresh; **frozen**, **paused**, and
**stopped** pages preserve the last trustworthy checkpoint and the reason the
campaign ended.

```{raw} html
<div class="campaign-gallery">
  <a id="nanobeam-waveform-atlas-card" class="campaign-card campaign-paused" href="nanobeam_waveform_scratch_dashboard.html" data-atlas-live-state="loading">
    <div class="campaign-card-head"><span id="nanobeam-waveform-atlas-status" aria-live="polite">Checking live state · adaptive carrier-transported waveform · single-GPU serial</span><h2>8.4 µm silicon nanobeam · adaptive waveform frontier</h2></div>
    <div class="campaign-card-images campaign-card-images-three"><img src="../_static/generated/nanobeam_waveform_scratch_geometry.png" alt="Frozen plateau, current frontier, and best feasible waveform topology"><img src="../_static/generated/nanobeam_waveform_scratch_fields.png" alt="Latest same-lineage checkpoint-labeled forward Ey field audit"><img src="../_static/generated/nanobeam_waveform_frontier_progress.png" alt="Adaptive waveform target, Q, loading, wavelength, and mode-identity progress"></div>
    <p>The live frontier starts from the immutable Q≈30k topology-free plateau and uses no held-out reference geometry. Its carrier-transported temporal target stays at twice the measured Q and re-anchors automatically; loading, mode identity, and broad wavelength membership are buffered constraints. Exact replay accepts every material change. The predecessor and all failed oversized escapes remain frozen evidence.</p>
  </a>
  <script>
  (() => {
    const card = document.getElementById("nanobeam-waveform-atlas-card");
    const badge = document.getElementById("nanobeam-waveform-atlas-status");
    if (!card || !badge) return;
    const refreshNanobeamStatus = () => fetch("../_static/generated/nanobeam_waveform_scratch_status.json", {cache: "no-store"})
      .then((response) => {
        if (!response.ok) throw new Error(`status ${response.status}`);
        return response.json();
      })
      .then((status) => {
        const controller = String(status?.optimizer_status || "unknown");
        const controllerAtRest = controller === "complete" || controller === "stopped"
          || controller === "ready_for_projection_continuation" || controller.startsWith("blocked_")
          || controller.startsWith("needs_");
        const serviceLive = status?.live?.optimizer_service_active === true
          && status?.terminal !== true && !controllerAtRest;
        const state = controller === "complete" ? "Complete"
          : controller.startsWith("blocked_") ? "Blocked"
          : controller === "stopped" ? "Stopped"
          : controller === "ready_for_projection_continuation" ? "Handoff ready"
          : controller.startsWith("needs_") ? "Audit needed"
          : serviceLive ? "Live" : "Paused";
        const checkpoint = Number.isInteger(status?.checkpoint_sequence)
          ? `numerical checkpoint ${status.checkpoint_sequence}` : "checkpoint unavailable";
        const measuredQ = Number(status?.current_progress?.current_carrier_tracked_q);
        const qLabel = Number.isFinite(measuredQ) && measuredQ > 0
          ? `Q ${Math.round(measuredQ).toLocaleString()}` : "Q pending";
        const audit = status?.field_trace_audit;
        const auditLag = Number(audit?.active_checkpoint_lag);
        const auditLabel = audit?.stale ? "field audit withheld as stale"
          : audit?.available && Number.isInteger(audit?.source_checkpoint_sequence)
            ? `field audit checkpoint ${audit.source_checkpoint_sequence}${
                Number.isInteger(auditLag) && auditLag > 0 ? ` (lag ${auditLag})` : ""
              }` : "field audit pending";
        card.classList.toggle("campaign-live", serviceLive);
        card.classList.toggle("campaign-paused", !serviceLive);
        card.dataset.atlasLiveState = serviceLive ? "live" : "paused";
        badge.textContent = `${state} · adaptive carrier-transported waveform · ${qLabel} · single-GPU serial · ${checkpoint} · ${auditLabel} · Q≈30k predecessor frozen`;
      })
      .catch(() => {
        card.dataset.atlasLiveState = "status-unavailable";
      });
    refreshNanobeamStatus();
    window.setInterval(refreshNanobeamStatus, 60000);
  })();
  </script>

  <a class="campaign-card campaign-live" href="atom_gap_2d_q_over_a_dashboard.html">
    <div class="campaign-card-head"><span>Live · true 2D x-z test · symmetric TE input</span><h2>20 µm SiN/SiO2 atom gap · temporal → Q/Aeff</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_gap_2d_q_over_a_geometry.png" alt="Full 20 micron x-z atom-gap material cross-section"><img src="../_static/generated/atom_gap_2d_q_over_a_field.png" alt="Pole-filtered full-domain Ey field of the 2D atom-gap cavity"></div>
    <p>A y-invariant 300 nm patterned Si3N4 core inside a fixed 1 µm SiO2 stack, interrupted by a hard 1 µm air gap. The density-0.5 start first matches the fixed atom response to a solved inward TE-mode pulse; after a trustworthy atom pole and measured slowdown it hands off to replay-gated same-pole Q/Aeff. The field panel is pole-frequency filtered.</p>
  </a>

  <a class="campaign-card campaign-live" href="atom_strip_cladded_tm_q_dashboard.html">
    <div class="campaign-card-head"><span>Live · fresh SiO2-slab restart · Ey</span><h2>8 × 8 µm SiN/SiO2 atom slot · y dipole</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_strip_cladded_tm_q_geometry.png" alt="Initial and live 8 by 8 micron TM atom-slot geometry"><img src="../_static/generated/atom_strip_cladded_tm_q_fields.png" alt="Live atom-slot Ez field profiles"></div>
    <p>A density-0.5, component-aware Yee campaign with a hard one-micron atom strip. The SiO2 slab fills the complete x–y plane through the PMLs, while an explicit air cutout removes the entire strip. It discovers the absolute Ey temporal response, then moves through pixel-only Q, joint SiN/SiO2-thickness Q, gradual fixed-beta Q, and topology-free smooth-boundary Q. The failed z-dipole vacuum branch is preserved separately.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="composite_waveguide_gap_sweep_dashboard.html">
    <div class="campaign-card-head"><span>Parameter sweep · complete · validated</span><h2>Composite Si₃N₄/SiO₂ waveguide · air-gap transmission</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/composite_waveguide_gap_sweep_schematic.png" alt="Coaxial composite square-waveguide air-gap geometry"><img src="../_static/generated/composite_waveguide_gap_sweep_comparison.png" alt="Composite-waveguide insertion loss versus air gap"></div>
    <p>A fixed-port 780 nm FDTD sweep measures same-mode transmission, reflection, and uncollected power across 1–2 µm air gaps. It includes 150 nm Si₃N₄ cores in 1 and 2 µm square SiO₂ guides, a 100 nm / 3 µm slice, and SiO₂-only 1 and 2 µm controls. The stored tensors are ready to expand into a full three-parameter seed search.</p>
  </a>

  <a class="campaign-card campaign-prepared" href="atom_strip_cladded_air_start_width_qv_dashboard.html">
    <div class="campaign-card-head"><span>Prepared · not started · air-start Yee</span><h2>16 × 2 µm SiN/SiO2 atom strip · variable oxide guide width</h2></div>
    <p>A clean successor to the live component-aware campaign. The SiN topology starts at density zero; after temporal discovery and pixel-only Q/V, SiN thickness, exact-index-1.44 SiO2 thickness, and analytic SiO2 width optimize jointly. Beta acts only on SiN, so oxide and nitride cannot collapse into an ambiguous ternary density.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="atom_strip_cladded_qv_yee_restart_dashboard.html">
    <div class="campaign-card-head"><span>Unqualified success · Tidy3D-validated · update 2562</span><h2>16 × 2 µm SiN/SiO2 atom strip · component-aware Yee landmark</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_strip_cladded_qv_yee_restart_geometry.png" alt="Final component-aware Yee atom-strip smooth-boundary geometry"><img src="../_static/generated/atom_strip_cladded_qv_yee_restart_fields.png" alt="Final component-aware Yee atom-strip Ey field profiles"></div>
    <p>Frozen at accepted update 2562 with the one-micron atom-access strip intact. The exact final geometry gives Tidy3D Q=3,894, V=5.993 (λ/n_air)³, Q/V=649.8, Purcell=49.38, and λ=826.21 nm; the same checkpoint gives FDTDX Q=5,807, V=4.612, and Q/V=1,259. The complete optimizer state, smooth boundary, validation record, and immutable success archive are preserved for later continuation.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="atom_strip_cladded_qv_dashboard.html">
    <div class="campaign-card-head"><span>Stopped · legacy scalar-material predecessor</span><h2>16 × 2 µm, 100 nm SiN in SiO2 · xyz-symmetric y atom</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_strip_cladded_qv_geometry.png" alt="XYZ-symmetric cladded SiN atom-strip initial and optimized geometry"><img src="../_static/generated/atom_strip_cladded_qv_fields.png" alt="XYZ-symmetric cladded atom-strip Ey field profiles"></div>
    <p>The original density-0.5 campaign is preserved at accepted update 583 for comparison. Its scalar-material FDTDX result motivated the independent Tidy3D validation and the component-aware Yee restart above; it is no longer the running optimizer.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="atom_strip_temporal_collection_dashboard.html">
    <div class="campaign-card-head"><span>Qualified success · paused and resumable</span><h2>16 × 2 × 0.5–0.7 µm SiN atom strip · xyz-symmetric y atom</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_strip_temporal_collection_geometry.png" alt="XYZ-symmetric SiN atom-strip initial and optimized geometry"><img src="../_static/generated/atom_strip_temporal_collection_fields.png" alt="XYZ-symmetric atom-strip Ey field profiles"></div>
    <p>Paused at trustworthy Q=1,978, V=4.210 λ³, Q/V=469.9, and atom-local Purcell=35.71 after 118 accepted pole updates. Its frozen-volume Q controller ultimately reached a trust-floor plateau; the exact full-state checkpoint and immutable qualified-success archive are preserved.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="atom_hole_xyz_temporal_q_dashboard.html">
    <div class="campaign-card-head"><span>Qualified success · paused and resumable</span><h2>12 × 4 µm SiN atom hole · xyz-symmetric y atom</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_hole_xyz_temporal_q_geometry.png" alt="XYZ-symmetric SiN atom-hole initial and optimized geometry"><img src="../_static/generated/atom_hole_xyz_temporal_q_fields.png" alt="XYZ-symmetric atom-hole Ey field profiles"></div>
    <p>Paused at trustworthy Q=10,057, V/λ³=61.24, atom-local Purcell=12.49, and 48.70× the handoff reciprocal guided-Purcell proxy; best observed Q was 10,499. The useful coupling gain is a qualified success, while the dielectric-centered mode and excessive Q requirement motivate the air-strip successor. Its exact optimizer state and hash-addressed checkpoint are preserved.</p>
  </a>

  <a class="campaign-card campaign-paused" href="atom_hole_temporal_q_dashboard.html">
    <div class="campaign-card-head"><span>Paused · resumable at Q=1,767</span><h2>300 nm SiN z-atom hole · one-sided V-constrained Q</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_hole_temporal_q_geometry.png" alt="One-sided SiN atom-hole initial and optimized geometry"><img src="../_static/generated/atom_hole_temporal_q_fields.png" alt="One-sided atom-hole Ez field profiles"></div>
    <p>The corrected one-sided branch reached Q=1,767 with physical atom volume 1.940 µm³ below its frozen 2.068 µm³ ceiling. Its exact optimizer state, cached trust controller, and immutable hash-addressed checkpoint are preserved for later continuation.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fryett_shifted_q_dashboard.html">
    <div class="campaign-card-head"><span>Landmark success · paused at Q=138,284</span><h2>Fryett 16 µm nanobeam · extended-mirror lifetime</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_shifted_q_geometry.png" alt="Pole-directed Fryett seed and current geometry"><img src="../_static/generated/fryett_shifted_q_fields.png" alt="Pole-directed Fryett tracked cavity field"></div>
    <p>The successful temporal-to-pole sequence reached Q=6,322 in 8 µm, inferred and repeated its learned 256.589 nm outer mirror cell through a 16 µm fork, and climbed from an independently audited Q=42,361 seed to Q=138,284. The exact unbinarized 25 nm checkpoint is preserved for later binarization and smooth-boundary continuation.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fryett_symmetric_dipole_mse_dashboard.html">
    <div class="campaign-card-head"><span>Stopped after plateau audit</span><h2>Fryett nanobeam · symmetric ports → central dipole</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_symmetric_dipole_mse_geometry.png" alt="Symmetric direct-pixel port-to-dipole geometry"><img src="../_static/generated/fryett_symmetric_dipole_mse_response.png" alt="Fixed finite-Q ideal and simulated dipole response"></div>
    <p>An x/y/z-symmetric unit-power bright TE port pair drives a central y dipole. The sole objective is full-vector temporal MSE against a fixed 780 nm, Q=10,000, V=0.1 um3 analytical response. Each raw Adam update is one 2 ps forward/adjoint pair over 3,200 independent 25 nm density pixels; Q, V, and two-port beta are telemetry only.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fryett_modal_source_joint_adam_dashboard.html">
    <div class="campaign-card-head"><span>Stopped at update 4</span><h2>Fryett nanobeam · pulsed midpoint-guide/source Adam</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_modal_source_joint_adam_geometry.png" alt="Prepared simultaneous Adam geometry"><img src="../_static/generated/fryett_modal_source_joint_adam_fields.png" alt="Prepared simultaneous Adam field placeholder"></div>
    <p>A broadband Gaussian pulse replaces the CW-like burst. Direct 25 nm pixels start as a density-0.5 midpoint-permittivity waveguide and its solved fundamental mode source. The ideal target is recomputed as the exact causal pulse convolution with a unity-amplitude lossless cosine pole. One reverse pass returns both gradients and commits one unconditional Adam step.</p>
  </a>

  <a class="campaign-card campaign-live" href="fryett_modal_source_unbinarized_dashboard.html">
    <div class="campaign-card-head"><span>Live</span><h2>Fryett nanobeam · unbinarized source-equilibrated discovery</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_modal_source_unbinarized_geometry.png" alt="Unbinarized nested-source geometry"><img src="../_static/generated/fryett_modal_source_unbinarized_fields.png" alt="Unbinarized nested-source field"></div>
    <p>A fresh midpoint-guide campaign keeps continuous filtered density forever—no projection, beta schedule, binary pressure, or boundary handoff. The 2,700-coordinate source is locally equilibrated around every material decision while agreement-controlled trust radii grow without a scheduled release.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fryett_modal_source_frozen_sprint_dashboard.html">
    <div class="campaign-card-head"><span>Stopped at Q≈781</span><h2>Fryett nanobeam · frozen evolved source → smooth boundary</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_modal_source_frozen_sprint_geometry.png" alt="Frozen-source sprint geometry"><img src="../_static/generated/fryett_modal_source_frozen_sprint_fields.png" alt="Frozen-source sprint field"></div>
    <p>A rollback-safe fork freezes the evolved source, jumps directly to beta 64, and removes all source inner solves. Direct material trust-region steps rapidly test the present architecture before fail-closed replay into a smooth subpixel level-set boundary.</p>
  </a>

  <a class="campaign-card campaign-paused" href="fryett_modal_source_bilevel_dashboard.html">
    <div class="campaign-card-head"><span>Paused rollback checkpoint</span><h2>Fryett nanobeam · source-equilibrated dielectric</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_modal_source_bilevel_geometry.png" alt="Prepared nested source-equilibrated dielectric geometry"><img src="../_static/generated/fryett_modal_source_bilevel_fields.png" alt="Pending nested campaign field"></div>
    <p>A nested variable-projection successor starts from a 450 nm midpoint-permittivity guide in a 1 µm-wide region and equilibrates all 2,700 fixed-norm source coordinates before every material decision. Density trust and beta release slowly over accepted commits, then a stable beta-64 topology replays into a smooth cut-cell boundary.</p>
  </a>

  <a class="campaign-card campaign-live" href="fryett_modal_temporal_dashboard.html">
    <div class="campaign-card-head"><span>Live</span><h2>Fryett nanobeam · reciprocal modal current</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_modal_temporal_geometry.png" alt="Prepared neutral modal-current campaign geometry"><img src="../_static/generated/fryett_modal_temporal_fields.png" alt="Pending modal-current campaign field"></div>
    <p>A fixed fundamental-TE feedthrough profile is used as both the centered bidirectional electric current and its reciprocal scalar observable. One absolute causal lossless waveform drives the complete optimization, with Q≥100 pole tracking and a replay-gated smooth-boundary handoff.</p>
  </a>

  <a class="campaign-card campaign-paused" href="fryett_modal_source_codesign_dashboard.html">
    <div class="campaign-card-head"><span>Ready · not started</span><h2>Fryett nanobeam · joint dielectric and source</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_modal_source_codesign_geometry.png" alt="Prepared joint dielectric and modal-current geometry"><img src="../_static/generated/fryett_modal_source_codesign_fields.png" alt="Pending joint source co-design field"></div>
    <p>The fixed-waveform successor exposes all 2,700 physical transverse Ex/Ey/Ez source samples at fixed norm. Its dielectric is a topology-free cubic-spline phase field whose sharp limit is binary bulk plus physical gray cut cells on one continuous boundary; one reverse pass drives both trust-region L-BFGS blocks.</p>
  </a>

  <a class="campaign-card campaign-live" href="fryett_temporal_dashboard.html">
    <div class="campaign-card-head"><span>Live</span><h2>Fryett nanobeam · lossless temporal target</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_temporal_geometry.png" alt="Neutral eight-micron temporal-target campaign geometry"><img src="../_static/generated/fryett_temporal_fields.png" alt="Temporal-target campaign windowed electric field"></div>
    <p>One objective from the neutral start rewards the exact source-driven lossless single-pole waveform. Raw pixel gradients rapidly binarize, a trustworthy Q≥100 pole enables carrier tracking, and a replay-gated level set takes over at the mature boundary.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fryett_4d_qv_dashboard.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>Fryett nanobeam · full-field 4D Q/V</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_4d_qv_geometry.png" alt="Neutral eight-micron four-dimensional Q over V campaign geometry"><img src="../_static/generated/fryett_4d_qv_fields.png" alt="Four-dimensional campaign windowed electric field"></div>
    <p>All 3,200 design pixels participate from initialization. Pulse-on-assisted discovery gives way to a fail-closed pole capture and eigenfrequency-shifted Q/V objective built from six-component fields in multiple time windows.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fryett_hybrid_scratch_dashboard.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>Fryett nanobeam · antenna first, cavity second</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_hybrid_scratch_geometry.png" alt="Eight-micron hybrid Fryett scratch geometry"><img src="../_static/generated/fryett_hybrid_scratch_field_xy.png" alt="Eight-micron hybrid Fryett scratch Ey field"></div>
    <p>Stage 1 gives LDOS only a 1 µm, 320-pixel antenna box. After the antenna is frozen, Stage 2 activates 234 collective Bragg controls for Q growth and dark-state refinement.</p>
  </a>

  <a class="campaign-card campaign-live" href="fryett_encapsulated_qv_dashboard.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>Fryett nanobeam · parametric → fixed-LDOS boundary</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fryett_boundary_geometry.png" alt="Fryett boundary-stage handoff and current geometry"><img src="../_static/generated/fryett_boundary_fields.png" alt="Fryett boundary-stage Ey cavity fields"></div>
    <p>Frozen 151-parameter ellipse discovery followed by topology-preserving normal-boundary optimization of LDOS at the immutable 780.550 nm emitter line.</p>
  </a>

  <a class="campaign-card campaign-paused" href="device23_anthracene_ldos_dashboard.html">
    <div class="campaign-card-head"><span>Paused</span><h2>Device 23 · anthracene LDOS</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/device23_anthracene_density.png" alt="Device 23 reconstructed and optimized SiN geometry"><img src="../_static/generated/device23_anthracene_fields.png" alt="Device 23 anthracene and SiN field profiles"></div>
    <p>Pixelated Si₃N₄ cavity below anthracene/PVA, driven by an Ey molecular emitter and optimized through shifted Q/V.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fourier_atom_hole_sin150_campaign.html">
    <div class="campaign-card-head"><span>Frozen</span><h2>150 nm SiN · Fourier-to-pixel</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fourier_atom_hole_sin150_final_geometry.png" alt="Final 150 nm SiN Fourier and pixel geometry"><img src="../_static/generated/fourier_atom_hole_sin150_final_field.png" alt="Final 150 nm SiN electric field"></div>
    <p>Large isotropic spectral discovery followed by exact pixel handoff and 1,283 total updates around a protected atom hole.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fourier_atom_hole_sin200_campaign.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>200 nm SiN · isotropic Fourier</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fourier_atom_hole_sin200_final_geometry.png" alt="Final 200 nm SiN Fourier geometry"><img src="../_static/generated/fourier_atom_hole_sin200_final_field.png" alt="Final 200 nm SiN x-y and x-z field"></div>
    <p>Thin-slab thickness experiment with full x-z field inspection and an independently doubled vertical-domain audit.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fourier_atom_hole_axis_seed_campaign.html">
    <div class="campaign-card-head"><span>Rejected</span><h2>Axis-biased Fourier seed</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fourier_atom_hole_axis_seed_final_geometry.png" alt="Axis-biased Fourier seed geometry"><img src="../_static/generated/fourier_atom_hole_axis_seed_final_field.png" alt="Axis-biased Fourier seed field"></div>
    <p>Twenty-update diagnostic exposing why the earlier spectral seed activated kx- and ky-aligned waves disproportionately.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="fourier_atom_hole_solid_start_campaign.html">
    <div class="campaign-card-head"><span>Rejected</span><h2>Solid-start Fourier atom hole</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fourier_atom_hole_solid_start_final_geometry.png" alt="Solid-start Fourier atom-hole geometry"><img src="../_static/generated/fourier_atom_hole_solid_start_final_field.png" alt="Solid-start Fourier atom-hole electric field"></div>
    <p>The 665-update solid-slab initialization study that motivated gradual, isotropic spectral discovery.</p>
  </a>

  <a class="campaign-card campaign-paused" href="fourier_atom_hole_ldos_dashboard.html">
    <div class="campaign-card-head"><span>Paused</span><h2>SiO₂-clad 250 nm SiN · spectral LDOS</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/fourier_atom_hole_geometry.png" alt="SiO2-clad Fourier atom-hole geometry"><img src="../_static/generated/fourier_atom_hole_field.png" alt="SiO2-clad Fourier atom-hole fields"></div>
    <p>Forty-thousand-coefficient Fourier mask with differentiable SiN and symmetric SiO₂ thickness controls.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="curved_mirror_beta_campaign.html">
    <div class="campaign-card-head"><span>Qualified success</span><h2>12-front curved mirrors · beta</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/curved_mirror_beta_final_geometry.png" alt="Final curved-mirror beta geometry"><img src="../_static/generated/curved_mirror_beta_final_field.png" alt="Final curved-mirror beta electric field"></div>
    <p>Pixel-polished facing mirrors with a frozen 53.13% two-guide coupling result and full channel accounting.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="curved_mirror_ldos_dashboard.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>12-front curved mirrors · LDOS</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/curved_mirror_ldos_geometry.png" alt="Curved-mirror LDOS optimized geometry"><img src="../_static/generated/curved_mirror_ldos_field.png" alt="Curved-mirror LDOS field profile"></div>
    <p>Warm-started 12-front cavity preserved after 795 LDOS updates and 294 bounded-beta updates.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="curved_mirror_doubling.html">
    <div class="campaign-card-head"><span>Ablation</span><h2>Curved mirrors · 8 / 12 / 16 fronts</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/curved_mirror_doubling_geometry.png" alt="Curved mirror geometries with different segment count"><img src="../_static/generated/curved_mirror_doubling_evidence.png" alt="Curved mirror field and resonance evidence"></div>
    <p>Controlled mirror-count comparison at fixed transverse domain size, including the 12-front seed used downstream.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="curved_mirror_study.html">
    <div class="campaign-card-head"><span>Study</span><h2>Facing curved mirrors · parameterization</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/curved_mirror_fabrication_schematic.png" alt="Fabricable curved alternating-material mirror geometry"><img src="../_static/generated/curved_mirror_intermediate_evidence.png" alt="Promising curved-mirror fields and resonance evidence"></div>
    <p>The geometry model distilled from the most promising intermediate structures: spacing, curvature, apodization, thickness, and embedding.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="curved_mirror_sweep.html">
    <div class="campaign-card-head"><span>Sweep complete</span><h2>Curved mirrors · 12-hour parameter sweep</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/curved_mirror_parameter_examples.png" alt="Curved-mirror sweep geometry variants"><img src="../_static/generated/curved_mirror_sweep_best.png" alt="Best curved-mirror sweep field and response"></div>
    <p>The broad thickness, gap, mirror-spacing, apodization, segment-count, and SiO₂-embedding screen that selected the optimization seed.</p>
  </a>

  <a class="campaign-card campaign-paused" href="axisymmetric_atom_hole_ldos_dashboard.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>Axisymmetric atom hole · radial Q/V</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/axisymmetric_atom_hole_geometry.png" alt="Axisymmetric radial atom-hole geometry"><img src="../_static/generated/axisymmetric_atom_hole_field.png" alt="Axisymmetric atom-hole field profile"></div>
    <p>One-dimensional cosine controls revolved through 2π, with Cartesian audits exposing the reduced-model discrepancy.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="atom_hole_waveguide_dashboard.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>Square atom hole · waveguide beta</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_hole_waveguide_geometry.png" alt="Square atom-hole waveguide geometry"><img src="../_static/generated/atom_hole_waveguide_field.png" alt="Square atom-hole waveguide field"></div>
    <p>7.5 µm no-slot beta-first campaign with a protected central hole, attached guide, and pole surveillance.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="scratch_atom_beta_dashboard.html">
    <div class="campaign-card-head"><span>Frozen</span><h2>From-scratch atom coupler · beta first</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/scratch_atom_beta_geometry.png" alt="From-scratch atom beta geometry"><img src="../_static/generated/scratch_atom_beta_field.png" alt="From-scratch atom beta field"></div>
    <p>Single-etch atom-to-waveguide topology discovery, guided branching-ratio objective, and shifted-pole continuation.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="scratch_atom_beta_final_validation.html">
    <div class="campaign-card-head"><span>Validation</span><h2>Atom coupler · beta sign-off</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/scratch_atom_beta_geometry.png" alt="Validated atom beta geometry"><img src="../_static/generated/scratch_atom_beta_field.png" alt="Validated atom beta field profile"></div>
    <p>The independent spectrum, branching-ratio, hard-mask, and density checks applied to the beta-first final checkpoint.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="scratch_atom_ldos_dashboard.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>From-scratch atom cavity · LDOS first</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/scratch_atom_ldos_geometry.png" alt="From-scratch atom LDOS geometry"><img src="../_static/generated/scratch_atom_ldos_field.png" alt="From-scratch atom LDOS field"></div>
    <p>Historical free-form LDOS campaign whose nearly static gray-material stage motivated stronger continuation controls.</p>
  </a>

  <a class="campaign-card campaign-paused" href="atom_gap_dashboard.html">
    <div class="campaign-card-head"><span>Paused</span><h2>One-micron atom gap cavity</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_gap_current_geometry.png" alt="One-micron atom-gap optimized geometry"><img src="../_static/generated/atom_gap_current_ey.png" alt="One-micron atom-gap Ey field"></div>
    <p>The original hard-gap free-form campaign, retained with its literal geometry and evidence that the mode remained dielectric-centered.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="scratch_atom_final_validation.html">
    <div class="campaign-card-head"><span>Validation contract</span><h2>Atom gap · Q and Purcell sign-off</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/atom_gap_current_geometry.png" alt="Atom-gap candidate geometry"><img src="../_static/generated/atom_gap_current_ey.png" alt="Atom-gap candidate Ey field"></div>
    <p>The fail-closed replay requirements for claiming Q≥1,000 and Purcell≥5 in the one-micron dielectric gap.</p>
  </a>

  <a class="campaign-card campaign-paused" href="folded_sin_atom_waveguide_dashboard.html">
    <div class="campaign-card-head"><span>Paused</span><h2>Folded SiN cavity · atom-to-waveguide</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/folded_sin_atom_waveguide_geometry.png" alt="Folded SiN atom-waveguide geometry"><img src="../_static/generated/folded_sin_atom_waveguide_field.png" alt="Folded SiN atom-waveguide field"></div>
    <p>Opposed 780 nm quasi-2D cavities with the exit guide inside the design region and directional-loss telemetry.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="sin_slot_atom_beta_campaign.html">
    <div class="campaign-card-head"><span>Qualified success</span><h2>SiN atom slot · perpetual beta campaign</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/sin_slot_campaign_geometries.png" alt="SiN atom-slot beginning and ending geometries"><img src="../_static/generated/sin_slot_campaign_fields.png" alt="SiN atom-slot beginning and ending fields"></div>
    <p>The complete multi-stage beta campaign: accepted methods, rejected branches, Pareto checkpoints, and terminal morphology failure.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="sin_slot_atom_beta_dashboard.html">
    <div class="campaign-card-head"><span>Final telemetry</span><h2>SiN atom slot · compact checkpoint</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/sin_slot_atom_beta_geometry.png" alt="Final SiN atom-slot geometry"><img src="../_static/generated/sin_slot_atom_beta_field.png" alt="Final SiN atom-slot field"></div>
    <p>The last literal mask, field, guided channels, loss directions, pole, and shifted-beta metrics from the stopped campaign.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="paper_semi2d_ldos_dashboard.html">
    <div class="campaign-card-head"><span>Stopped</span><h2>Zhang-style quasi-2D cavity · remote atom</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/paper_semi2d_ldos_geometry.png" alt="Zhang-style remote atom cavity geometry"><img src="../_static/generated/paper_semi2d_ldos_field.png" alt="Zhang-style remote atom cavity field"></div>
    <p>Parametric bulge and movable-hole optimization for an atom constrained 0.5 µm from the silicon device edge.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="paper_semi2d_phc_cavity.html">
    <div class="campaign-card-head"><span>Seed reconstruction</span><h2>Zhang-style cavity · unloaded baseline</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/paper_semi2d_geometry.png" alt="Unloaded Zhang-style photonic crystal cavity geometry"><img src="../_static/generated/paper_semi2d_mode_ldos.png" alt="Unloaded cavity mode and remote atom LDOS field"></div>
    <p>The full paper cavity without its bus waveguide, including the pole and atom LDOS 0.5 µm from the dielectric edge.</p>
  </a>

  <a class="campaign-card campaign-paused" href="sin_quasi2d_ldos_dashboard.html">
    <div class="campaign-card-head"><span>Paused</span><h2>780 nm SiN quasi-2D cavity</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/sin_quasi2d_ldos_geometry.png" alt="780 nm SiN quasi-2D optimized geometry"><img src="../_static/generated/sin_quasi2d_ldos_field.png" alt="780 nm SiN quasi-2D field profile"></div>
    <p>Material-scaled remote-atom campaign targeting a 780.24 nm resonance and enhanced evanescent coupling.</p>
  </a>

  <a class="campaign-card campaign-frozen" href="ez_bandgap_quasi2d_dashboard.html">
    <div class="campaign-card-head"><span>Screened</span><h2>Ez bandgap · quasi-2D atom interface</h2></div>
    <div class="campaign-card-images"><img src="../_static/generated/ez_bandgap_unit_cell_screen.png" alt="Ez photonic bandgap unit cell geometry screen"><img src="../_static/generated/ez_quasi2d_field_summary.png" alt="Ez quasi-2D cavity field summary"></div>
    <p>Fabricability-aware slab-hole/post screen and edge taper built to couple a z-oriented atom to a quasi-2D mode.</p>
  </a>
</div>
```

```{toctree}
:hidden:
:maxdepth: 1

nanobeam_waveform_scratch_dashboard
fryett_temporal_dashboard
fryett_modal_temporal_dashboard
fryett_modal_source_codesign_dashboard
fryett_modal_source_bilevel_dashboard
fryett_modal_source_frozen_sprint_dashboard
fryett_modal_source_unbinarized_dashboard
fryett_modal_source_joint_adam_dashboard
atom_hole_temporal_q_dashboard
atom_hole_xyz_temporal_q_dashboard
atom_strip_temporal_collection_dashboard
atom_strip_cladded_qv_dashboard
atom_strip_cladded_qv_yee_restart_dashboard
atom_strip_cladded_air_start_width_qv_dashboard
atom_strip_cladded_tm_q_dashboard
atom_gap_2d_q_over_a_dashboard
composite_waveguide_gap_sweep_dashboard
fryett_shifted_q_dashboard
fryett_symmetric_dipole_mse_dashboard
fryett_port_dipole_adam_dashboard
fryett_4d_qv_dashboard
fryett_hybrid_scratch_dashboard
fryett_encapsulated_qv_dashboard
device23_anthracene_ldos_dashboard
fourier_atom_hole_sin150_campaign
fourier_atom_hole_sin200_campaign
fourier_atom_hole_axis_seed_campaign
fourier_atom_hole_solid_start_campaign
fourier_atom_hole_ldos_dashboard
curved_mirror_beta_campaign
curved_mirror_ldos_dashboard
curved_mirror_doubling
curved_mirror_study
curved_mirror_sweep
axisymmetric_atom_hole_ldos_dashboard
atom_hole_waveguide_dashboard
scratch_atom_beta_dashboard
scratch_atom_beta_final_validation
scratch_atom_ldos_dashboard
atom_gap_dashboard
scratch_atom_final_validation
folded_sin_atom_waveguide_dashboard
sin_slot_atom_beta_campaign
sin_slot_atom_beta_dashboard
paper_semi2d_ldos_dashboard
paper_semi2d_phc_cavity
sin_quasi2d_ldos_dashboard
ez_bandgap_quasi2d_dashboard
```
