No paired runtimeA different physical lens for 250 m observation and direction control; it does not replace the 3 m design. Loading a recording never runs the solver. New synthesis is offered after an accepted recorded 3 m case is inspected.
Editable input geometry · runtime preview unavailable
No field response is attached to this geometry. Evaluate or wait for the current analytical preview.
Historical runs are immutable records. Opening a result never overwrites the editable draft. This browser retains the last 12 inspected results for offline use. Save workspace includes its selected numerical result for portable reopening.
No retained lens runs on this runtime.
3 km lens cap — 60 GHz — 10,000 × 10,000 cells
Scalar propagation with an assumed local dielectric slab. This is a design approximation; it is not a full-wave Maxwell solution, a validated fabricated radome, or calibrated power/gain. Saved runtime and LLM settings
Selecting a preset loads its saved physical parameters as an unsolved draft. Editing a draft does not change its named preset until you save it again. The recorded result stays bound to its original run and candidate.
The 10,000 × 10,000, 3 km design is a built-in selectable preset. Named presets, your edited draft, selected run and worker reference persist in this browser. Save current step writes a portable workspace; reopening it follows the retained run without starting another solve.
Actual design: 100,000,000 cells · pitch 0.624568 mm · projected square D 6.245676 m · spherical radius 4.684257 m. Point source inside the upper cap at the sphere center; this cap does not enclose a complete sphere.
Loading bounded geometry preview…
Moving the source translates this source-centered cap. Requested target coordinates do not alter the camera scale.
ADO uses the saved runtime LLM backend. It proposes the four bounded coefficients; the runtime evaluates the full physical grid and scores actual target intensity. No browser-generated history or simulated progress is shown.
This preset uses one prescribed scalar point source at the cap center. It has no independently driven phased-array elements or beamforming weights. The 10,000 × 10,000 cells are passive dielectric thickness samples, not 100 million antenna feeds.
Pitch 0.62457 mm · square projected aperture 6.2457 m · cap radius 4.6843 m.
Unsolved input bounds: 0.4–9 mm. Step 0.1 mm. The diagram shows input bounds, not computed cell heights.
The source-to-target path sets a phase prescription. Four bounded coefficients change global, radial-quadratic, radial-quartic and x/y-astigmatic phase before quantization into dielectric thickness.
Available route: scalar Kirchhoff surface propagation with a local slab response. Full-wave Maxwell and working-size PNGF are not admitted for this 100-million-cell preset. No backend switch silently changes its grid or geometry.
No numerical field or lens effect has been computed for this draft. Evaluate to obtain a full-grid scalar result.
No simulated field is attached to the unsolved geometry. The 3D view above shows only the declared cap envelope and source/target direction.
Opening a run never submits a new solve. The compact workspace keeps this preset independent of your earlier Lens Designer drafts, search mode and results.
schema: metahz.lens-procedural-radome/v1
id: radome-60ghz-10000x10000-3km
name: '3 km lens cap — 60 GHz — 10,000 × 10,000 cells'
revision: 1
frequency_Hz: 60000000000
source_m:
- 0
- 0
- 0
target_m:
- 0
- 0
- 3000
grid_shape:
- 10000
- 10000
pitch_m: 0.0006245676208333333
cap_radius_m: 4.68425715625
material:
name: Bambu PLA (assumed)
epsilon_r: 2.57
loss_tangent: 0.015
status: user_assumed
reference_frequency_Hz: 60000000000
realization:
min_thickness_m: 0.0004
max_thickness_m: 0.009
thickness_step_m: 0.0001
design:
coefficients_rad:
- 0
- 0
- 0
- 0
bounds_rad:
-
- -3.2
- 3.2
-
- -1.6
- 1.6
-
- -1.6
- 1.6
-
- -1.6
- 1.6
coefficient_step_rad: 0.1
solver:
fidelity: analytical
cell_chunk_size: 65536
observation:
offsets_m:
-
- 0
- 0
- 0
-
- 3
- 0
- 0
-
- -3
- 0
- 0
-
- 0
- 3
- 0
-
- 0
- -3
- 0
-
- 0
- 0
- -1000
-
- 0
- 0
- 1000
optimization:
mode: ado
seed: 19
max_evaluations: 6
max_seconds: 900
max_memory_mb: 512
Hemispherical radome · 60 GHz · approximately 100 million patches · 3 km
The 2.49 m-radius shell is a physical design input. Its 99,860,316 irregular wedge patches are derived on demand. The available whole-dome evaluator is scalar and approximate; bounded vector full-wave panel results are reported separately and do not validate the whole dome or a 3,000 m focus.
Geometry: radius 2.49 m · hemisphere area 38.95638 m² · 6,262 polar rings · 99,860,316 variable-width patches · nominal pitch 0.6245676 mm. The circular rim lies at z=0 m and has diameter 4.98 m.
Loading bounded dome view…
At 60 GHz the roughly 5 m aperture is about 1,000 wavelengths wide. The 3 km target is within its approximate 10 km far-field boundary, so a target calculation must keep complex Fresnel-region phase.
The shell is centered on the base plane, open underneath for mounting. The prescribed source is at 0, 0, 0 m. This abstract point source is not a solved antenna feed.
Polar-ring sectors change width with latitude, so patch outlines follow the curved surface. A compact ring-count rule defines all 99,860,316 patches; the display renders only a bounded geometry sample. Junctions between rings need conforming edge splitting before fabrication.
Four smooth coefficients represent the target-directed phase correction. The default radial terms preserve rotational symmetry. A Gaussian surface basis could add localized corrections later; a Dirichlet/Voronoi partition would define patch boundaries, not solve Maxwell’s equations.
The runtime whole-dome route uses scalar propagation with assumed local slab response. A small coupled-current vector full-wave panel can test local dielectric behavior. Neither route is a validated complete-dome Maxwell solution, and the material values need measurement.
Reopening reads an existing run; it does not start a new evaluation.