Construction workbench: modes, evidence and limits
Open the bundled workbench · Belt launch preset · Construction gallery
These are separate mathematical experiments, not stages of one physical nuclear model. Select Construction process in the Process panel, then the mode's View menu where available. If the panel is closed, use Process to reopen it. The viewer retains settings on reload. Only the belt launch link above promises an explicit query preset; choose the other processes through the controls rather than inventing preset URLs.
Mode directory
| Experiment | Selection in the bundled workbench | Read with it |
|---|---|---|
| Native N-cut belt | Use Belt launch preset above. Geometry and Wiring are different readouts. | Belt counts, tip contacts and open transport. |
| Be-8 | Construction process → Be-8; White-rooted hex simplex; enable Show Be-8 reference. | Eight-route reference versus layer construction. |
| Be-9 | Construction process → Be-9; View → White-rooted hex simplex; enable Show Be-9 reference. | The Central neutron sleeve (trial) switch changes four internal passages, not the eight identities or endpoint permutation. |
| Retained expansion | Construction process → Sulfur-32, View → Retained expansion. | Workflow steps retain construction records; they are not physical time or a stability proof. |
| Cellulation steps | Construction process → Sulfur-32, View → Cellulation. | Order dependence and the blocked strict source-role patch. Legacy replay is not a repaired role-compatible shell. |
| Whole-cell reference | Open the separate cellulation reference, or select 12x12 fibration and follow its cellulation link. | This displays the whole-cell reference, not the animated shell's event search or a proof of confluence. |
| NNPE trees | Construction process → Sulfur-32, View → NNPE composition. | Assigned inventory, attachment ports and allowed joins. |
| Hex strands | Construction process → Sulfur-32, View → Hex strands · 3D. | Retained seed identities and display layers are not a derivation of all inter-layer connections. |
| Wiring and cutaway | Construction process → Sulfur-32, View → Wiring or Channels / cutaway. | Conditional local continuation result. A displayed reference route is not independently generated global wiring. |
Drag to orbit and use the scene controls to zoom or change view. Visibility, lane offsets and simplified/collapsed drawings do not by themselves change model identities or supply interaction laws. Assets are bundled locally; the interactive scene requires WebGL. See the per-family dependency table.
Beryllium routing and the sleeve trial
The Be-8 reference is an independent octahedral routing candidate: four rim entry ports and two retained turn choices give eight complete routes across twelve native edges. Each route keeps its original identity while its output slot is permuted. Four passes restore every slot; this is a slot-routing law, not a physical period or a theorem that the complete construction history becomes empty.
The reference and the two fixed-direction component rows retain the same guides, IDs and endpoint slots. Component colors change under the declared reading convention; this does not reverse the paths or make crossings into new connections. Separate display lanes are drawing offsets, not additional edges.
Be-9 starts from the same eight reference identities. Its sleeve trial replaces the middle segment of four mixed P/E routes by explicit center passages; the other four guides and the full endpoint permutation stay unchanged. Internal edge–strand incidence therefore changes even when endpoint data agrees. Toggle the trial to compare with the unsleeved reference. It does not calculate a binding energy, lifetime or isotope-stability law.
Be-8 source, matrix and tests · Be-9 retained passages and tests.
NNPE inventory is not port counting
The trial begins with sixteen labelled generators. Each is assigned inventory N:1/P:1 independently of its N/P/E attachment-port labels. Thus the model starts with 16N/16P. E is an attachment type here, not an extra independently counted electron; joining P to E does not annihilate generator inventory.
Use the labelled port buttons, or their corresponding projected scene targets, to choose two attachments. N joins N; P joins E. Only different connected components may join: occupied ports and loop-producing joins are rejected. A complete sixteen-generator tree has fifteen joins and eighteen external ports while retaining all sixteen inventories and generator identities. No unique connecting tree is selected by this convention.
Collapse composites changes the display, not the retained binary history or inventory. Undo removes the last join; Reset restores separate generators. Join and collapse state persist locally. The half-cube surfaces are a representation of this trial, not a physical particle decomposition or a full higher-operadic certificate.
NNPE construction, counting convention and tests · Separate operation-level composition audit.
Sulfur local continuation: what was predicted
The source reports a reference-calibrated angular rule selecting 16/16 target next edges under leave-one-carrier-out tests: both records of the target carrier are excluded from that carrier's calibration. The plane-only rule retains the correct edge but leaves three departures in every case. The completed rule receives the native neighborhood, known carrier orientation, incoming plane and polarity tuple—not a target next-port pair.
It also has 768 synthetic uniqueness checks over 48 native frames and sixteen polarity tuples. These establish local rule uniqueness at the stated inputs, not a globally conserving, collision-free or closed route. The rule was formulated during the investigation; its within-reference tests are not independent experimental validation or independent validation of the discovery procedure.
The remaining gate is to derive the incoming frame and polarity context from the layer construction without using target wiring as a lookup. Complete routes, carrier generation, inter-layer wiring and physical interpretation remain unproved by this one-step result. The viewer's existing reference geometry was not changed by the source audit.
Local rule, rejected alternatives, calibration and reproduction.
Verification boundary
Source-reported scientific checks above were not rerun to write this guide. Publication/browser checks exercise the named controls and packaged assets; they do not prove nuclear structure, validate measured masses, or close the source obligations. In particular, a passing browser check can display a deliberately unresolved or falsified construction.