Public research release · October 16, 2026 · Revision 1
The fusion gap, made inspectable.
A reproduced geometric result and a negative inventory check from the current 8T + 7O study. The result is partial: it does not demonstrate a physical fusion event.
The released scope is rigid docking geometry and cell bookkeeping. The original computational study predates this public release; the release date records when these selected materials became public.
What was actually reproduced
| Result | Value | Meaning |
|---|---|---|
| Distinct tracked cells | 30 | Two unchanged 15-cell candidates: 16T + 14O, one connected face graph. |
| Contact at the chosen endpoint | 2 shared faces | Rigid geometric contact under the declared FCC construction. |
| Continuous translation check | 225 cross-cell pairs; 0 positive-volume intersections | Exact rational separating-axis intervals for conservative convex host-center cells. |
| Search scope | 8 orientation classes; 200 face-matched candidates | 1 ranked candidate certified for the displayed approach; no global minimum-energy claim. |
| Invalid-motion control | Inward overtravel rejected | Moving beyond contact violates the geometric bound. |
Classification: computed under explicit assumptions. These are exact geometric checks within the model, not measured particle trajectories. The conservative cell bounds enclose the modeled curved interstices; clear bounds certify a clear path for those interstices. A failed bound alone does not establish a physical collision.
A useful negative result: six inputs are insufficient.
Two current protons provide 16T + 14O. Under the retained product assumptions—proton 8T + 7O, neutron 20T + 10O, positron 2T + 4O, and outgoing neutral packet 2T + O—the products require 32T + 22O.
| Incoming neutral packets (2T + O each) | Total input | Shortfall against retained products |
|---|---|---|
| 0 | 16T + 14O | 16T + 8O |
| 6 | 28T + 20O | 4T + 2O |
| 8 | 32T + 22O | Balanced inventory |
The earlier six-packet scheme cannot be carried over unchanged: it is short by 4T + 2O. Eight incoming packets balance this ledger, with one outgoing packet, so net absorption is seven. Charge stays +2e under q/e = (2O − T)/6.
What remains unresolved: the neutron here is a retained bookkeeping target; a transition law, source and coupling for neutral inputs, bound-state energy, reaction rate, product signatures, and net energy still need derivation and testing. Balancing the inventory is necessary for this proposed accounting and does not cause a nuclear reaction.
Reproduce the artifact
- Download the Python checker and save it as
fusion-evidence-checker.py. - Place fusion-evidence-inputs.json in the same folder. It contains only the selected input geometry.
- Run
python fusion-evidence-checker.pyusing Python 3.9 or later. Only the standard library is used. - Compare the generated result with the published audit. File hashes and provenance are in the release manifest.
The public checker preserves the geometric algorithm and changes file I/O for standalone use. It additionally recomputes the reaction inventory from constituent counts. Its geometric output was compared field-for-field with the retained audit.
Coordinates and scope
Doubled FCC coordinates: host A = (2,0,0), host B = (−2,−2,−2); relative rotation diag(−1,−1,1); translation B(s) = B + s(−4,−2,−2), s ≥ 0. The preliminary rotation is bounded at a larger separation. These are model coordinates without a physical length, speed, force or time scale.
Release ID: dclxvi-fusion-evidence-2026-10-16-r1. The remaining staff studies and their original URLs retain their access restrictions. This release is complete for the stated geometric and inventory check.
The next result must go beyond geometry.
Specify an explicit transition rule and a quantitative observable that separates the proposed mechanism from the null model. Reject transitions that lose or duplicate cells, violate the stated conservation laws, cross forbidden regions, or require unaccounted energy. Then design the apparatus, calibration, backgrounds, controls, detection thresholds and independent replication around that prediction.
A negative result at any gate is publishable progress. A first credible reaction signal, if obtained, would still be distinct from net energy, a practical power source or commercial valuation.