Summary
The existing ethanol–coronene candidate, called G96 in the project, is not a confirmed local minimum or stationary point. A separate signed decision accepted it as energetically adequate for a limited final-energy route after scans of all ten unresolved soft directions. Other checks support the numerical route. These decisions answer narrower questions than a minimum test, so the frequency classification remains 119 positive, 0 negative and 10 undetermined.
A ladder of checks
| Check | Accepted result | Boundary |
|---|---|---|
| Repeatability | Repeated energy-and-force calculations meet pre-declared tolerances. | Repeatability does not establish physical accuracy. |
| Energy and derivative consistency; grid response | Finite energy differences agree with the implemented derivatives; prescribed grid-response tests passed. | Numerical consistency is not absolute force accuracy or convergence to the radial-grid limit. |
| Local geometry contribution | A separately accepted rule for a local geometry-budget estimate passed. | An upper heuristic in a local model, not a bound on arbitrary joint displacements. |
| Ten soft-direction scans | No decrease beyond the accepted threshold within the tested scope. | Curvature signs, stationarity and a local minimum remain unconfirmed. |
The targeted energy test
For G96 a minimum was not confirmed. The sign of the curvature of its ten softest modes at r2SCAN-3c could not be established with the required reliability. On a denser angular grid, estimates from the two measured Hessian columns changed by 29% and 7%, both increasing. This response did not establish the curvature signs.
Along each of the ten modes, single-point energies were computed in both displacement directions using finite rigid-fragment rotations. The amplitudes were chosen so that a negative curvature of the same magnitude would lower the energy by approximately 0.016 kcal/mol. No energy decrease larger than 0.01 kcal/mol was found: the four softest modes were checked on the denser angular grid; the remaining six used the standard grid with a grid-transfer allowance measured on those four modes.
Acceptance is specific to G96 and this scan. Combinations of modes, points beyond the selected amplitudes and radial grid convergence were not tested. The scan follows the r2SCAN-3c surface, not the final single-point energy surface. It supplies no general geometric-error bound outside the scanned lines and amplitudes.
The separate local geometry estimate
The separate geometry-budget rule uses the analytic Hessian for displacements along 119 stiffer modes. For the ten soft modes it uses an envelope of local-minimum estimates from cubic fits to the energy scans and Newton steps at three numerical settings. This upper heuristic does not cover combinations of modes or points outside the scanned amplitudes. Passing the rule does not change G96’s structural classification.
The targeted scan and local geometry estimate are separate pieces of evidence. Neither establishes physical accuracy of the final energy method or an error bound for later target-material differences.
What the candidate can be used for
The project accepted the prescribed numerical checks and the nine final single-point calculations for this candidate. Calculation completion is separate from agreement with published reference work. This note reports no interaction or binding energy; the literature comparison and its follow-up require their own report and disclosure decision.
The practical use is to examine the pilot’s computational route before extending it to target-material comparisons. The candidate remains one structure from one starting position. The accepted checks do not establish a local or global minimum, selectivity, water effects or cartridge performance. Scientific acceptance is an internal project decision by the project lead; independent external scientific review of the computational work has not been performed.
Change history
v1.0 ·
First publication.