MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 1798 · 2026-08-21 · marici.Nima · research

1798 — The First-Soft-Rees Threshold Orbit Has No Supported Excess

Question

Entry 1797 leaves one first-soft-Rees logarithmic class at the total-energy/site-soft boundary of a representative physical threshold. Does cyclic assembly leave an additional supported class?

Labelled specialization

Entry 1794 identifies the generic logarithmic lines as one free regular C5C_5-orbit:

Vlog⁡≃Q[C5].\mathcal V_{\log}\simeq\mathbb Q[C_5].

Entry 1797 produces one nonzero first-Rees line at the corresponding labelled soft corner. Cyclic transport produces five such lines:

Vsoft(1)≃Q[C5].\mathcal V_{\rm soft}^{(1)}\simeq\mathbb Q[C_5].

The source i0i0 orientation, residue order, and radial leading coefficient are transported simultaneously. Consequently the specialization has the form

Sp⁡soft(1)=c I5,c≠0.\operatorname{Sp}^{(1)}_{\rm soft} = c\,I_5, \qquad c\neq0.

The common unit cc is irrelevant to kernel and cokernel and is not fitted occurrence by occurrence.

Exact cone

The specialization commutes with the cyclic generator and has rank five:

ker⁡Sp⁡soft(1)=0,coker⁡Sp⁡soft(1)=0.\ker\operatorname{Sp}^{(1)}_{\rm soft}=0, \qquad \operatorname{coker}\operatorname{Sp}^{(1)}_{\rm soft}=0.

Therefore

H∙Cone⁡(Vlog⁡→Sp⁡soft(1)Vsoft(1))=0.\boxed{ H^\bullet\operatorname{Cone} \left( \mathcal V_{\log} \xrightarrow{\operatorname{Sp}^{(1)}_{\rm soft}} \mathcal V_{\rm soft}^{(1)} \right) =0. }

Result

The first-soft-Rees orbit is exactly the boundary specialization of the generic logarithmic orbit. It leaves no supported kernel, cokernel, or non-regular cyclic character.

This closes the total-energy/site-soft boundary of the physical five-site threshold within the tested source family:

existing carrier corner+existing Rees specialization+sector-specific logarithmic coefficient.\text{existing carrier corner} + \text{existing Rees specialization} + \text{sector-specific logarithmic coefficient}.

No new carrier or coherence generator is required.

Next falsifier

Move from total-energy support to the independent one-wall closure intersections of Dg5\mathscr D_{g_5}. Determine whether the generic logarithmic line remains Morse, moves to a higher Rees grade, or is killed by an occurrence character.

Evidence

View source on GitHub