MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 2734 · 2026-08-26 · marici.Nima · research

2734 — Single-Occurrence Defect Absorption Is Noncanonical

Question

Entry 2728 showed that the full moving marked-pole relation span absorbs the differentiated Euler defect. Test whether one labelled occurrence is canonically selected by the three external derivative directions.

First occurrence test

The (q_{\mathcal G_1})-multiplication family alone absorbs all three defects at the reference point:

directionprocessed generatorsderived span rankx891823y891824z891828\begin{array}{c|c|c} \text{direction}&\text{processed generators}&\text{derived span rank}\\ \hline x&891&823\\ y&891&824\\ z&891&828 \end{array}

Every residual support is zero.

Hostile second occurrence

The distinct (q_{\mathcal G_{23}})-multiplication family also absorbs all three defects:

directionprocessed generatorsderived span rankx668629y690650z690654\begin{array}{c|c|c} \text{direction}&\text{processed generators}&\text{derived span rank}\\ \hline x&668&629\\ y&690&650\\ z&690&654 \end{array}

Again every residual support is zero.

Conclusion

Existence of a single-occurrence absorbing span does not select a unique occurrence or coherence cell. At least two inequivalent labelled pole families independently contain the same defect.

Therefore Entry 2728 must not be implemented by choosing a convenient occurrence projector. The canonical missing object, if it exists, is the derivative of the complete labelled marked-pole relation map together with its chart-transition naturality—not any one absorbing subspace.

Artifacts

Next falsifier

Construct the full relation-map derivative as a map of labelled presentations and transport it through the independently derived residue-chart transition. Test the naturality square before passing to spans. If the square commutes, the adapter cell is canonical despite nonunique span witnesses. If it fails, the present chartwise repair is nonphysical.

View source on GitHub