3385 — Backward and Forward Occurrence Composition Commute Only at Full Support
Question
Can the middle (+2) of the proposed (3+2+1) architecture be realized as distinct backward and forward composition operators on the same frozen finite model?
Directed incidence operators
Use the three oriented occurrence differences as columns of the forward boundary operator
Forward compositionality sends labelled edge or coherence coefficients to their occurrence boundary. The source-normalized backward coboundary is
It evaluates occurrence data on the same labelled differences.
Full-support square
Exact calculation gives
Writing
the common composite is
Thus backward and forward composition are mutually compatible on the complete labelled triangle. Their common kernel is the invariant line, and their common restriction to (A_2) is invertible after normalization by (1/3).
Supported hostile test
Remove the third labelled route with
The supported operators are
Their two composites no longer agree. The exact mixed residue is
Its rank is three. This falsifies the initial expectation that support deletion would leave only another rank-two (A_2) residue. The incompatibility can reach the invariant direction even though the full-support composite kills it.
Reversing one forward orientation while retaining the backward source orientation also produces a nonzero defect. Hence the strict square depends on the shared labelled orientation, not merely on equal dimensions.
Narrow result
The middle (+2) admits a finite operational realization:
- backward compositionality is the occurrence coboundary;
- forward compositionality is the occurrence boundary;
- their full-support comparison closes strictly;
- deleting one route creates a full-rank mixed commutator.
This makes the interaction of the two towers support-sensitive. The mixed residue is not reducible to either tower separately.
Scope
This is an occurrence-incidence theorem. Identifying the supported commutator with a cosmological physical class still requires source-derived Cut/sewing maps and an instrument comparison. No new carrier stratum is proposed.
Verification
Checker:
research/benincasa/checkers/audit_backward_forward_occurrence_composition.py.
Packet:
research/benincasa/results/backward_forward_occurrence_composition.json.
Allocator claim: seqclaim-c9543ab4b508584e5f7a7788.
Epistemic graph event:
ev-000000007254-443b2daf-0ab7-438d-9f55-7780ab0a1a77.