AQ23 and borrow forward propagation
AQ23 and inherited borrow metadata can preserve sparse exact structure into later low-bit transitions.
Diagnostic work, exact algebra and local capabilities are not treated as end-to-end mining advantage.
What was tested?
AQ23 and inherited borrow metadata can preserve sparse exact structure into later low-bit transitions.
Why the test is meaningful
Sparse algebraic structure is useful only if it survives its actual consumers. V19 propagates AQ23 programs and borrow metadata through R6–R8, tracking where new nonlinear terms appear rather than assuming that low degree is conserved.
ANF(F∘f) computed exactly on each 16-point cubesupport(f)={S:a_S=1}selector claim requires stable mapping from borrow/context to exact propagated programHow it was tested
Build propagation and nonlinearity atlases, compile sparse R6 programs, assess borrow as a selector and test exact propagation through R8.
What happened
AQ23-to-R8 propagation passed exactness gates. Borrow retained semantic value but descriptive associations were not promoted to shortcut claims.
Exactness and statistical controls
Propagation outputs were checked exhaustively within each source cube and high-24 parent frequencies were labelled as sampled. Exact formulas passed, but descriptive borrow associations were not promoted to a selector theorem.
What the result means
Sparse programs exist in sampled registered contexts; a universal selector and runtime economics remain pending.
Limitations
- High-24 prevalence estimates are empirical.
- A universal selector remains unproved.
- Program sparsity is not a native hardware benchmark.
Evidence trail
Exact within each 16-point source cube; high-24 parent frequencies are explicitly marked as sampled.
Canonical variants
SUBENGINE-V19ASUBENGINE-V19BSUBENGINE-V19CSUBENGINE-V19DSUBENGINE-V19ESUBENGINE-V19FSUBENGINE-V19GSUBENGINE-V19HSource: internally audited canonical reports. Local filesystem structure, private headers and operational identifiers are excluded from publication.