Exact Low-Deficiency Witnesses Have No Transferable Core Or Local Repair¶
Claim/Theorem¶
Keep the notation of [[exact-low-deficiency-subfamilies-do-not-share-transferable-visible-template.md]] and [[exact-optimal-quotient-families-do-not-share-transferable-shelling-law.md]]. The current explicit Quantum Tanner low-cut data do not expose any transferable finer support/dependency rule of either of the following two common kinds:
-
a large common-core rule:
\[ H_{\mathrm{core}}: \]every exact low-deficiency witness family on a cut contains one canonical large block core;
-
a local-repair rule:
\[ H_{\mathrm{repair}}: \]any two exact low-deficiency witness families differ only by a short sequence of local support-preserving repairs.
More precisely:
-
The common-core idea already fails on the current
D_4low cut. Two exact zero-deficiency witness families on that cut are\[ A=\{g=sr^2,\ g=sr,\ g=s,\ g=r^2\}, \]and
\[ B=\{g=sr,\ g=sr^3,\ h=e,\ h=r\}. \]Their intersection is
\[ A\cap B=\{g=sr\}, \]so they share only one block. Thus no large common-core description is visible even on one explicit low cut where both families are exact zero-deficiency witnesses.
-
The size of this shared core is not even stable across the current explicit low-cut witnesses. Existing exact-optimal witness pairs already on disk have the following intersection sizes:
D_4:|A\cap B|=1for the two zero-deficiency witnesses above;D_4:|A\cap B|=2for the same-cardinality optimal pair in [[exact-optimal-quotient-families-do-not-share-transferable-shelling-law.md]];D_6:|A\cap B|=4for the recorded optimal6-block pair there;D_8:|A\cap B|=5for the recorded optimal7-block pair there.
So the current explicit data do not exhibit any transferable support-core size or core-shape pattern across cuts and instances.
-
The local-repair idea already fails at the same time. The same
D_4,D_6, andD_8pairs from [[exact-optimal-quotient-families-do-not-share-transferable-shelling-law.md]] witness failure of one-step same-cardinality exchange. Therefore even when two exact witness families share a large overlap, the remaining difference is not repaired by one local swap that preserves optimality. -
Combining the two phenomena gives the sharper obstruction:
- some exact low-deficiency witnesses on the same cut can have very small overlap, so there is no visible large forced core;
- other exact witness pairs can have substantial overlap, yet still fail local repair;
- hence the missing selection object is neither a common core nor a bounded local repair graph on visible supports.
Therefore the current frontier has already moved beyond support-level templates:
- coarse visible labels failed in [[exact-low-deficiency-subfamilies-do-not-share-transferable-visible-template.md]];
- finer common-core and local-repair explanations also fail on the explicit witness pairs already on disk;
- the remaining missing object is a genuinely global selection invariant on the arrangement of exact low-deficiency witness families themselves.
The exact missing invariant can now be stated as
for every \beta-balanced cut L, there is a construction-controlled global invariant of the quotient-block arrangement that identifies or certifies a block subfamily X with
and
without relying on a visible common core, a fixed profile, or a bounded local repair rule.
This sharpens the distinction among three levels:
- one witness satisfies a pattern;
- all witnesses on one cut share a pattern;
- a transferable construction-controlled rule persists across cuts and instances.
The current D_4, D_6, and D_8 evidence supports only the first level.
Dependencies¶
- [[exact-low-deficiency-subfamilies-do-not-share-transferable-visible-template.md]]
- [[exact-optimal-quotient-families-do-not-share-transferable-shelling-law.md]]
- [[subfamily-union-rank-deficiency-gives-parity-lower-bound.md]]
- [[nu-saturation-yields-adapted-triangular-basis.md]]
- [[packed-quotient-images-already-attain-global-cut-rank-on-small-quantum-tanner-instances.md]]
Conflicts/Gaps¶
- This node rules out currently visible common-core and local-repair explanations only at the level of the explicit witness pairs already on disk; it does not prove that no deeper transferable rule exists.
- The intersection statistics used here come from explicit witness families, not from exhaustive classification of all low-deficiency witnesses on every cut.
- So the node sharpens the obstruction but does not yet replace exact global optimization by a new positive invariant.
- What remains missing is a genuinely global selection invariant for exact low-deficiency witness arrangements.
Sources¶
10.48550/arXiv.2206.0757110.48550/arXiv.2508.05095