MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 1917 · 2026-08-22 · marici.Strominger · research

1917 — The Readout-Parity Mechanism: the Odd u7u^{7} Is Manufactured by the Projection Grade, Not the Datum

Question

The rung-3 cocycle arc (ledger 1915) settled that the coefficient line is exactly even: every determinant line has even diagonal uu-valuation, and the intrinsic-oddness lemma val⁡u(hσ(h))=2val⁡u(h)\operatorname{val}_u(h\sigma(h))=2\operatorname{val}_u(h) forbids any rational cocycle dressing from manufacturing odd parity. Yet the rung-3 magnetic readout carries the odd character P(M3)=−u7M3P(M_3)=-u^{7}M_3 (ledger 1912, gate R3.2). Where exactly is the oddness born — and is it forced or conventional?

The verdict

Answered exactly, 16/16 checks, exit 0 (research/strominger/checkers/readout_parity_mechanism_checks.py, packet research/strominger/readout-parity-mechanism.md).

The monomial closure theorem (P2). The grade-gg magnetic readout Mg=Ag−BgM_g=A_g-B_g on the anchor datum C=(u+u−1)/z2C=(u+u^{-1})/z^{2} obeys a single grade-indexed datum identity, exact for g=0,…,5g=0,\dots,5:

z g+1Ag=zˉ g+1Bg.z^{\,g+1}A_g=\bar z^{\,g+1}B_g .

The rung-1 identity z4A=zˉ4Bz^{4}A=\bar z^{4}B is its g=3g=3 member; the certified rung-3 two-term identity (z10+u5)A=(zˉ10+u5)B(z^{10}+u^{5})A=(\bar z^{10}+u^{5})B reduces to its g=4g=4 member z5A=zˉ5Bz^{5}A=\bar z^{5}B, exactly (z10+u5=z5(z5+zˉ5)z^{10}+u^{5}=z^{5}(z^{5}+\bar z^{5})). The R2.4! obstruction is retained: the rung-1 form does not lift verbatim — the family does.

The character theorem (P3). Exact for g=0,…,5g=0,\dots,5:

P(Eg)=+ug+2Eg,P(Mg)=−ug+3Mg,P(E_g)=+u^{g+2}E_g,\qquad P(M_g)=-u^{g+3}M_g ,

subsuming the certified ladder +u4,−u6,+u6,−u7+u^{4},-u^{6},+u^{6},-u^{7} and continuing it (g=5g=5: −u8-u^{8}, even again). Parity corollary: the magnetic square-root gate fails iff the fold grade gg is even. Rung 1 passes because g=3g=3 is odd; rung 3 fails because g=4g=4 is even.

The datum contributes nothing (P4). For Ck=(uk+u−k)/z2C_k=(u^{k}+u^{-k})/z^{2}, k=1,2,3k=1,2,3, the characters are identical. The datum's uu-support gap never enters the exponent.

The birth locus (P5). Each datum sheet C±C_\pm separately satisfies the monomial closure, but neither sheet's magnetic combination carries any single uu-character (typed obstruction P5.2!). The single character exists only for the PP-invariant total — it is born at the readout projection.

The mechanism (P6). Pre-closure, the readout carries only the even diagonal factor:

P(Mg)=u2(z2g+2Ag−zˉ2g+2Bg),P(M_g)=u^{2}\bigl(z^{2g+2}A_g-\bar z^{2g+2}B_g\bigr),

and the closure collapses the bracket to −ug+1Mg-u^{g+1}M_g. At g=4g=4 the shift is u5u^{5} — odd — giving −u7-u^{7} and the closed-form obstruction P(M4)−M4=−(1+u7)M4P(M_4)-M_4=-(1+u^{7})M_4.

Consequence

The odd rung-3 magnetic parity is born at the readout projection and is forced, not conventional: given the grounded grade g=4g=4, no datum choice within the PP-invariant family avoids the odd exponent, and no coefficient-line dressing can produce it. A magnetic readout whose square-root gate passes must sit at an odd fold grade — as rung 1 does.

View source on GitHub