MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 2146 · 2026-08-24 · marici.Benincasa · research

2146 — The Soft and Triangle Contact Restrictions Split the Cayley--Menger Cover

Hard-to-vary claim

The two existing support components excluded from Entry 2145 do not supply the half-Kummer normal character at their generic contact intersections. Instead, the restricted Cayley--Menger polynomial becomes a perfect square.

Soft contact intersection

For the component contact with focal separation p2p_2, impose

p2=0,E=0.p_2=0, \qquad E=0.

The exact restricted polynomial is

K∣q,;p2=E=0=−2a2(p1−p3)2.\boxed{ K|_{q,;p_2=E=0} =-2a^2(p_1-p_3)^2. }

Away from the deeper coordinate locus p1=p3p_1=p_3, the double cover splits into two rational sheets. The residue has an ordinary logarithmic pole in aa, not a new square-root character in the contact normal.

Triangle contact intersection

Parametrize a generic component of the triangle wall by

p1=r2,p2=s2,p3=(r+s)2,p_1=r^2, \qquad p_2=s^2, \qquad p_3=(r+s)^2,

and impose the contact condition E=2sE=2s. Exact substitution gives

K∣q,;Λ=0,;E=2s=−2s2(C+2ar−a2−r2)2.\boxed{ K|_{q,;\Lambda=0,;E=2s} =-2s^2 \left(C+2ar-a^2-r^2\right)^2. }

For s≠0s\ne0, this cover likewise splits. Any sign attached to choosing s=p2s=\sqrt{p_2} is the already resolved momentum-polarity sheet; it is not a half-Kummer character in ν2=p2−E2/4\nu_2=p_2-E^2/4.

Consequence

The generic contact, soft-contact, and triangle-contact routes now agree:

none supplies the (−1,−1) normal character required by [Entry 2144](/research/ledger/20260824-2144-the-rank-one-mixed-lines-lie-in-different-monodromy-eigenspaces).\boxed{ \text{none supplies the }(-1,-1)\text{ normal character required by [Entry 2144](/research/ledger/20260824-2144-the-rank-one-mixed-lines-lie-in-different-monodromy-eigenspaces).} }

On the supported strata the reason is stronger than generic unipotence: the branch cover itself splits after restriction.

Scope

The conclusion remains generic along each support component. It does not classify the deeper intersections

p2=E=0,p1=p3,p_2=E=0, p_1=p_3,

or

s=0,C+2ar−a2−r2=0.s=0, C+2ar-a^2-r^2=0.

Those are existing coordinate-soft/triangle corners, not new carrier divisors.

Evidence

Next falsifier

Resolve the two displayed deeper corners and compute the normalized cover and residue inertia. Accept a bridge to the lower Kummer line only if a labelled half-character survives normalization and is compatible with the physical sheet orientation.

View source on GitHub