Follow the source behind the result.

The registered Python source is shown as text. Displaying it does not execute the code.

"""v122 -- The margins are theorems: the established, frozen selectors
(D4 annihilator + two diamond determinants) pin R uniquely among hexagon
matrices.  The v121 atom margins lose their input status.  [I] exact
census; the H2 residue reduces to selectors already in the ledger.

v121 left "derive the six atom margins + det = rank" as the residue.
This module closes it: the margins FOLLOW from constraints that were
established and frozen long before the address question was even posed.

  THE THREE ESTABLISHED SELECTORS (all frozen pre-v120):
    (S1) the D4 annihilator n = (5, -9, 6):  n^T R = (8, 0, 0)
         (v94/tfpt_2: n kills the unchanged (c2, c3) plane; n.a = 8);
    (S2) det R = 8 = rank E8       (v71 determinant quartet (3,4,8,20));
    (S3) det K = 4 = |mu_4| with K = R + Q Sigma   (v71/v94 diamond).
    Plus the definitional range: entries in the hexagon {0..5}
    (residues mod p2 = 6).

  [I] 1. CENSUS.  Hexagon columns with n.col = 8: exactly 4; with
         n.col = 0: exactly 4  =>  64 candidates satisfy (S1).
         Imposing (S2) leaves 12.  Imposing (S3) leaves EXACTLY ONE --
         and it is R.
  [I] 2. COROLLARY: THE MARGINS ARE THEOREMS.  The atom margins
         (4, 8, 10) / (4, 13, 5), the anchor column R e_1 = a, and all
         v120/v121 sum rules (16, 10, 14, 24, 40) are CONSEQUENCES of
         (S1)-(S3) -- their input status is revoked.
  [I] 3. BONUS LEMMA: THE L-DETERMINANT IS AUTOMATIC.  On the
         (S1)+(S2) locus, det(M + 2U) = 20 holds for ALL 12 candidates
         -- the fourth quartet determinant det L = 20 is NOT an
         independent selector but a consequence of the annihilator and
         the rank determinant.  The quartet (3, 4, 8, 20) carries
         redundancy on this locus, exactly the overdetermination
         pattern the compiler shows elsewhere.
  [P] 4. RESIDUE (recorded): what remains is the establishment of the
         selectors themselves -- n = (5, -9, 6) (the D4-equivariant
         annihilator) and the diamond determinants (already inventoried
         ledger objects, CONTRACT.U / FLAV.*).  H2 introduces NO new
         residual class: the address question is fully reduced to
         pre-existing inventory.  FIREWALL: n, the quartet, R, L and
         all words frozen in v4/v18/v20/v71/v94 -- nothing fitted.
"""
from itertools import product

import sympy as sp

from tfpt_constants import check, summary, reset

R = sp.Matrix([[1, 3, 0], [1, 5, 2], [2, 5, 3]])
Q = sp.Matrix([[3, 1, 0], [3, 2, 0], [3, 2, 1]])
SIG = sp.diag(1, -1, -1)
U_AX = 3 * sp.ones(3, 1) * sp.Matrix([[1, 0, 0]])
N_VEC = (5, -9, 6)


def hex_cols(target):
    return [(x, y, z) for x, y, z in product(range(6), repeat=3)
            if 5 * x - 9 * y + 6 * z == target]


def run():
    reset()
    print("v122 margin theorem (established selectors pin R)")

    # selectors hold for R (established, re-verified)
    nrow = sp.Matrix([[5, -9, 6]])
    check("ESTABLISHED SELECTORS re-verified on R: n^T R = (8,0,0) "
          "(D4 annihilator, v94), det R = 8 (v71), det(R + Q Sigma) = "
          "det K = 4 (v71/v94 diamond), n.a = 8",
          (nrow * R) == sp.Matrix([[8, 0, 0]])
          and R.det() == 8 and (R + Q * SIG).det() == 4
          and (nrow * sp.Matrix([1, 1, 2]))[0] == 8)

    # 1. census
    c8, c0 = hex_cols(8), hex_cols(0)
    cands = []
    for c1 in c8:
        for c2 in c0:
            for c3 in c0:
                m = sp.Matrix(3, 3, lambda i, j: [c1, c2, c3][j][i])
                if m.det() == 8:
                    cands.append(m)
    final = [m for m in cands if (m + Q * SIG).det() == 4]
    check("CENSUS: hexagon columns with n.col = 8: exactly 4, with "
          "n.col = 0: exactly 4 => 64 (S1)-candidates; det = 8 leaves "
          "12; det(M + Q Sigma) = 4 leaves EXACTLY ONE -- and it is R",
          len(c8) == 4 and len(c0) == 4
          and len(cands) == 12 and len(final) == 1 and final[0] == R)

    # 2. corollary: margins are theorems
    m = final[0]
    check("COROLLARY: the atom margins are THEOREMS -- the unique "
          "solution has row sums (4, 8, 10) = (e1, rank, p3), column "
          "sums (4, 13, 5) = (e1, Delta_Q, g_car), and anchor column "
          "M e_1 = a = (1,1,2); the v120/v121 sum rules follow -- "
          "their input status is revoked",
          [sum(m.row(i)) for i in range(3)] == [4, 8, 10]
          and [sum(m.col(j)) for j in range(3)] == [4, 13, 5]
          and [m[i, 0] for i in range(3)] == [1, 1, 2])

    # 3. bonus lemma: det L automatic
    check("BONUS LEMMA: on the (S1)+(S2) locus det(M + 2U) = 20 = "
          "det L holds for ALL 12 candidates -- the fourth quartet "
          "determinant is NOT an independent selector but a "
          "consequence (the quartet carries redundancy: the compiler's "
          "overdetermination pattern)",
          all((mm + 2 * U_AX).det() == 20 for mm in cands))

    # 4. residue
    check("RESIDUE [P] (recorded): H2 reduces to the establishment of "
          "the selectors themselves -- n = (5,-9,6) and the diamond "
          "determinants, both pre-existing inventoried ledger objects; "
          "the address question introduces NO new residual class. "
          "FIREWALL: n, quartet, R, L, words frozen in "
          "v4/v18/v20/v71/v94 -- nothing fitted", True)

    return summary("v122 margin theorem")


if __name__ == "__main__":
    raise SystemExit(1 if run() else 0)

d35cbd2ed636412594800fb3d1bcba6320f78df11dc856cc1bc3dc076c008bdb

The public registry includes 116 document views, 6,693 section records and 985 physical LaTeX sources. Active publication documents, research material and archives retain their recorded roles. This is source navigation, not a new semantic review or proof run.