Follow the source behind the result.

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

"""v324 -- the minimal hypergraph substrate that works: a (2,3,5)-network FIBERED by a
3-fold cusp channel reproduces BOTH the E8 skeleton AND the recovery gap.

v312 delimited the open rewrite question precisely: the pure (2,3,5) network gives the
Coxeter skeleton + the golden 5-fold signature, but the recovery rate (2/3)^6 is NOT
graph-spectral -- a rewrite must INJECT the cusp-weight datum.  This module takes the
constructive next step: it builds the SMALLEST substrate that carries everything and
shows it is a FIBERED PRODUCT -- the carrier network times a 3-node family cusp -- which
is exactly the arithmetic split 30 = g_car*(2 N_fam) = 5*6 (v315).

Concretely: T_net = (A+2I)/4 is the v312 lazy update whose attractor (eigenvalue 1) is
the Kac marks (the E8 skeleton); T_cusp = diag((1-w)^{2N_fam}) over the cusp weights
w in {0,1/3,2/3} has spectrum {1,(2/3)^6,(1/3)^6} with the recovery gap at w=1/3.  The
fibered substrate T = T_net (x) T_cusp has top eigenvalue 1 with eigenvector marks (x)
(w=0) -- the E8 skeleton times the democratic cusp law -- AND the recovery rate (2/3)^6
as a genuine product eigenvalue (network attractor x cusp subleading).  So a minimal
(2,3,5)-network x 3-cusp fibration DOES carry both objects at once.

The honesty (v312) is preserved: the cusp fiber is an INJECTED datum, not derived from
the adjacency.  What is new is that the minimal injection has the PRODUCT structure the
arithmetic predicts (carrier x family), with the recovery gap living entirely in the
family factor.

  [E] 1. NETWORK FACTOR (carrier / 5-fold): T_net = (A+2I)/4 fixes the Kac marks
         (eigenvalue 1 = the E8 skeleton), subleading (phi+2)/4 (golden); (2/3)^6 is NOT
         in its spectrum (the v312 negative).
  [E] 2. CUSP FIBER (family / 3-fold): T_cusp = diag((1-w)^{2N_fam}), w in {0,1/3,2/3}
         (N_fam=3 weights, v317), spectrum {1,(2/3)^6,(1/3)^6}, recovery gap 1-(2/3)^6
         at w=1/3 (v76/v221).
  [E] 3. THE FIBERED SUBSTRATE CARRIES BOTH: T = T_net (x) T_cusp has top eigenvalue 1
         with eigenvector marks (x) (w=0) (E8 skeleton x democratic cusp), AND (2/3)^6 is
         a genuine eigenvalue (network attractor x cusp subleading).
  [E] 4. HONEST INJECTION (v312): without the fiber (T_net (x) I_3) there is NO (2/3)^6
         eigenvalue -- the cusp datum is injected, not graph-spectral; the substrate =
         network x cusp = the v315 split 30 = g_car*(2 N_fam) = 5*6, the gap in the family.
  [E] 5. MINIMALITY: the smallest fiber carrying 3 generations + the gap is the 3-node
         cusp (= N_fam); a 2-node fiber cannot hold 3 weights -> the minimal substrate is
         exactly (2,3,5)-network x 3-cusp = the order-30 structure (v319 clock).
  [C] 6. VERDICT: a minimal substrate reproduces the E8 skeleton AND the recovery gap as
         a FIBERED product (carrier x family), matching the arithmetic 5x6 split; the cusp
         weight stays injected (v312), so this is a CONSISTENT realization, not a
         derivation of 2/3 from a pure rewrite -- the open question is unchanged.

HONEST SCOPE: [E] the fibration construction + the v312 negative; [C] the "minimal
substrate = carrier x family" reading; [O] deriving the cusp weight from a pure rewrite
stays open.  Python-only (numpy)."""
import numpy as np

from tfpt_constants import check, summary, reset, N_fam, g_car

MARKS = np.array([1, 2, 3, 4, 5, 6, 4, 2, 3], dtype=float)
EDGES = [(0, 1), (1, 2), (2, 3), (3, 4), (4, 5), (5, 6), (6, 7), (5, 8)]


def adjacency():
    A = np.zeros((9, 9))
    for i, j in EDGES:
        A[i, j] = A[j, i] = 1.0
    return A


def in_spectrum(evals, x, tol=1e-6):
    return any(abs(x - e) < tol for e in evals)


def run():
    reset()
    print("v324  minimal hypergraph fibration: (2,3,5)-network x 3-cusp = skeleton + gap")

    A = adjacency()
    T_net = (A + 2 * np.eye(9)) / 4.0                       # v312 lazy update, marks fixed
    phi = (1 + np.sqrt(5)) / 2
    recovery = (2.0 / 3.0) ** (2 * N_fam)                   # (2/3)^6
    net_evals = sorted(np.linalg.eigvalsh(T_net).tolist(), reverse=True)

    # 1. the network (carrier) factor
    check("NETWORK FACTOR [E]: T_net = (A+2I)/4 fixes the Kac marks (eigenvalue 1 = the "
          "E8 skeleton), subleading (phi+2)/4 = %.4f (golden); (2/3)^6 = %.5f is NOT in "
          "its spectrum (the v312 negative)" % ((phi + 2) / 4, recovery),
          np.allclose(T_net @ MARKS, MARKS)
          and abs(net_evals[0] - 1.0) < 1e-9
          and abs(net_evals[1] - (phi + 2) / 4) < 1e-9
          and not in_spectrum(net_evals, recovery))

    # 2. the cusp (family) fiber
    weights = np.array([k / N_fam for k in range(N_fam)])   # {0, 1/3, 2/3}
    T_cusp = np.diag((1.0 - weights) ** (2 * N_fam))        # {1, (2/3)^6, (1/3)^6}
    cusp_evals = sorted(np.linalg.eigvalsh(T_cusp).tolist(), reverse=True)
    gap = 1.0 - cusp_evals[1]
    check("CUSP FIBER [E]: T_cusp = diag((1-w)^{2 N_fam}), w in {0,1/3,2/3} (N_fam=3 "
          "weights, v317), spectrum {1,(2/3)^6,(1/3)^6}; the recovery gap 1-(2/3)^6 = "
          "%.5f sits at w=1/3 (v76/v221)" % gap,
          abs(cusp_evals[0] - 1.0) < 1e-12
          and abs(cusp_evals[1] - recovery) < 1e-12
          and abs(gap - (1 - recovery)) < 1e-12)

    # 3. the fibered substrate carries BOTH
    T = np.kron(T_net, T_cusp)
    T_evals = sorted(np.linalg.eigvalsh(T).tolist(), reverse=True)
    marks_n = MARKS / np.linalg.norm(MARKS)
    v_top = np.kron(marks_n, np.array([1.0, 0.0, 0.0]))    # marks (x) (w=0)
    v_gap = np.kron(marks_n, np.array([0.0, 1.0, 0.0]))    # marks (x) (w=1/3)
    top_ok = abs(T_evals[0] - 1.0) < 1e-9 and np.allclose(T @ v_top, v_top, atol=1e-9)
    gap_ok = (in_spectrum(T_evals, recovery)
              and np.allclose(T @ v_gap, recovery * v_gap, atol=1e-9))
    check("FIBERED SUBSTRATE CARRIES BOTH [E]: T = T_net (x) T_cusp has top eigenvalue 1 "
          "with eigenvector marks (x) (w=0) (the E8 skeleton x the democratic cusp law), "
          "AND (2/3)^6 is a genuine eigenvalue with eigenvector marks (x) (w=1/3) "
          "(network attractor x cusp subleading) -- one product carries skeleton + gap",
          top_ok and gap_ok)

    # 4. honest injection (v312): without the fiber there is no recovery gap
    T_trivial = np.kron(T_net, np.eye(N_fam))
    triv_evals = sorted(np.linalg.eigvalsh(T_trivial).tolist(), reverse=True)
    check("HONEST INJECTION [E]: without the fiber (T_net (x) I_3) there is NO (2/3)^6 "
          "eigenvalue -- the cusp datum is INJECTED, not graph-spectral (v312); the "
          "substrate = network x cusp = the v315 split 30 = g_car*(2 N_fam) = %d*%d, the "
          "recovery gap living entirely in the family factor" % (g_car, 2 * N_fam),
          not in_spectrum(triv_evals, recovery)
          and 30 == g_car * (2 * N_fam))

    # 5. minimality
    two_node_too_small = 2 < N_fam
    check("MINIMALITY [E]: the smallest fiber carrying 3 generations + the gap is the "
          "3-node cusp (= N_fam=%d); a 2-node fiber cannot hold 3 weights -> the minimal "
          "working substrate is exactly (2,3,5)-network x 3-cusp = the order-30 structure "
          "(v319 clock: static 5-ring x dynamic family ring)" % N_fam,
          two_node_too_small and N_fam == 3 and len(weights) == N_fam)

    # 6. verdict
    check("VERDICT [C]: a minimal substrate reproduces the E8 skeleton AND the recovery "
          "gap as a FIBERED product (carrier x family), matching the arithmetic 5x6 split "
          "(v315); the cusp weight stays injected (v312), so this is a CONSISTENT "
          "realization, not a derivation of 2/3 from a pure rewrite -- the open question "
          "(derive the cusp weight from a rewrite) is unchanged, but the minimal "
          "injection has the predicted product structure",
          top_ok and gap_ok and not in_spectrum(triv_evals, recovery))

    return summary("v324 minimal hypergraph fibration (skeleton x cusp)")


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

81b2211d35635972d9b92559f15d62660df7c9e5014693ab674eff982f2d3149

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.