The photon-only and shared photon/electron/positron hypotheses have different pair-production thresholds. A direction-uniform analytic bound controls their full three-dimensional thresholds; high-precision collinear witnesses check the arithmetic independently. The calculation supplies no photon flux or empirical rejection. Its physical scale, clock, field attachment and massive dispersion are explicit hypotheses.
The decision contribution and proof state the inputs, exact domain and implication for issue #1025.
From the repository root with the pinned requirements and PYTHONPATH=code:
python -m dispersion_feasibility.bounds
python -m dispersion_feasibility.kinematics
python -m dispersion_feasibility.check
python -m pytest -q code/dispersion_feasibility
kinematics prints the complete report. check reads the retained report or an
explicit path. It imports no producer, reconstructs the thirty directions by a
different formula, and verifies energy conservation, stationarity, positive
collinear curvature and the analytic bracket. Numerical witnesses are not
global optimization certificates; the proof supplies the global bound.
The report's residual field is the producer's internal solve diagnostic, not
the residual of its rounded output: the checker recomputes conservation from
the 35-digit serialized witness to 1e-29 eV. Decimal fields have bounded
length and exponent and reject fraction expressions and nonfinite values.
This contribution occupies one bounded feasibility session. Numerical work is limited to 15 cumulative local CPU minutes, with no cloud run, flux transport, future-data access or Lean expansion. It does not assert completion of #1025, activate #1026 or amend the immutable FZ-12 registration. The P value comes from the exposed alpha calibration; the length conversion and electron mass are declared nominal inputs without a propagated calibration uncertainty.