ALP-ttbar Toy MC v1.3: Post-registration Results for Vertex Resolution and Tails from a Mini Tracking Simulation of Displaced Light ALP → e+e− (companion to arXiv:2610.38932)

Personal, numpy-only toy Monte Carlo written after reading arXiv:2610.38932 (J. Alimena et al., "Hunting for Long-Lived ALPs in Top Events via Displaced Dielectron Pairs at the LHC"). Not affiliated with the authors or the CMS Collaboration; only qualitative trends are meaningful. v1.3 replaces the assumed vertex-resolution tails of v1.1–v1.2 with a mini tracking simulation (helices in 3.8 T, multiple scattering, bremsstrahlung, hits, joint two-track vertex fit), under a public preregistration frozen before any result (github.com/wspaik5/ALP_ttbar_Toy_MC_v1.3_Preregistration_Mini_Tracking). Without bremsstrahlung, conversion vertices are located to about 2 mm and the conversion veto works (about 2x loss). With bremsstrahlung that the fit does not model, vertices shift outwards by tens of mm and the loss in reach is 183x at 17 MeV and 17x at 200 MeV, worse than every v1.1 scenario. Four of five preregistered predictions were refuted, one confirmed. The decisive quantity is the bremsstrahlung treatment of the electron fit. A numerical bug found in the first run and two methodological additions are reported as deviations. Code: github.com/wspaik5/ALP_ttbar_Toy_MC_v1.3_Postregistration_Mini_TrackingPrevious versions: v1.2 10.5281/zenodo.23847880, v1.1 10.5281/zenodo.23833475, v1.0 10.5281/zenodo.23825276