Neutrino Longitudinal Momentum Reconstruction from CMS Open Data

Abstract

This independent technical note documents a reproducible reconstruction study based on the complete 100,000-event educational CMS W-to-muon-and-neutrino sample published through CERN Open Data (record 5205). All published event rows were analyzed without additional selection cuts. Under a fixed W-boson mass input of 80.3625 GeV, 88,243 events were classified as admitting two real longitudinal-neutrino candidates, while 11,757 admitted no real candidate under the imposed constraint. No event entered the defined numerical tangent band. A control run using 80.4 GeV gave 88,291 two-candidate events and 11,709 no-real-candidate events. Verification includes source-file identity checks, 22 passing automated tests, mass-condition checks, and a separately implemented numerical audit of candidate classes and labels. A no-real classification does not, by itself, mean that a collision was invalid or the source data were corrupted. This document reports a reconstruction and reproducibility study, not a measurement of the W-boson mass, a detector-level analysis, or an official CMS/CERN publication. It has not undergone external peer review. Software, evidence and reproduction instructions: https://github.com/cxclrfx/Neutrino-Longitudinal-Momentum-Reconstruction. Original data: https://opendata.cern.ch/record/5205.