CMS dE/dx Estimator Optimization and prototyping for Anomalous Ionization Searches*

Heavy Stable Charged Particles (HSCPs) predicted by supersymmetric and other beyond-Standard-Model scenarios traverse the CMS silicon tracker at sub-relativistic speeds and deposit unusually large ionization energy (dE/dx), the key observable for reconstructing their mass. However, because the per-hit energy-loss distribution is Landau-skewed, conventional estimators such as the CMS harmonic-2 average try to down weight the skewed tail but may leave residual high-tail bias, limiting mass resolution and discovery reach. Using 2018 CMS HSCP gluino simulations (√s = 13 TeV), we analyze existing estimators and prototype two alternative per-track estimators: a Landau-MPV fit that extracts the most-probable dE/dx value via Minuit-based regression with adaptive Freedman–Diaconis binning, and a variable-truncation (VT) estimator that iteratively removes non-Gaussian tails. The VT method demonstrates superior bias suppression and stability, achieving visually narrower reconstructed-mass peaks than harmonic-2 , while the Landau-MPV fit provides accurate results when convergent but reduced robustness on low-hit or highly skewed tracks.Current mass reconstruction still relies on Bethe–Bloch parameterization constants derived from the harmonic-2 estimator; ongoing work focuses on re-extracting these constants using the new estimators to achieve more reliable and unbiased mass measurements in CMS data.