Recoil uncertainty in top mass measurements

We present the effect of treating the top quark as the recoiler during gluon emission in $t\rightarrow bW$ decay on direct top quark mass measurements. We study this effect using the Pythia8 Monte Carlo event generator within the CMSSW framework, and compare to recoil configurations used during LHC Run 2 and legacy Pythia8 configurations. We implement various analyses using the Rivet analysis toolkit to study this effect. The top mass is extracted from the lepton+jet hadronic top mass distribution, normalized jet mass in boosted top quark decays, the lepton+soft muon invariant mass distribution, and the invariant lepton+bottom quark mass distribution. The strong coupling constant $\alpha_s$ is tuned to CMS data separately for each of the three recoil configurations using the highly correlated variable $\Delta R_g$. For each of the recoil models with and without tuned $\alpha_s$, we extract a mass from each of the mentioned mass distributions, and calculate the difference between models. This process is done for both the Pythia8 Monash and CP5 tunes, as well as POWHEG+Pythia8 matching. Our study suggests that recoiling to the top quark introduces a systematic shift of at least -200 MeV shift to Run 2 top mass measurements.