Improving the Logarithmic Accuracy of the Angular-Ordered Parton Shower

Monte Carlo event generators are a key tool for making theoretical predictions that can be compared with the results of collider experiments, our most accurate probes of fundamental particle physics. New developments in the way parton shower accuracy is assessed have led us to re-examine the accuracy of the angular-ordered parton shower in the Herwig 7 event generator, focussing on the way recoil is handled after successive emissions. We first discuss how the evolution variable is defined in the Herwig angular-ordered shower and how the choice of this definition determines the recoil scheme. We then show how the recoil scheme can affect the logarithmic accuracy of final-state radiation produced by the algorithm. As part of this investigation we consider a new interpretation of the evolution variable intended to mitigate problems with previous iterations of the shower. To test this, simulated events for each scheme are compared with experimental data from both LEP and the LHC. Next we extend our analysis to initial-state radiation and perform the same process of assessing the logarithmic accuracy of different interpretations of the evolution variable. This time, we compare simulated events for each scheme with LHC data for the vector boson production. Additionally, we consider the impact that the choice of NLO matching scheme has on the accuracy of these simulations, with reference to the same LHC data.