Optimizing Higgs Boson CP Measurement in $H\to \tau \tau $ decay with ML Techniques

Current measurements of the CP state of the Higgs boson have favoured a scalar Higgs boson but are not able to exclude a mixing of scalar and pseudoscalar Higgs boson states. A measurement of possible mixed CP states of the Higgs boson is best done through the $H \to ττ$ decay mode. The decay products of $τ$ leptons, produced in $H \to ττ$ decays, encode the CP information of the Higgs. This presents a challenge as a large proportion of τ decays involve cascade decays to three pions. This results in increased complexity in defining a CP sensitive observable. Deep learning tools (through neural networks) have been employed to extract as much sensitivity as possible. This neural network approach has been shown to effectively separate scalar and pseudoscalar hypothesis with decays of $τ$ to three pions. Assessing the effectiveness of this approach involves studies into detector resolution and $τ$ -decay modelling. Improvements to the approach are sought through the use of $E^{miss}_T$