Lessons from Machine Learning of Higgs CP measurements in H --> tautau; tau -->3pi nu, for the TauSpinner algorithm and 'truth' simulation of leptons

par Zbigniew Was (IFJ PAN Krakow)

jeudi 30 nov. 2017, 11:00 → 12:00 Europe/Paris

Petit Amphi (Annecy-le-Vieux)

In recent papers we have looked for the Higgs CP parity measurement in the cascade decay H --> tau+tau- ; tau+- --> 3(2)pi nu.
The sensitive observable is of multi dimensional nature. Machine learning approach enabled to nicely integrate information available in each kinematical variable into valid estimator.
To start the design and to confirm correctness of the estimate, weighted event sample was used. For that purpose, the TauSpinner algorithm wasused.
This result of interest in itself, underline importance for future Monte Carlo programs to emulate fully differential distributions.
Comments in context of the observables relying on precisely reconstructed leptons will be made.