Speaker
Mathias Kuschick
Description
Promoting predictions for top quark production to include off-shell effects comes at an exceptionally high computational cost. On an example of $l^+\nu l^-\bar{nu} b \bar{b}$ process, dominated by top quark pair production and decay in the dilepton channel, we show how to reliably encode off-shell effects in a generative neural network based on direct diffusion. This network, constructed using POWHEG/bb4l and POWHEG/hvq samples, can be applied to new or existing POWHEG/hvq samples at almost no computational cost.
Authors
Anja Butter
Mathias Kuschick
Michael Klasen
Sofia Palacios Schweitzer
(ITP, University Heidelberg)
Tilman Plehn
Tomas Jezo
(WWU ITP)
